Contrast-enhanced radiology systems, methods, and computer program products using machine learning

Abstract

A method for providing a prediction of a representation of an examination region, the prediction being generated using a medical imaging technique involving a contrast agent, the method comprising: receiving a first representation of the examination region of an examination subject in a frequency space; receiving a second representation of the examination region of an examination subject in a frequency space; providing the first and second representations as inputs to a predictive machine learning model, the predictive machine learning model being configured to provide a prediction as output of a representation of the examination region in a frequency space after a certain amount of contrast agent has been applied during the medical imaging technique; receiving the output of the predictive machine learning model based on the inputs; and converting the output of the predictive machine learning model into a representation of the examination region of the examination subject in a physical space.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit of European patent application No. EP21160325.3, filed on 2 March 2021, and European patent application No. EP21167116.9, filed on 7 April 2021, the entire disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] This disclosure generally relates to generating radiographic images, and in some non-limiting embodiments relates to providing systems, methods, and computer program products for generating predictions of representations of radiographic images, said predictions being generated using medical imaging techniques involving contrast agents. Background Technology

[0004] Radiology can refer to a branch of medicine that uses medical imaging techniques to diagnose and treat diseases. Radiology can involve the application of electromagnetic radiation and mechanical waves, such as generating images during ultrasound procedures for diagnostic, therapeutic, and / or scientific purposes. In some embodiments, ionizing radiation such as X-rays, gamma rays, and / or electrons can be used to generate images during medical imaging procedures. Radiology also includes other imaging methods such as computed tomography (CT), positron emission tomography (PET), ultrasound, magnetic resonance imaging (MRI) (also known as nuclear magnetic resonance imaging (NMRI)), and / or similar methods, although these medical imaging techniques do not use ionizing radiation. In some cases, contrast agents (e.g., radioactive contrast agent materials, radioactive contrast agents, contrast media, etc.) and / or flushing agents (such as saline) can be used in medical imaging techniques such as angiography, CT, ultrasound, MRI, and / or similar techniques. Contrast agents can be used (e.g., injected into the patient's bloodstream) to provide contrast enhancement to images generated based on medical imaging techniques. For example, contrast agents may include those used to improve the depiction of a patient's (e.g., human or animal) physical structure and / or function during medical imaging techniques.

[0005] CT scans (e.g., computed axial computed tomography (CAT) scans) can include medical imaging techniques used in radiology to non-invasively acquire detailed images of the body for diagnostic purposes. One device used to perform a CT scan is a CT scanner. A CT scanner uses a rotating X-ray tube and a row of detectors placed in a gantry to measure the attenuation of X-rays by different tissues within a patient's body. Multiple X-ray measurements acquired from different angles can be processed on a computer using reconstruction algorithms to generate tomographic (e.g., cross-sectional) images of the body.

[0006] MRI can include a medical imaging technique, particularly used in medical diagnostics, to depict the structure and function of tissues and / or organs within a patient's body. In MRI, the magnetic moments of protons in the subject are aligned in a fundamental magnetic field, resulting in macroscopic magnetization along the longitudinal direction. The magnetic moments in the subject are then deflected from a resting position (e.g., a relaxed position) by irradiation with high-frequency (HF) pulses (e.g., by excitation with HF pulses of electromagnetic radiation). The recovery or magnetization dynamics of the protons from the excited state to the resting position can then be detected as a relaxation signal by one or more HF receiver coils of the MRI machine. For spatial encoding, rapidly switching magnetic gradient fields can be superimposed on the fundamental magnetic field. The relaxation signal is initially presented as raw data (e.g., detected MRI data) in frequency space (e.g., frequency domain, spatial frequency space, Fourier space, Fourier depiction, etc.) and can be transformed to entity space (e.g., image space) via an inverse Fourier transform. In one embodiment, tissue contrast is formed during the use of the original MRI by different relaxation times (e.g., spin-lattice relaxation time and spin-spin relaxation time, referred to as T1 and T2) and / or proton density. Spin-lattice relaxation time describes the transition of longitudinal magnetization to its equilibrium magnetization state, where the spin-lattice relaxation time is measured as the time required to reach 63.21% equilibrium magnetization before resonant excitation. Spin-spin relaxation time describes the transition of transverse magnetization to its equilibrium magnetization state in a similar manner.

[0007] During CT imaging procedures, iodine-containing solutions can be used as contrast agents. During MRI imaging procedures, superparamagnetic materials (e.g., iron oxide nanoparticles, superparamagnetic iron-platinum particles (SIPPs), etc.) and / or paramagnetic materials (e.g., gadolinium chelates, manganese chelates, etc.) can be used as contrast agents. In ultrasound imaging procedures, liquids containing gas-filled microbubbles can be used, with the liquid administered intravenously. MRI contrast agents work by altering the relaxation time of structures that take up the contrast agent. A distinction can be made between two groups of materials: paramagnetic and superparamagnetic. Both groups have unpaired electrons that induce magnetic fields around individual atoms or molecules. Superparamagnetic contrast agents cause a significant shortening of T2, while paramagnetic contrast agents primarily cause a shortening of T1. The effect of contrast agents can be indirect, as the contrast agent itself does not emit a signal but only affects the signal intensity in its vicinity. One example of a superparamagnetic contrast agent is iron oxide nanoparticles (e.g., superparamagnetic iron oxide (SPIO)). Examples of paramagnetic contrast agents are gadolinium chelates, such as gadopentetate dimeglumine (e.g., ), gadoteric acid (e.g., , , ), gadodiamine (e.g., ), gadoterol (e.g., ) and gadobutrol ( ). Summary of the Invention

[0008] Accordingly, systems, methods, and computer program products for providing predictions of representations of examination areas are disclosed, the predictions being generated using medical imaging techniques involving contrast agents.

[0009] According to some non-limiting embodiments, a system is provided for providing a prediction of a representation of an examination region, the prediction being generated using a medical imaging technique involving contrast agents. The system includes at least one processor programmed or configured to receive a first representation of an examination region of a subject in a frequency space, wherein the first representation includes a representation of an examination region for which no amount of contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; and to receive a second representation of the examination region of the subject in a frequency space, wherein the second representation includes a representation of an examination region for which a second amount of contrast agent was applied during the medical imaging technique, wherein the second amount of contrast agent is different from the first amount of contrast agent. The system provides input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent has been applied during a medical imaging technique, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent; receives the output of the predictive machine learning model based on the input; converts the output of the predictive machine learning model into a predicted representation in entity space of an examination region of an examination object; and provides a predicted representation in entity space of an examination region of an examination object.

[0010] According to some non-limiting embodiments, a computer program product is provided for providing a prediction of a representation of an examination region, the prediction being generated using a medical imaging technique involving contrast agents. The computer program product includes at least one non-transitory computer-readable medium comprising one or more instructions, which, when executed by at least one processor, cause the at least one processor to receive a first representation of an examination region of an examination object in a frequency space, wherein the first representation includes a representation of an examination region for which no amount of contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique, and to receive a second representation of the examination region of the examination object in a frequency space, wherein the second representation includes a representation of an examination region for which a second amount of contrast agent was applied during the medical imaging technique. The representation of a region, wherein the second amount of contrast agent differs from the first amount of contrast agent, provides input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent has been applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent, receives the output of the predictive machine learning model based on the input, converts the output of the predictive machine learning model into a predicted representation of the examination region of the examination object in physical space, and provides a predicted representation of the examination region of the examination object in physical space.

[0011] According to some non-limiting embodiments, a method is provided for providing a prediction of a representation of an examination region, the prediction being generated using a medical imaging technique involving contrast agents. The method includes receiving, with at least one processor, a first representation of the examination region of an examination subject in a frequency space, wherein the first representation includes a representation of an examination region for which no amount of contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; receiving, with the at least one processor, a second representation of the examination region of the examination subject in a frequency space, wherein the second representation includes a representation of an examination region for which a second amount of contrast agent was applied during the medical imaging technique, wherein the second amount of contrast agent is different from the first amount of contrast agent; and providing input to a predictive machine learning model using the at least one processor. The input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation. The predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of the representation in frequency space of an examination region to which a third amount of contrast agent has been applied during medical imaging, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent. The at least one processor receives the output of the predictive machine learning model based on the input, converts the output of the predictive machine learning model into a predicted representation in entity space of the examination region of the examination object, and provides the predicted representation in entity space of the examination region of the examination object.

[0012] Further non-restrictive embodiments are set forth in the following numbered clauses:

[0013] Item 1: A system for providing a prediction of an examination region, the prediction being generated using a medical imaging technique involving contrast agents, the system comprising at least one processor programmed or configured to perform the following operations: receiving a first representation of an examination region of an examination subject in a frequency space, wherein the first representation includes a representation of an examination region for which no amount of contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; receiving a second representation of an examination region of an examination subject in a frequency space, wherein the second representation includes a representation of an examination region for which a second amount of contrast agent was applied during the medical imaging technique, wherein the second amount of contrast agent is different from the first amount of contrast agent; providing a prediction. The method includes taking input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent has been applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent; receiving the output of the predictive machine learning model based on the input; converting the output of the predictive machine learning model into a predicted representation of the examination region of the examination object in physical space; and providing a predicted representation of the examination region of the examination object in physical space.

[0014] Item 2: The system as described in Item 1, wherein the at least one processor is further programmed or configured to perform the following operations: specifying a portion of the frequency space center of the first representation, a portion of the frequency space center of the second representation, or portions of the frequency space center of both the first and second representations, to provide a simplified representation of the inspection region of the object in the frequency space; wherein the input to the predictive machine learning model includes the simplified representation of the inspection region of the object in the frequency space; and wherein, when the input to the predictive machine learning model is provided, the at least one processor is programmed or configured to: provide the simplified representation of the inspection region of the object in the frequency space as the input to the predictive machine learning model.

[0015] Item 3: The system as described in Item 2, wherein the at least one processor is further programmed or configured to: supplement the output of the predictive machine learning model with: a portion of the first representation that does not contain a frequency space center and is not specified for providing a simplified representation of the inspection region of the object in frequency space; a portion of the second representation that does not contain a frequency space center and is not specified for providing a simplified representation of the inspection region of the object in frequency space; or portions of the first representation and the second representation that respectively do not contain a frequency space center and are not specified for providing a simplified representation of the inspection region of the object in frequency space, to provide supplementary output of the predictive machine learning model; wherein, when the output of the predictive machine learning model is converted into a predicted representation of the inspection region of the object in entity space, the at least one processor is programmed or configured to: convert the supplementary output of the predictive machine learning model into a predicted representation of the inspection region of the object in entity space.

[0016] Item 4: The system as described in any of Items 1-3, wherein the first quantity of contrast agent applied during the medical imaging technique is greater than zero, and wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

[0017] Item 5: The system as described in any of Items 1-3, wherein the first quantity of contrast agent applied during the medical imaging technique is zero, and wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

[0018] Item 6: A system as described in any one of items 1-5, wherein the at least one processor is further programmed or configured to perform the following operations: generating a first representation in frequency space of the examination area of ​​the examination object regarding a first result of a radiological examination; and generating a second representation in frequency space of the examination area of ​​the examination object regarding a second result of a radiological examination.

[0019] Item 7: The system as described in Item 6, wherein the radiological examination is a magnetic resonance imaging examination, a computed tomography examination, or an ultrasound examination.

[0020] Item 8: The system as described in Item 6, wherein the radiological examination is a magnetic resonance imaging (MRI) examination, wherein, when receiving a first representation of the examination area of ​​the examination object in the frequency space, the at least one processor is programmed or configured to: receive first k-space data associated with the MRI examination of the examination area of ​​the examination object; and wherein, when receiving a second representation of the examination area of ​​the examination object in the frequency space, the at least one processor is programmed or configured to: receive second k-space data associated with the MRI examination of the examination area of ​​the examination object.

[0021] Item 9: A system as described in any one of items 1-7, wherein, when receiving a first representation of the examination area of ​​an object in the frequency space, the at least one processor is programmed or configured to perform the following operations: receiving a first representation of the examination area of ​​the object in the physical space, wherein the first representation in the physical space includes a representation of an examination area in which no amount of contrast agent was applied during the medical imaging technique or in which a first amount of contrast agent was applied during the medical imaging technique; converting the first representation in the physical space into a first representation of the examination area of ​​the object in the frequency space; wherein, when receiving a second representation of the examination area of ​​the object in the frequency space, the at least one processor is programmed or configured to perform the following operations: receiving a second representation of the examination area of ​​the object in the physical space, wherein the second representation in the physical space includes a representation of an examination area in which a second amount of contrast agent was applied during the medical imaging technique, wherein the second amount of contrast agent is different from the first amount of contrast agent; converting the first representation in the physical space into a first representation of the examination area of ​​the object in the frequency space.

[0022] Item 10: A system as described in any one of items 1-9, wherein the at least one processor is further programmed or configured to perform the following operations: training the predictive machine learning model based on a training dataset, wherein the training dataset comprises: a set of reference representations in frequency space of the examination region of each of a plurality of examination objects, each set of reference representations in frequency space of the examination region of the examination object comprising: a first reference representation in frequency space of the examination region of the examination object; a second reference representation in frequency space of the examination region of the examination object; and a third reference representation in frequency space of the examination region of the examination object; and wherein the first reference representation comprises a reference representation in frequency space of the examination region for which no contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; wherein the second reference representation comprises a reference representation in frequency space of the examination region for which a second amount of contrast agent was applied during the medical imaging technique; and wherein the third reference representation comprises a reference representation in frequency space of the examination region for which a third amount of contrast agent was applied during the medical imaging technique.

[0023] Item 11: The system as described in Item 10, wherein, when training the predictive machine learning model, the at least one processor is programmed or configured to perform the following operation: minimizing the amount of error provided by an error function, wherein the error function quantifies the deviation between a predicted representation of the examination region of an examination subject in frequency space during medical imaging techniques, wherein a third amount of contrast agent has been applied, and a third reference representation of the examination region of the examination subject in frequency space.

[0024] Item 12: A system as described in any one of items 1-11, wherein the at least one processor is further programmed or configured to perform the following operations: training the predictive machine learning model based on a training data set, wherein the training data set comprises: a set of simplified reference representations of the inspection regions of each of a plurality of inspection objects in frequency space, each set of simplified reference representations of the inspection regions of the inspection objects comprising: a first simplified reference representation of the inspection region of the inspection object in frequency space; a second simplified reference representation of the inspection region of the inspection object in frequency space; and a third simplified reference representation of the inspection region of the inspection object in frequency space; and wherein the first simplified reference representation comprises a frequency included in the first reference representation of the inspection region in frequency space. A reference representation of a portion of the center of a space, wherein the first reference representation includes a reference representation of an examination region in which no contrast agent was applied or in which a first amount of contrast agent was applied during the medical imaging technique; wherein the second simplified reference representation includes a reference representation of the examination region in the frequency space that includes a portion of the center of the frequency space, wherein the second reference representation includes a reference representation of the examination region in the frequency space that includes a portion of the center of the frequency space, wherein the third simplified reference representation includes a reference representation of the examination region in the frequency space that includes a portion of the center of the frequency space, wherein the third reference representation includes a reference representation of the examination region in the frequency space that includes a third amount of contrast agent applied during the medical imaging technique.

[0025] Item 13: The system as described in Item 12, wherein, when training the predictive machine learning model, the at least one processor is programmed or configured to perform the following operation: minimizing the amount of error provided by an error function, wherein the error function quantifies the deviation between a predicted simplified representation of the examination region of the subject to be examined during medical imaging techniques, in frequency space, and a third simplified reference representation of the examination region of the subject to be examined in frequency space.

[0026] Item 14: A computer program product for providing a prediction of a representation of an examination region, the prediction being generated using a medical imaging technique involving contrast agents, the computer program product comprising at least one non-transitory computer-readable medium, the at least one non-transitory computer-readable medium comprising one or more instructions, the one or more instructions causing the at least one processor, when executed, to perform the following operations: receiving a first representation of an examination region of an examination subject in a frequency space, wherein the first representation includes a representation of an examination region for which no amount of contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; receiving a second representation of an examination region of an examination subject in a frequency space, wherein the second representation includes an examination region for which a second amount of contrast agent was applied during the medical imaging technique. The representation, wherein the second quantity of contrast agent is different from the first quantity of contrast agent; providing input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of the first representation and at least a portion of the second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third quantity of contrast agent has been applied during medical imaging techniques, wherein the third quantity of contrast agent is greater than the first quantity of contrast agent and the second quantity of contrast agent; receiving the output of the predictive machine learning model based on the input; converting the output of the predictive machine learning model into a predicted representation of the examination region of the examination object in physical space; and providing a predicted representation of the examination region of the examination object in physical space.

[0027] Item 15: A computer program product as described in Item 14, wherein one or more instructions further cause the at least one processor to perform the following operations: specifying a portion of the frequency space center of the first representation, a portion of the frequency space center of the second representation, or portions of the frequency space center of both the first and second representations, to provide a simplified representation in frequency space of the inspection region of the object being inspected; wherein the input to the predictive machine learning model includes the simplified representation in frequency space of the inspection region of the object being inspected; and wherein the one or more instructions causing the at least one processor to provide the input to the predictive machine learning model cause the at least one processor to perform the following operations: providing the simplified representation in frequency space of the inspection region of the object being inspected as the input to the predictive machine learning model.

[0028] Item 16: A computer program product as described in Item 15, wherein one or more instructions further cause the at least one processor to perform the following operations: supplement the output of the predictive machine learning model with: a portion of the first representation that does not contain a frequency space center and is not specified for providing a simplified representation of the inspection region of the object in frequency space; a portion of the second representation that does not contain a frequency space center and is not specified for providing a simplified representation of the inspection region of the object in frequency space; or portions of the first representation and the second representation that respectively do not contain a frequency space center and are not specified for providing a simplified representation of the inspection region of the object in frequency space; provide a supplementary output of the predictive machine learning model; wherein the one or more instructions causing the at least one processor to convert the output of the predictive machine learning model into a predicted representation of the inspection region of the object in entity space cause the at least one processor to perform the following operations: convert the supplementary output of the predictive machine learning model into a predicted representation of the inspection region of the object in entity space.

[0029] Item 17: A computer program product as described in any of Items 14-16, wherein a first quantity of contrast agent applied during medical imaging techniques is greater than zero, and wherein a second quantity of contrast agent applied during medical imaging techniques is greater than the first quantity of contrast agent.

[0030] Item 18: A computer program product as described in any of Items 14-16, wherein the first quantity of contrast agent applied during medical imaging is zero, and wherein the second quantity of contrast agent applied during medical imaging is greater than the first quantity of contrast agent.

[0031] Item 19: A computer program product as described in any one of Items 14-18, wherein the one or more instructions further cause the at least one processor to perform the following operations: generating a first representation in frequency space of the examination area of ​​the radiological examination object in relation to a first result of the radiological examination; and generating a second representation in frequency space of the examination area of ​​the radiological examination object in relation to a second result of the radiological examination.

[0032] Item 20: A computer program product as described in Item 19, wherein the radiological examination is a magnetic resonance imaging examination, a computed tomography examination, or an ultrasound examination.

[0033] Item 21: A computer program product as described in Item 19, wherein the radiological examination is a magnetic resonance imaging (MRI) examination, wherein one or more instructions causing the at least one processor to receive a first representation of the examination area of ​​the examination object in the frequency space cause the at least one processor to perform the following operations: receiving first k-space data associated with the MRI examination of the examination area of ​​the examination object; and wherein causing the at least one processor to receive a second representation of the examination area of ​​the examination object in the frequency space causes the at least one processor to perform the following operations: receiving second k-space data associated with the MRI examination of the examination area of ​​the examination object.

[0034] Item 22: A computer program product as described in any one of items 14-20, wherein the one or more instructions causing the at least one processor to receive a first representation of an examination region of an object in frequency space cause the at least one processor to perform the following operations: receiving a first representation of an examination region of an object in physical space, wherein the first representation in physical space includes a representation of an examination region for which no amount of contrast agent was applied during medical imaging or for which a first amount of contrast agent was applied during medical imaging; converting the first representation in physical space into a first representation of an examination region of an object in frequency space; wherein the one or more instructions causing the at least one processor to receive a second representation of an examination region of an object in frequency space cause the at least one processor to perform the following operations: receiving a second representation of an examination region of an object in physical space, wherein the second representation in physical space includes a representation of an examination region for which a second amount of contrast agent was applied during medical imaging, wherein the second amount of contrast agent is different from the first amount of contrast agent; converting the first representation in physical space into a first representation of an examination region of an object in frequency space.

[0035] Item 23: A computer program product as described in any one of items 14-22, wherein the one or more instructions further cause the at least one processor to perform the following operations: training the predictive machine learning model based on a training data set, wherein the training data set comprises: a set of reference representations in frequency space of examination regions of each of a plurality of examination objects, each set of reference representations of examination regions of the examination objects comprising: a first reference representation of the examination region of the examination object in frequency space; a second reference representation of the examination region of the examination object in frequency space; and a third reference representation of the examination region of the examination object in frequency space; and wherein the first reference representation in frequency space includes a reference representation of an examination region for which no contrast agent was applied during medical imaging or for which a first amount of contrast agent was applied during medical imaging; wherein the second reference representation in frequency space includes a reference representation of an examination region for which a second amount of contrast agent was applied during medical imaging; and wherein the third reference representation in frequency space includes a reference representation of an examination region for which a third amount of contrast agent was applied during medical imaging.

[0036] Item 24: A computer program product as described in Item 23, wherein the one or more instructions that cause the at least one processor to train the predictive machine learning model cause the at least one processor to perform the following operation: minimizing the amount of error provided by an error function, wherein the error function quantifies the deviation between a predicted representation of the examination region of an examination subject in frequency space during medical imaging techniques, wherein a third amount of contrast agent has been applied, and a third reference representation of the examination region of the examination subject in frequency space.

[0037] Item 25: A computer program product as described in any one of items 14-24, wherein the one or more instructions further cause the at least one processor to perform the following operations: training the predictive machine learning model based on a training data set, wherein the training data set comprises: a set of simplified reference representations in frequency space of the inspection regions of each of a plurality of inspection objects, each set of simplified reference representations of the inspection regions of the inspection objects comprising: a first simplified reference representation of the inspection region of the inspection object in frequency space; a second simplified reference representation of the inspection region of the inspection object in frequency space; and a third simplified reference representation of the inspection region of the inspection object in frequency space; and wherein the first simplified reference representation comprises a first reference representation of the inspection region in frequency space. A reference representation including a portion of the center of frequency space, wherein the first reference representation includes a reference representation of an examination region in which no contrast agent was applied or in which a first amount of contrast agent was applied during medical imaging; wherein the second simplified reference representation includes a reference representation of an examination region in frequency space including a portion of the center of frequency space, wherein the second reference representation includes a reference representation of an examination region in which a second amount of contrast agent was applied during medical imaging; wherein the third simplified reference representation includes a reference representation of an examination region in frequency space including a portion of the center of frequency space, wherein the third reference representation includes a reference representation of an examination region in which a third amount of contrast agent was applied during medical imaging.

[0038] Item 26: A computer program product as described in Item 25, wherein the one or more instructions that cause the at least one processor to train the predictive machine learning model cause the at least one processor to perform the following operation: minimizing the amount of error provided by an error function, wherein the error function quantifies the deviation between a predicted simplified representation of the examination region of the subject to be examined during medical imaging techniques, in frequency space, and a third simplified reference representation of the examination region of the subject to be examined in frequency space.

[0039] Item 27: A method for providing a prediction of an examination region, the prediction being generated using a medical imaging technique involving contrast agents, the method comprising: receiving, with at least one processor, a first representation of an examination region of an examination subject in a frequency space, wherein the first representation includes a representation of an examination region for which no amount of contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; receiving, with the at least one processor, a second representation of the examination region of the examination subject in a frequency space, wherein the second representation includes a representation of an examination region for which a second amount of contrast agent was applied during the medical imaging technique, wherein the second amount of contrast agent is different from the first amount of contrast agent; and providing, with the at least one processor, input to a predictive machine learning model, wherein the predictive machine learning model... The input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent has been applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent; receiving the output of the predictive machine learning model based on the input using the at least one processor; converting the output of the predictive machine learning model into a predicted representation in entity space of the examination region of the examination object using the at least one processor; and providing the predicted representation in entity space of the examination region of the examination object using the at least one processor.

[0040] Item 28: The method of Item 27 further comprises: specifying that the first representation includes a portion of the frequency space center, the second representation includes a portion of the frequency space center, or the first representation and the second representation respectively include portions of the frequency space center, providing a simplified representation of the inspection region of the object in the frequency space; wherein the input of the predictive machine learning model includes the simplified representation of the inspection region of the object in the frequency space; and wherein providing the input of the predictive machine learning model comprises: providing the simplified representation of the inspection region of the object in the frequency space as the input of the predictive machine learning model.

[0041] Item 29: The method of Item 28 further comprises: supplementing the output of the predictive machine learning model with the following: a portion of the first representation that does not contain a frequency space center and is not designated to provide a simplified representation of the inspection region of the object in frequency space; a portion of the second representation that does not contain a frequency space center and is not designated to provide a simplified representation of the inspection region of the object in frequency space; or portions of the first representation and the second representation that respectively do not contain a frequency space center and are not designated to provide a simplified representation of the inspection region of the object in frequency space, to provide supplementary output of the predictive machine learning model; wherein converting the output of the predictive machine learning model into a representation of the inspection region of the object in entity space comprises: converting the supplementary output of the predictive machine learning model into a representation of the inspection region of the object in entity space.

[0042] Item 30: The method of any one of items 27-29, wherein the first quantity of contrast agent applied during the medical imaging technique is greater than zero, and wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

[0043] Item 31: The method of any one of items 27-29, wherein the first quantity of contrast agent applied during the medical imaging technique is zero, and wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

[0044] Item 32: The method of any one of items 27-31 further comprises: generating a first representation in frequency space of the examination area of ​​the radiological examination subject in relation to a first result of the radiological examination; and generating a second representation in frequency space of the examination area of ​​the radiological examination subject in relation to a second result of the radiological examination.

[0045] Item 33: The method as described in Item 32, wherein the radiological examination is a magnetic resonance imaging examination, a computed tomography examination, or an ultrasound examination.

[0046] Item 34: The method of Item 32, wherein the radiological examination is a magnetic resonance imaging examination, wherein receiving a first representation of the examination area of ​​the examination object in the frequency space includes: receiving first k-space data associated with the magnetic resonance imaging examination of the examination area of ​​the examination object; and wherein receiving a second representation of the examination area of ​​the examination object in the frequency space includes: receiving second k-space data associated with the magnetic resonance imaging examination of the examination area of ​​the examination object.

[0047] Item 35: The method of any one of items 27-33, wherein a first representation of the examination area of ​​the subject in the frequency space comprises: a first representation of the examination area of ​​the subject in the physical space, wherein the first representation in the physical space comprises a representation of the examination area in which no amount of contrast agent was applied during the medical imaging technique or in which a first amount of contrast agent was applied during the medical imaging technique; converting the first representation in the physical space into a first representation of the examination area of ​​the subject in the frequency space; and wherein a second representation of the examination area of ​​the subject in the frequency space comprises: a second representation of the examination area of ​​the subject in the physical space, wherein the second representation in the physical space comprises a representation of the examination area in which a second amount of contrast agent was applied during the medical imaging technique, wherein the second amount of contrast agent is different from the first amount of contrast agent; converting the first representation in the physical space into a first representation of the examination area of ​​the subject in the frequency space.

[0048] Item 36: The method of any one of items 27-35, further comprising: training the predictive machine learning model based on a training data set, wherein the training data set comprises: a set of reference representations of the examination regions of each of a plurality of examination objects in frequency space, each set of reference representations of the examination regions of the examination objects in frequency space comprising: a first reference representation of the examination region of the examination object in frequency space; a second reference representation of the examination region of the examination object in frequency space; and a third reference representation of the examination region of the examination object in frequency space; and wherein the first reference representation comprises a reference representation of an examination region for which no contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique; wherein the second reference representation comprises a reference representation of an examination region for which a second amount of contrast agent was applied during the medical imaging technique; and wherein the third reference representation comprises a reference representation of an examination region for which a third amount of contrast agent was applied during the medical imaging technique.

[0049] Item 37: The method as described in Item 36, wherein training the predictive machine learning model is based on: minimizing an error amount provided by an error function, wherein the error function quantifies the deviation between a predicted representation of the examination region of the subject in the frequency space during medical imaging techniques, wherein a third amount of contrast agent has been applied, and a third reference representation of the examination region of the subject in the frequency space.

[0050] Item 38: The method of any one of items 27-37, further comprising: training the predictive machine learning model based on a training data set, wherein the training data set comprises: a set of simplified reference representations of the inspection regions of each of a plurality of inspection objects in frequency space, each set of simplified reference representations of the inspection regions of the inspection objects in frequency space comprising: a first simplified reference representation of the inspection regions of the inspection objects in frequency space; a second simplified reference representation of the inspection regions of the inspection objects in frequency space; and a third simplified reference representation of the inspection regions of the inspection objects in frequency space; and wherein the first simplified reference representation comprises a portion of the frequency space center included in the first reference representation of the inspection region in frequency space. The reference representations are as follows: the first reference representation includes a reference representation of an examination area in which no contrast agent was applied or in which a first amount of contrast agent was applied during the medical imaging technique; the second simplified reference representation includes a reference representation of the examination area in the frequency space that includes a portion of the center of the frequency space, wherein the second reference representation includes a reference representation of the examination area in the frequency space that includes a portion of the center of the frequency space, wherein the third simplified reference representation includes a reference representation of the examination area in the frequency space that includes a portion of the center of the frequency space, wherein the third reference representation includes a reference representation of the examination area in the frequency space that includes a third amount of contrast agent applied during the medical imaging technique.

[0051] Item 39: The method as described in Item 38, wherein training the predictive machine learning model is based on: minimizing an error amount provided by an error function, wherein the error function quantifies the deviation between a prediction of a simplified representation of the examination region of the subject in the frequency space during medical imaging techniques, and a third simplified reference representation of the examination region of the subject in the frequency space.

[0052] These and other features of this disclosure, as well as the methods of operation and functions of the related structural elements and the economy of combination and manufacture of components, will become more apparent upon reference to the accompanying drawings, which form part of this specification, wherein like reference numerals denote corresponding components in the figures. However, it should be clearly understood that the drawings are for illustrative purposes only and are not intended to limit the scope of this disclosure. As used in the detailed description and claims, the singular forms “a,” “an,” and “the” include the plural forms unless the context clearly specifies otherwise. Attached Figure Description

[0053] The additional advantages and details of the non-limiting embodiments are explained in more detail below with reference to the exemplary embodiments illustrated in the accompanying drawings, wherein:

[0054] Figure 1 This is a schematic diagram of a non-limiting implementation of an environment in which the systems, methods, and / or computer program products described herein can be implemented in accordance with this disclosure;

[0055] Figure 2 yes Figure 1 A schematic diagram of a non-limiting embodiment of a component of one or more devices and / or one or more systems;

[0056] Figure 3 This is a flowchart of a non-limiting embodiment of a process for providing a prediction of an examination area, the prediction being generated using medical imaging techniques involving contrast agents;

[0057] Figure 4 This is a flowchart of a non-limiting implementation of a process for training a predictive machine learning model to provide a representation of the examined region for prediction;

[0058] Figures 5A to 5C This is a schematic diagram of a non-limiting embodiment of a process for providing a representation of an examination area, said prediction being generated using medical imaging techniques involving contrast agents; and

[0059] Figures 6A to 6B This is a schematic diagram of a non-limiting implementation of a process for providing a representation of an examination area, the prediction being generated using medical imaging techniques involving contrast agents. Detailed Implementation

[0060] For ease of description below, the terms “end,” “upper,” “lower,” “right,” “left,” “vertical,” “horizontal,” “top,” “bottom,” “lateral,” “longitudinal,” and their derivatives shall be used in relation to the orientation of this disclosure in the accompanying drawings. However, it should be understood that this disclosure may have many alternative variations and sequences of steps unless expressly stated otherwise. It should also be understood that the specific devices and processes illustrated in the accompanying drawings and described in the following specification are merely exemplary embodiments or aspects of this disclosure. Therefore, specific dimensions and other physical characteristics relating to the embodiments or aspects of the embodiments disclosed herein should not be considered limiting unless otherwise indicated.

[0061] The terms "aspect," "component," "element," "structure," "behavior," "step," "function," "instruction," and / or similar as used herein should not be construed as critical or essential unless explicitly stated otherwise. Furthermore, as used herein, the articles "a" and "an" mean to include one or more items and may be used interchangeably with "one or more" and "at least one." The singular forms of "a," "an," and "the" as used in the specification and claims include plural references, for example, unless the context clearly specifies otherwise. Furthermore, as used herein, the terms "set" and "group" are intended to include one or more items (e.g., related items, unrelated items, combinations of related and unrelated items, etc.) and may be used interchangeably with "one or more" or "at least one." The term "one" or similar language is used if referring to only one item. Furthermore, as used herein, the terms "have," "has," "having," or similar terms are intended as open-ended terms. Additionally, the phrase "based on" is intended to mean "at least partially based on," unless explicitly stated otherwise. Furthermore, the phrase “based on” can mean “in response to” and indicates the conditions for automatically triggering the specific operation of electronic devices (e.g., controllers, processors, computing devices, etc.) appropriately mentioned herein.

[0062] As used herein, the term "system" can refer to one or more computing devices or a combination of computing devices, such as, but not limited to, processors, servers, client devices, software applications, and / or other similar components. Furthermore, as used herein, references to "server" or "processor" can refer to the previously cited server and / or processor referred to as performing a preceding step or function, different servers and / or processors, and / or combinations of servers and / or processors. For example, as used in the specification and claims, a first server and / or processor referred to as performing a first step or function can refer to the same or different server and / or processor referred to as performing a second step or function.

[0063] As used herein, the terms "communication" and "communicate" can refer to the receipt, transmission, transfer, provision, and / or similar of information (e.g., data, signals, messages, instructions, commands, and / or similar information). For one unit (e.g., a device, system, component of a device or system, combinations thereof, and / or the like) to communicate with another unit means that the first unit is able to receive information directly or indirectly from and / or transmit information to the other unit. This can refer to a direct or indirect connection that is inherently wired and / or wireless. Furthermore, two units can communicate with each other, even if the transmitted information may be modified, processed, relayed, and / or routed between the first and second units. For example, the first unit can communicate with the second unit even if the first unit passively receives information and does not actively send information to the second unit. As in another embodiment, the first unit can communicate with the second unit if at least one intermediate unit (e.g., a third unit located between the first and second units) processes information received from the first unit and sends the processed information to the second unit. In some non-limiting embodiments, information may refer to network packets (e.g., data packets and / or the like) that include data.

[0064] This document describes some non-limiting implementations related to thresholds. As used herein, a threshold can refer to a value that is greater than, more than, higher than, greater than or equal to, less than, less than, lower than, less than or equal to, or equal to the threshold.

[0065] In some non-limiting embodiments, the machine learning model can be used as a method to reduce the amount of contrast agent used to generate radiographic images. For example, during a first step of the process, a training dataset can be generated. For multiple individuals (e.g., patients, people, etc.), the training dataset may include each individual's original radiographic image (e.g., a radiographic image generated during medical imaging without a certain amount of contrast agent applied, such as a zero-contrast image), a low-contrast radiographic image generated during medical imaging based on the application of a low amount of contrast agent (e.g., subsequently), and a full-contrast radiographic image generated during medical imaging based on the application of a standard amount of contrast agent (e.g., a full-contrast image). In a second step, a machine learning model (such as an artificial neural network) can be trained to predict a predicted radiographic image (e.g., an artificial radiographic image) for each individual included in the training dataset based on the original radiographic image and the low-contrast radiographic image, which shows the examined area after the application of a standard amount of contrast agent. When training the predictive machine learning model, the full-contrast radiographic image is used as a reference (e.g., a baseline ground truth) for each individual. In the third step, the trained machine learning model can be used to predict a predicted radiographic image for a new individual based on the original radiographic image and the low-contrast radiographic image. This predicted radiographic image shows the examination area of ​​the new individual as it would be during the medical imaging procedure with a standard amount of contrast agent already applied. A similar process can be found in International Patent Application No. PCT / US2018 / 055034, filed October 9, 2018, which is incorporated herein by reference in its entirety.

[0066] However, during this process, co-registration may be required to match individual radiographic images from different individuals, ensuring pixel / voxel correspondence. For example, co-registration of radiographic images may be necessary so that image elements (e.g., pixels or voxels) from a radiographic image of an individual's examination area correspond (e.g., are shown as identical) to image elements from another radiographic image of the same individual's examination area. If the image elements of the radiographic images do not correspond, artifacts may appear in the predicted radiographic image, potentially obscuring, distorting, and / or mimicking anatomical structures in the examination area.

[0067] Furthermore, the process described above may require training the predictive machine learning model and generating the predicted radiographic images using complete radiographic images. For example, training the predictive machine learning model and generating the predicted radiographic images may require using complete radiographic images after various amounts of contrast agent have been applied. In this way, the computational complexity for generating the predicted radiographic images increases rapidly. Moreover, the computation of the predicted radiographic images may take a significant amount of time and may require expensive hardware to complete the computation within the desired timeframe. Additionally, the complete radiographic images may need to be scaled down to local regions (e.g., patches), and these local regions may need to be processed separately to avoid overloading computer memory with excessively large radiographic images. However, when these local regions are processed separately and reconnected to form a complete radiographic image, this approach may result in artifacts (such as stitching artifacts) at the interfaces. The subsequent removal of these stitching artifacts may require additional computational resources. Furthermore, this method may introduce errors into the predicted radiographic images, which radiologists may misinterpret, potentially leading to misdiagnosis based on the complete radiographic images.

[0068] This invention provides improved systems, methods, and computer program products for providing predictions of representations of examination regions, the predictions being generated using medical imaging techniques involving contrast agents. Embodiments of this disclosure may include a system comprising at least one processor programmed or configured to: receive a first representation of an examination region of an object in a frequency space, wherein the first representation includes a representation of an examination region during the medical imaging technique without the application of a certain amount of contrast agent or with the application of a first amount of contrast agent during the medical imaging technique; receive a second representation of the examination region of the object in a frequency space, wherein the second representation includes a representation of an examination region during the medical imaging technique with the application of a second amount of contrast agent, wherein the second amount of contrast agent is different from the first amount of contrast agent; and provide input to a predictive machine learning model, wherein the... The input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent has been applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent; receive the output of the predictive machine learning model based on the input; convert the output of the predictive machine learning model into a predicted representation in entity space of the examination region of the examination object; and provide a predicted representation in entity space of the examination region of the examination object.

[0069] In some non-limiting embodiments, the at least one processor is further programmed or configured to perform the following operations: specifying a portion of the frequency space center of the first representation, a portion of the frequency space center of the second representation, or portions of the frequency space center of both the first and second representations, providing a simplified representation of the inspection region of the object in the frequency space, wherein the input to the predictive machine learning model includes the simplified representation of the inspection region of the object in the frequency space, and wherein, when the input to the predictive machine learning model is provided, the at least one processor is programmed or configured to provide the simplified representation of the inspection region of the object in the frequency space as input to the predictive machine learning model. In some non-limiting embodiments, the at least one processor is further programmed or configured to perform the following operations: supplement the output of the predictive machine learning model to provide a supplementary output of the predictive machine learning model: a first representation that does not contain a frequency space center and is not selected to provide a simplified representation of the inspection region of the object in frequency space; a second representation that does not contain a frequency space center and is not selected to provide a simplified representation of the inspection region of the object in frequency space; or both the first and second representations that do not contain a frequency space center and are not selected to provide a simplified representation of the inspection region of the object in frequency space, wherein, when the output of the predictive machine learning model is converted into a predicted representation of the inspection region of the object in entity space, the at least one processor is programmed or configured to convert the supplementary output of the predictive machine learning model into a predicted representation of the inspection region of the object in entity space.

[0070] In some non-limiting embodiments, the first quantity of contrast agent applied during the medical imaging technique is greater than zero, wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

[0071] In some non-limiting embodiments, the first quantity of contrast agent applied during the medical imaging technique is zero, wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent (i.e., greater than zero).

[0072] In some non-limiting embodiments, the at least one processor is further programmed or configured to generate a first representation in frequency space of the examined area of ​​the radiological examination subject with respect to a first result of the radiological examination, and to generate a second representation in frequency space of the examined area of ​​the radiological examination subject with respect to a second result of the radiological examination. In some non-limiting embodiments, the radiological examination is a magnetic resonance imaging (MRI) examination, a computed tomography (CT) scan, or an ultrasound examination.

[0073] In some non-limiting embodiments, when receiving a first representation of the examination area of ​​the subject in the frequency space, the at least one processor is programmed or configured to receive first k-space data associated with a magnetic resonance imaging (MRI) examination of the examination area of ​​the subject, and wherein, when receiving a second representation of the examination area of ​​the subject in the frequency space, the at least one processor is programmed or configured to receive second k-space data associated with an MRI examination of the examination area of ​​the subject.

[0074] In some non-limiting embodiments, when receiving a first representation of the examination area of ​​the subject in the frequency space, the at least one processor is programmed or configured to perform the following operations: receiving a first representation of the examination area of ​​the subject in the physical space, wherein the first representation in the physical space includes a representation of the examination area during medical imaging techniques where no amount of contrast agent was applied or a first amount of contrast agent was applied during medical imaging techniques; converting the first representation in the physical space into a first representation of the examination area of ​​the subject in the frequency space; and wherein, when receiving a second representation of the examination area of ​​the subject in the frequency space, the at least one processor is programmed or configured to perform the following operations: receiving a second representation of the examination area of ​​the subject in the physical space, wherein the second representation in the physical space includes a representation of the examination area during medical imaging techniques where a second amount of contrast agent was applied, wherein the second amount of contrast agent is different from the first amount of contrast agent; converting the first representation in the physical space into a first representation of the examination area of ​​the subject in the frequency space.

[0075] In some non-limiting embodiments, the at least one processor is further programmed or configured to train the predictive machine learning model based on a training dataset, wherein the training dataset comprises a set of reference representations in frequency space of the examination region of each of a plurality of examination objects, each set of reference representations of the examination region of the examination object comprising a first reference representation, a second reference representation, and a third reference representation in frequency space of the examination region of the examination object, and wherein the first reference representation comprises a reference representation of the examination region for which no contrast agent was applied or for which a first amount of contrast agent was applied during the medical imaging technique, wherein the second reference representation comprises a reference representation of the examination region for which a second amount of contrast agent was applied during the medical imaging technique, and wherein the third reference representation comprises a reference representation of the examination region for which a third amount of contrast agent was applied during the medical imaging technique.

[0076] In some non-limiting embodiments, when training the predictive machine learning model, the at least one processor is programmed or configured to minimize the amount of error provided by an error function, wherein the error function quantifies the deviation between a predicted representation of the examination region of the subject in frequency space during medical imaging techniques, where a third amount of contrast agent has been applied, and a third reference representation of the examination region of the subject in frequency space.

[0077] In some non-limiting embodiments, the at least one processor is further programmed or configured to train the predictive machine learning model based on a training dataset, wherein the training dataset comprises a set of simplified reference representations in frequency space of the inspection region of each of a plurality of inspection objects, each set of simplified reference representations in frequency space comprising a first simplified reference representation, a second simplified reference representation, and a third simplified reference representation in frequency space of the inspection region of the inspection object, and wherein the first simplified reference representation comprises a reference to the first reference representation of the inspection region in frequency space that includes a portion of the center of frequency space. The first reference representation includes a reference representation of an examination area in which no contrast agent was applied or in which a first amount of contrast agent was applied during the medical imaging technique. The second simplified reference representation includes a reference representation of the examination area in the frequency space that includes a portion of the center of the frequency space of the second reference representation in the frequency space. The second reference representation includes a reference representation of an examination area in which a second amount of contrast agent was applied during the medical imaging technique. The third simplified reference representation includes a reference representation of the examination area in the frequency space that includes a portion of the center of the frequency space of the third reference representation in the frequency space. The third reference representation includes a reference representation of an examination area in which a third amount of contrast agent was applied during the medical imaging technique.

[0078] In some non-limiting embodiments, when training the predictive machine learning model, the at least one processor is programmed or configured to minimize the amount of error provided by an error function, wherein the error function quantifies the deviation between a predicted simplified representation of the examination region of the subject to be examined during medical imaging techniques, in frequency space, and a third simplified reference representation of the examination region of the subject to be examined in frequency space.

[0079] In this way, the system can provide a predictive representation of the examined region of the object in physical space, which can reduce the use of contrast agent to provide this representation. Furthermore, the system can provide a predictive representation of the examined region of the object in physical space in a manner tolerable in co-registration errors, thereby avoiding the risk of artifacts (especially stitching artifacts). Moreover, compared to systems that do not use machine learning models configured to provide predictions of the representation of the examined region in frequency space during medical imaging with a specific amount of contrast agent applied, the system may require fewer computational resources to provide the predictive representation, and the system can match the computational complexity to the system's hardware configuration. Furthermore, by using a simplified representation of the inspection region of the inspection object in the frequency space to provide a predictive representation in the entity space and / or by using a simplified reference representation of the inspection region of the inspection object in the frequency space to train a machine learning model, this system can reduce the amount of time required to provide a predictive representation in the entity space and / or to train a machine learning model compared to systems that do not use a simplified representation of the inspection region of the inspection object in the frequency space to provide a predictive representation in the entity space or do not use a simplified reference representation of the inspection region of the inspection object in the frequency space to train a machine learning model.

[0080] Now for reference Figure 1 , Figure 1 This is a schematic diagram of a non-limiting embodiment of an environment 100 in which the apparatus, system, method, and / or computer program product described herein can be implemented. For example... Figure 1 As shown, environment 100 includes a radiology machine learning system 102, a medical imaging system 104, and a user equipment 106. In some non-limiting embodiments, the radiology machine learning system 102 may be interconnected with the medical imaging system 104 and / or the user equipment 106 via a communication network 108 (e.g., establishing a connection to communicate with it and / or similar). In some non-limiting embodiments, the radiology machine learning system 102 may be interconnected with the medical imaging system 104 and the user equipment 106 via a wired connection, a wireless connection, or a combination of wired and wireless connections.

[0081] The radiology machine learning system 102 may include one or more computing devices configured to communicate with the medical imaging system 104 and / or user equipment 106 via a communication network 108. For example, the radiology machine learning system 102 may include a server, a group of servers, and / or other similar devices. In some non-limiting embodiments, the radiology machine learning system 102 may be a component of the medical imaging system 104. In some non-limiting embodiments, the radiology machine learning system 102 may include a cloud computing system.

[0082] The medical imaging system 104 may include one or more computing devices capable of communicating with the radiological machine learning system 102 and / or user equipment 106 via a communication network 108. For example, the medical imaging system 104 may include one or more scanners, such as CT scanners and / or MRI scanners, capable of communicating via the communication network 108 and performing medical imaging procedures involving the use of contrast agents (such as radiographic contrast materials).

[0083] User equipment 106 may include one or more computing devices configured to communicate with radiological machine learning system 102 and / or medical imaging system 104 via communication network 108. For example, user equipment 106 may include a desktop computer (e.g., a client device communicating with a server) and / or a portable computer, such as a laptop, tablet, smartphone, and / or similar device. In some non-limiting embodiments or aspects, user equipment 106 may be associated with a user (e.g., an individual operating the device).

[0084] The communication network 108 may include one or more wired and / or wireless networks. For example, the communication network 108 may include cellular networks (e.g., Long Term Evolution (LTE) networks, third-generation (3G) networks, fourth-generation (4G) networks, fifth-generation (5G) networks, Code Division Multiple Access (CDMA) networks and / or similar networks), local area networks (LANs), wide area networks (WANs), wireless local area networks (WLANs), private networks, ad hoc networks, intranets, the Internet, fiber optic networks, Ethernet, Universal Serial Bus (USB) networks, cloud computing networks and / or similar networks, and / or some or all of these networks or other types of networks.

[0085] Figure 1 The number and arrangement of systems and / or equipment shown are provided as examples. Figure 1 Compared to the systems and / or devices shown, there may be additional systems and / or devices, fewer systems and / or devices, different systems and / or devices, or systems and / or devices with different arrangements. Furthermore, Figure 1 The two or more systems and / or devices shown can be implemented in a single system or a single device, or Figure 1 The single system or single device shown may be implemented as multiple distributed systems or devices. Additionally or alternatively, a group of systems or a group of devices in environment 100 (e.g., one or more systems, one or more devices) may perform one or more functions described as being performed by another group of systems or another group of devices in environment 100.

[0086] Now for reference Figure 2 , Figure 2 This is a schematic diagram of example components of device 200. Device 200 may correspond to radiology machine learning system 102, medical imaging system 104, and / or user equipment 106. In some non-limiting embodiments, radiology machine learning system 102, medical system 104, and / or user equipment 106 may include at least one device 200 and / or at least one component of device 200. Figure 2 As shown, device 200 may include bus 202, processor 204, memory 206, storage unit 208, input unit 210, output unit 212 and communication interface 214.

[0087] Bus 202 may include components that allow communication between parts of device 200. In some non-limiting embodiments, processor 204 may be implemented in hardware, software, or a combination of hardware and software. For example, processor 204 may include a processor (e.g., a central processing unit (CPU), graphics processing unit (GPU), accelerated processing unit (APU), and / or similar device), microprocessor, digital signal processor (DSP), and / or any processing component that can be programmed to perform functions (e.g., a field-programmable gate array (FPGA), application-specific integrated circuit (ASIC), and / or similar device). Memory 206 may include random access memory (RAM), read-only memory (ROM), and / or another type of dynamic or static storage device for storing information and / or instructions used by processor 204 (e.g., flash memory, magnetic storage, optical storage, and / or similar devices).

[0088] Storage component 208 may store information and / or software related to the operation and use of device 200. For example, storage component 208 may include hard disks (such as magnetic disks, optical disks, magneto-optical disks, solid-state disks and / or the like), optical disks (CDs), digital versatile optical disks (DVDs), floppy disks, magnetic tape cassettes and / or other types of computer-readable media and corresponding drives.

[0089] Input component 210 may include components that allow device 200 to receive information, such as via user input (e.g., a touchscreen display, keyboard, keypad, mouse, buttons, switches, microphone, camera, and / or similar devices). Additionally or alternatively, input component 210 may include sensors for sensing information (e.g., a Global Positioning System (GPS) component, accelerometer, gyroscope, actuator, and / or similar device). Output component 212 may include components that provide output information from device 200 (e.g., a display, speaker, one or more light-emitting diodes (LEDs), and / or similar devices).

[0090] Communication interface 214 may include transceiver-like components (e.g., a transceiver, separate receiver and transmitter, and / or similar components) that enable device 200 to communicate with other devices, such as via a wired connection, a wireless connection, or a combination of wired and wireless connections. Communication interface 214 may allow device 200 to receive information from and / or provide information to other devices. For example, communication interface 214 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, etc. Interfaces, cellular network interfaces and / or similar interfaces.

[0091] Device 200 can perform one or more of the processes described herein. Device 200 can perform these processes based on software instructions stored on a computer-readable medium (such as memory 206 and / or storage component 208) executed by processor 204. A computer-readable medium (e.g., a non-transitory computer-readable medium) is defined herein as a non-transitory storage device. A non-transitory storage device may include storage space located within a single physical storage device or storage space distributed across multiple physical storage devices.

[0092] Software instructions may be read into memory 206 and / or storage unit 208 from another computer-readable medium or another device via communication interface 214. Upon execution, the software instructions stored in memory 206 and / or storage unit 208 cause processor 204 to perform one or more processes described herein. Additionally or alternatively, hardwired circuitry may be used in place of or in combination with software instructions to perform one or more processes described herein. Therefore, the embodiments or aspects described herein are not limited to any particular combination of hardware circuitry and software.

[0093] The memory 206 and / or storage component 208 may include data storage or one or more data structures (e.g., a database and / or similar structures). The device 200 is capable of retrieving information from the data storage or one or more data structures in the memory 206 and / or storage component 208, storing information in the data storage or one or more data structures in the memory 206 and / or storage component 208, or searching for information stored in the data storage or one or more data structures in the memory 206 and / or storage component 208.

[0094] Figure 2 The number and arrangement of components shown are provided as examples. In some non-limiting embodiments, with Figure 2Compared to the components shown, device 200 may include additional components, fewer components, different components, or components arranged differently. Additionally or alternatively, a set of components of device 200 (e.g., one or more components) may perform one or more functions as described herein that are performed by another set of components of device 200.

[0095] Now for reference Figure 3 , Figure 3 This is a flowchart of a non-limiting embodiment of a process 300 for providing a prediction of an examination area, the prediction being generated using medical imaging techniques involving contrast agents. In some non-limiting embodiments or aspects, one or more functions described with respect to process 300 may (e.g., wholly, partially, etc.) be performed by the radiographic machine learning system 102. In some non-limiting embodiments or aspects, one or more steps of process 300 may (e.g., wholly, partially, and / or similarly) be performed by another device or set of devices independent of and / or including the radiographic machine learning system 102, such as the medical imaging system 104 and / or user equipment 106.

[0096] like Figure 3 As shown, in step 302, process 300 may include receiving a first representation of the examination region in frequency space. For example, the radiology machine learning system 102 may receive a first representation of the examination region (e.g., field of view (FOV), image volume, etc.) in frequency space from the medical imaging system 104. In some non-limiting embodiments, the radiology machine learning system 102 may receive a first representation of the examination region of the subject in frequency space from the medical imaging system 104. In some non-limiting embodiments, the first representation may include a representation in frequency space of the examination region for which no amount of contrast agent was applied during the medical imaging technique (e.g., a medical imaging technique to be performed during a radiological examination) or for which a first amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the first amount of contrast agent applied during the medical imaging technique is greater than zero.

[0097] In some non-limiting embodiments, the object of examination may include a biological organism, such as a mammal. In one embodiment, the object of examination may include a human. In some non-limiting embodiments, the examination area may be part of a radiological examination of the object (e.g., a radiological examination involving medical imaging techniques). In one embodiment, the examination area may include an organ or part of an organ of the object. In some non-limiting embodiments, the radiological examination may include magnetic resonance imaging (MRI), computed tomography (CT), and / or ultrasound. In some non-limiting embodiments, the examination area may include a volume imaged in a radiological image. The examination area may be defined by a radiologist (e.g., manually defined by a radiologist), for example, defined on an overview image (locator). Additionally or alternatively, the examination area may also be automatically defined (e.g., automatically defined by the radiological machine learning system 102, automatically defined by the medical imaging system 104, etc.). For example, the examination area may be automatically defined based on a protocol (e.g., a specific protocol of the medical imaging system 104).

[0098] In some non-limiting embodiments, the first representation of the examination area in frequency space may include data associated with a radiological examination performed by the medical imaging system 104. For example, the first representation of the examination area in frequency space may include k-space data associated with an MRI examination performed by the medical imaging system 104 (e.g., an MRI examination of the examination area of ​​the subject).

[0099] In some non-limiting embodiments, the radiometric machine learning system 102 can convert a first representation of the examined region in frequency space into a first representation of the examined region in entity space (e.g., image space). For example, the radiometric machine learning system 102 can use an inverse Fourier transform to convert the first representation of the examined region in frequency space into a first representation of the examined region in entity space.

[0100] In some non-limiting embodiments, the radiometric machine learning system 102 can generate a first representation of the examination region in frequency space. For example, the radiometric machine learning system 102 can receive a first representation of the examination region in entity space (e.g., a first radiometric image of the examination region), and the radiometric machine learning system 102 can generate a first representation of the examination region in frequency space based on the first representation of the examination region in entity space using a Fourier transform. In some non-limiting embodiments, the radiometric machine learning system 102 can receive the first representation of the examination region in entity space as a two-dimensional (2D) radiometric image or a three-dimensional radiometric image of the examination region in entity space, and the radiometric machine learning system 102 can generate a first representation of the examination region in frequency space based on the first representation of the examination region in entity space using a 2D Fourier transform or a 3D Fourier transform (as applicable).

[0101] like Figure 3 As shown, in step 304, process 300 may include receiving a second representation of the examination region in frequency space. For example, the radiology machine learning system 102 may receive a second representation of the examination region in frequency space (e.g., the same examination region as the first representation) from the medical imaging system 104. In some non-limiting embodiments, the radiology machine learning system 102 may receive a second representation of the examination region of an examination object (e.g., the same examination object as the examination region in the first representation) in frequency space from the medical imaging system 104. In some non-limiting embodiments, the second representation may include a representation of the examination region to which a second amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the second amount of contrast agent is different from the first amount of contrast agent applied during the medical imaging technique and associated with the first representation. In some non-limiting embodiments, the second amount of contrast agent is greater than the first amount of contrast agent applied during the medical imaging technique.

[0102] In some non-limiting embodiments, the second representation of the examination area in frequency space may include data associated with a radiological examination performed by the medical imaging system 104. For example, the second representation of the examination area in frequency space may include k-space data associated with an MRI examination performed by the medical imaging system 104.

[0103] In some non-limiting embodiments, the radiology machine learning system 102 may receive k-space data associated with a first representation of the examination region in frequency space and k-space data associated with a second representation of the examination region in frequency space. For example, the radiology machine learning system 102 may receive the first representation of the examination region in frequency space as first k-space data associated with an MRI examination of the examination region (e.g., first k-space data generated by the MRI examination), and the radiology machine learning system 102 may receive the second representation of the examination region in frequency space as second k-space data associated with an MRI examination of the examination region (e.g., second k-space data generated by the MRI examination). In some non-limiting embodiments, the radiology machine learning system 102 may generate the first representation of the examination region in frequency space for a first result of a radiology examination, and the radiology machine learning system 102 may generate the second representation of the examination region in frequency space for a second result of a radiology examination. For example, the radiology machine learning system 102 may generate the first representation of the examination region in frequency space based on the k-space data associated with the first result of the MRI examination, and the radiology machine learning system 102 may generate the second representation of the examination region in frequency space based on the k-space data associated with the second result of the MRI examination. In some non-limiting embodiments, the first representation of the examination region in frequency space may include first k-space data associated with an MRI examination of the examination region and / or the second representation of the examination region in frequency space may include second k-space data associated with an MRI examination of the examination region.

[0104] In some non-limiting embodiments, the radiometric machine learning system 102 can convert a second representation of the examined region in frequency space into a second representation of the examined region in entity space. For example, the radiometric machine learning system 102 can use an inverse Fourier transform to convert the second representation of the examined region in frequency space into a second representation of the examined region in entity space.

[0105] In some non-limiting embodiments, the radiometric machine learning system 102 can generate a second representation of the examination region in the frequency space. For example, the radiometric machine learning system 102 can receive a second representation of the examination region in the entity space (e.g., a first radiometric image of the examination region), and the radiometric machine learning system 102 can generate a second representation of the examination region in the frequency space based on the second representation of the examination region in the entity space using a Fourier transform. In some non-limiting embodiments, the radiometric machine learning system 102 can receive the second representation of the examination region in the entity space as a two-dimensional (2D) radiometric image or a three-dimensional (3D) radiometric image of the examination region, and the radiometric machine learning system 102 can generate a second representation of the examination region in the frequency space based on the second representation of the examination region in the entity space using a 2D Fourier transform or a 3D Fourier transform (as applicable).

[0106] like Figure 3 As shown, in step 306, process 300 may include providing a first representation and a second representation as input to a predictive machine learning model. For example, radiological machine learning system 102 may provide input to a predictive machine learning model (e.g., a prediction algorithm, prediction model, prediction paradigm, etc.), wherein the input to the predictive machine learning model includes at least a portion of the first representation of the examination region in frequency space and at least a portion of the second representation of the examination region in frequency space. In some non-limiting embodiments, the predictive machine learning model may include a trained machine learning model configured to provide a prediction as output of the representation of the examination region in frequency space of an examination subject for which a third amount of contrast agent has been applied during a medical imaging technique (e.g., the same medical imaging technique associated with the first and second representations of the examination region in frequency space). In some non-limiting embodiments, the prediction of the representation of the examination area in frequency space of the examination area to which a third amount of contrast agent has been applied during medical imaging techniques includes a prediction of the examination area of ​​the subject, because if a specific amount of contrast agent (a standard amount of contrast agent to be applied to the subject based on the parameters of the medical imaging techniques and / or the examination area) has already been applied to the examination area (e.g., the examination area of ​​the subject), then the examination area will be represented without the need for that specific amount to actually be applied during the medical imaging techniques.

[0107] In some non-limiting embodiments, the third quantity of contrast agent differs from the first quantity and / or the second quantity of contrast agent. For example, the third quantity of contrast agent may be greater than the first quantity of contrast agent applied during medical imaging techniques and associated with a first representation of the examined area in frequency space, and greater than the second quantity of contrast agent applied during medical imaging techniques and associated with a second representation of the examined area in frequency space.

[0108] In some non-limiting embodiments, the predictive machine learning model may include an artificial neural network. In some non-limiting embodiments, the artificial neural network may include at least three layers of processing elements. For example, the artificial neural network may include a first layer with input neurons (e.g., nodes), an Nth layer with output neurons, and N-2 inner layers, where N is a natural number greater than 2. In some non-limiting embodiments, the input neurons may be configured to receive a first representation of the examination region of the subject in the frequency space and a second representation of the examination region of the subject in the frequency space. The output neurons may be configured to provide (e.g., as output) at least one prediction of the representation of the examination region in the frequency space of an examination region to which a specific amount of contrast agent has been applied during medical imaging techniques, based on the input. In some non-limiting embodiments, the processing elements of each layer between the input and output neurons are interconnected in a predetermined pattern with predetermined connection weights. In some non-limiting embodiments, the connection weights may be updated during training of the machine learning model (e.g., training of the artificial neural network).

[0109] In some non-limiting embodiments, the predictive machine learning model may include a convolutional neural network (CNN). In some non-limiting embodiments, the CNN is capable of processing input in the form of a matrix. The CNN may include one or more filters (e.g., one or more convolutional layers) and one or more aggregation layers (e.g., one or more pooling layers) that are repeated alternately, and the ends of the CNN may include one or more fully connected neurons (e.g., one or more densely / fully connected layers). Additionally or alternatively, the predictive machine learning model may also include a generative adversarial network (GAN).

[0110] In some non-limiting embodiments, the input to the predictive machine learning model may further include information associated with the subject of examination (e.g., the subject from which the examination area is obtained based on a radiological examination), information associated with the examination area (e.g., the examination area associated with a first representation and a second representation), and / or information associated with a radiological examination performed on the subject (e.g., the examination area of ​​the subject). For example, the input to the predictive machine learning model may include information associated with the subject's sex, age, weight, height, medical history, the nature and duration and amount of medications ingested, blood pressure, central venous pressure, respiratory rate, serum albumin, total bilirubin, blood glucose, iron content, respiratory capacity, and / or similar information. Additionally or alternatively, the input to the predictive machine learning model may include the original condition of the examination area, medical procedures associated with the examination area (such as partial resection of the examination area), organ conditions (such as whether a liver transplant has been performed, whether iron liver is present, whether fatty liver is present), and / or similar conditions. Additionally or alternatively, the input to the predictive machine learning model may include the type of radiological examination performed on the subject.

[0111] In some non-limiting embodiments, the radiation machine learning system 102 may specify (e.g., select, choose, determine, etc.) a portion of the center of frequency space included in a first representation of the examination region in frequency space (e.g., a region), a portion of the center of frequency space included in a second representation of the examination region in frequency space, or both the first and second representations of the examination region in frequency space may each include a portion of the center of frequency space to provide a simplified representation of the examination region of the object in frequency space. For example, the radiation machine learning system 102 may automatically specify a portion of the center of frequency space included in a first representation of the examination region in frequency space (e.g., a region), a portion of the center of frequency space included in a second representation of the examination region in frequency space, or both the first and second representations of the examination region in frequency space may each include a portion of the center of frequency space to provide a simplified representation of the examination region of the object in frequency space. In some non-limiting embodiments, the radiometric machine learning system 102 may specify a portion of the representation of the examination region in frequency space (e.g., a portion of a first representation of the examination region in frequency space, a portion of a second representation of the examination region in frequency space, a portion of a simplified representation of the examination region in frequency space, a portion of a reference representation of the examination region in frequency space, etc.) based on manual input received from a user (e.g., a user of user equipment 106).

[0112] In one embodiment, a simplified representation of the inspection area of ​​the inspection object in frequency space may include a first representation of the inspection area in frequency space containing a portion of the center of the frequency space, and a second representation of the inspection area in frequency space. In another embodiment, a simplified representation of the inspection area of ​​the inspection object in frequency space may include a first representation of the inspection area in frequency space and a second representation of the inspection area in frequency space containing a portion of the center of the frequency space. In yet another embodiment, a simplified representation of the inspection area of ​​the inspection object in frequency space may include a first representation of the inspection area in frequency space containing a portion of the center of the frequency space and a second representation of the inspection area in frequency space containing a portion of the center of the frequency space.

[0113] In some non-limiting embodiments, the input to the predictive machine learning model includes a simplified representation of the inspection region of the object in frequency space. In some non-limiting embodiments, the radiometric machine learning system 102 can provide the simplified representation of the inspection region of the object in frequency space as input to the predictive machine learning model. By using the simplified representation of the inspection region of the object in frequency space, the radiometric machine learning system 102 can reduce computational resource requirements and computational complexity, which are used when obtaining the output of the predictive machine learning model based on the simplified representation of the inspection region of the object in frequency space.

[0114] According to this disclosure, using the representation of the inspection region in frequency space may be more advantageous than using its representation in entity space. For example, co-registration of the representation in frequency space is not as critical as in entity space. Co-registration (e.g., image registration) can refer to a process in digital image processing that best matches (e.g., registers) two or more images of the same or similar regions. One image can be defined as a reference image, and the other image can be defined as an object image. The reference image and the object image to be matched may be different from each other because they were acquired from different locations, different time points, and / or with different sensors. To best match the object image with the corresponding reference image, a compensation transform can be computed.

[0115] During the training, validation, and prediction processes, the frequency-space representation of the examined region may be more tolerant of errors in co-registration than its entity-space representation. For example, if one representation in frequency space is superimposed with another representation in frequency space with lower accuracy, the impact of the lack of accuracy is smaller compared to the superposition of the entity-space representation of the examined region with lower accuracy. This reduced impact is due to the properties of the Fourier transform. For instance, after transforming an image using a Fourier transform, a flip or rotation of the entity-space representation (e.g., the image itself) may cause image information (e.g., information associated with visible structures) to be localized in different regions of the image. However, in frequency space, such a flip or rotation does not change the region providing contrast information because the contrast information of the Fourier-transformed image (e.g., the representation of the image in frequency space) is mapped around the center of frequency space.

[0116] The term "frequency space center" is synonymous with "frequency space origin." The frequency space center, for example, corresponds to a point with coordinates 0,0 in a two-dimensional Cartesian coordinate system. Points closer to the frequency space center represent lower frequencies than points farther from it. The region of frequency space surrounding the center contains more information about the contrast of the representation of the examined region than regions farther from it.

[0117] Another advantage of using the frequency-space representation of the inspection region is that contrast information is separated from detail information (e.g., fine structure). Therefore, it is possible to focus attention on the contrast information that will be learned by the predictive machine learning model during the training process, and on the contrast information that may be predicted by the predictive machine learning model during the prediction process. While the contrast information of the inspection region in entity space is typically distributed throughout the entire representation (e.g., each image element inherently contains information about contrast), the contrast information of the inspection region in frequency space is encoded in and around the center of the frequency space. Thus, the lower frequencies of the frequency-space representation of the inspection region are responsible for contrast information, while the higher frequencies contain detail information about fine structure. Using the frequency-space representation of the inspection region allows for the separation of contrast information to restrict training and prediction to contrast information and to reintroduce information about fine structure after the training and / or prediction processes.

[0118] like Figure 3As shown, in step 308, process 300 may include determining the output of a predictive machine learning model. For example, a radiological machine learning system 102 may provide input to a predictive machine learning model, and the predictive machine learning model may generate an output based on that input. The radiological machine learning system 102 may determine the output based on the predictive machine learning model that generates the output. In some non-limiting embodiments, the radiological machine learning system 102 may receive the output based on the predictive machine learning model that generates the output. In some non-limiting embodiments, the output of the predictive machine learning model may include a third representation of the examination region in the frequency space, which may be a prediction of the representation of the examination region in the frequency space (e.g., an artificial representation). In some non-limiting embodiments, the third representation may include a representation of the examination region in the frequency space of an examination region to which a third amount of contrast agent has been applied during medical imaging techniques.

[0119] In some non-limiting embodiments, the radiographic machine learning system 102 can determine the output of a predictive machine learning model based on k-space data associated with an MRI examination of the examination area. For example, the radiographic machine learning system 102 can provide the predictive machine learning model with first k-space data and second k-space data associated with an MRI examination of the examination area as input. The radiographic machine learning system 102 can determine the output based on the predictive machine learning model, which generates the output using the first k-space data and second k-space data associated with an MRI examination of the examination area as input.

[0120] like Figure 3 As shown, in step 310, process 300 may include converting the output of the predictive machine learning model into a predicted representation of the examination region in physical space. For example, the radiological machine learning system 102 may convert the output of the predictive machine learning model into a predicted representation of the examination region in physical space (e.g., a prediction of the representation of the examination region in physical space). In some non-limiting embodiments, the predicted representation of the examination region in physical space may include a predicted radiographic image of the examination region during medical imaging techniques (e.g., a medical imaging technique associated with a first representation of the examination region in frequency space and a second representation of the examination region in frequency space) with a third amount of contrast agent applied (e.g., a prediction of a radiographic image of the examination region). In some non-limiting embodiments, the radiological machine learning system 102 may use an inverse Fourier transform to convert the output of the predictive machine learning model into a predicted representation of the examination region in physical space.

[0121] In some non-limiting embodiments, the radiology machine learning system 102 can perform operations associated with a predicted representation of the examination region in physical space. For example, the radiology machine learning system 102 can provide a predicted representation of the examination region in physical space as its output. In such embodiments, the radiology machine learning system 102 can provide (e.g., transmit) the predicted representation of the examination region in physical space to user device 106 and / or medical imaging system 104 (e.g., so that user device 106 and / or medical imaging system 104 can output the predicted representation of the examination region in physical space). In some non-limiting embodiments, the radiology machine learning system 102 can store the predicted representation of the examination region in physical space in a data structure associated with the radiology machine learning system 102. In some non-limiting embodiments, the radiology machine learning system 102 can display the predicted representation of the examination region in physical space on a display device.

[0122] In some non-limiting embodiments, the radiometric machine learning system 102 may supplement the output of the predictive machine learning model to provide a supplementary output. For example, the radiometric machine learning system 102 may supplement the output of the predictive machine learning model with: a portion of the first representation of the inspected region in frequency space that does not contain a frequency space center and is not designated to provide a simplified representation of the inspected region in frequency space; a portion of the second representation of the inspected region in frequency space that does not contain a frequency space center and is not designated to provide a simplified representation of the inspected region in frequency space; or a portion of the first and second representations that respectively do not contain a frequency space center and are not designated to provide a simplified representation of the inspected region in frequency space. In some non-limiting embodiments, the radiometric machine learning system 102 may convert the supplementary output of the predictive machine learning model into a representation of the inspected region of the inspected object in entity space.

[0123] In some non-limiting embodiments, the radiological machine learning system 102 can convert the resulting signal intensity of various location coordinates of the predicted representation of the examined area in physical space into grayscale and / or color values ​​to provide a digital image in a common image format (e.g., the Medical Digital Imaging and Communications (DICOM) format).

[0124] Now for reference Figure 4 , Figure 4This is a flowchart of a non-limiting embodiment of a process 400 for training a predictive machine learning model configured to provide predictions of an examination area where a specific amount of contrast agent has been applied during a medical imaging technique. In some non-limiting embodiments, the predictive machine learning model may be the same as or similar to the predictive machine learning model described with respect to process 300. In some non-limiting embodiments or aspects, one or more functions described with respect to process 400 may (e.g., wholly, partially, etc.) be performed by the radiological machine learning system 102. In some non-limiting embodiments or aspects, one or more steps of process 300 may (e.g., wholly, partially, etc.) be performed by another device or set of devices independent of and / or including the radiological machine learning system 102, such as the medical imaging system 104 and / or user equipment 106.

[0125] like Figure 4 As shown, in step 402, process 400 may include receiving a training data set, which includes a set of reference representations in frequency space of the examination region of each of a plurality of examination objects. In some non-limiting embodiments, the radiology machine learning system 102 may receive the training data set from the medical imaging system 104 and / or the user equipment 106, which includes a set (e.g., multiple reference representations) of reference representations in frequency space of the examination region of each of a plurality of examination objects. In some non-limiting embodiments, the radiology machine learning system 102 may retrieve the training data set from a data structure associated with the radiology machine learning system 102. In some non-limiting embodiments, the examination region may be the same for all examination objects among the plurality of examination objects. In some non-limiting embodiments, the reference representation of the examination region (e.g., a simplified reference representation) may refer to the representation in frequency space of the examination region used during the training process of the predictive machine learning model.

[0126] In some non-limiting embodiments, each reference representation of the examination area of ​​each subject in the frequency space may further include information associated with the subject (e.g., the subject from which the examination area is obtained based on a radiological examination), information associated with the examination area (e.g., the examination area associated with the first and second representations), and / or information associated with the radiological examination performed on the subject (e.g., the examination area of ​​the subject). For example, information associated with the subject may include information associated with the subject's sex, age, weight, height, medical history, nature and / or duration and / or amount of medications ingested, blood pressure, central venous pressure, respiratory rate, serum albumin, total bilirubin, blood glucose, iron content, respiratory capacity, and / or similar information. Additionally or alternatively, information associated with the examination area may include the original condition of the examination area, medical procedures associated with the examination area (such as partial resection of the examination area), organ condition (such as whether a liver transplant has been performed, whether iron liver is present, whether fatty liver is present), and / or similar conditions. Additionally or alternatively, information associated with the radiological examination performed on the subject may include the type of radiological examination performed on the subject.

[0127] In some non-limiting embodiments, the training dataset may include a set (e.g., multiple reference representations) of the examination region of each of a plurality of examination subjects in frequency space. Each set of reference representations of the examination region of the examination subject in frequency space may include a first reference representation of the examination region of the examination subject in frequency space, a second reference representation of the examination region of the examination subject in frequency space, and a third reference representation of the examination region of the examination subject in frequency space. In some non-limiting embodiments, the first reference representation may include a reference representation of the examination region in frequency space for which no contrast agent was applied during the medical imaging technique or for which a first amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the second reference representation may include a reference representation of the examination region in frequency space for which a second amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the third reference representation may include a reference representation of the examination region in frequency space for which a third amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the third amount of contrast agent may be a standard amount of contrast agent applied to the examination subject during the medical imaging technique based on parameters of the medical imaging technique and / or the examination subject. In some non-limiting embodiments, the third reference representation may be a baseline truth representation.

[0128] In some non-limiting embodiments, the first quantity of contrast agent applied during the medical imaging technique is greater than zero. In some non-limiting embodiments, the first quantity of contrast agent applied during the medical imaging technique is zero. In some non-limiting embodiments, the second quantity of contrast agent differs from the first quantity of contrast agent applied during the medical imaging technique and associated with the first representation. In some non-limiting embodiments, the second quantity of contrast agent is greater than the first quantity of contrast agent applied during the medical imaging technique. In some non-limiting embodiments, the third quantity of contrast agent differs from the first quantity of contrast agent and / or the second quantity of contrast agent. For example, the third quantity of contrast agent may be greater than the first quantity of contrast agent applied during the medical imaging technique and associated with the first representation of the examined area in frequency space, and greater than the second quantity of contrast agent applied during the medical imaging technique and associated with the second representation of the examined area in frequency space.

[0129] In some non-limiting embodiments, the training dataset may include a set of simplified reference representations of the inspection region of each of a plurality of inspection objects in frequency space. Each set of simplified reference representations of the inspection region of an inspection object in frequency space may include a first simplified reference representation of the inspection region of the inspection object in frequency space, a second simplified reference representation of the inspection region of the inspection object in frequency space, and / or a third simplified reference representation of the inspection region of the inspection object in frequency space. In some non-limiting embodiments, the first simplified reference representation may include a representation in frequency space of a portion of the first reference representation of the inspection region in frequency space, including the center of the frequency space. In some non-limiting embodiments, the second simplified reference representation may include a representation in frequency space of a second reference representation of the inspection region in frequency space, including the center of the frequency space. In some non-limiting embodiments, the third simplified reference representation may include a representation in frequency space of a third reference representation of the inspection region in frequency space, including the center of the frequency space.

[0130] like Figure 4 As shown, in step 404, process 400 may include specifying a portion of a reference representation. For example, the radiation machine learning system 102 may specify a portion (e.g., a region) of a reference representation (e.g., the same portion of each reference representation) of the examination region of an object, which is used to provide a simplified reference representation of the examination region of the object in the frequency space. In this way, the radiation machine learning system 102 can provide a training dataset that includes simplified reference representations of the examination regions of the objects among a plurality of objects in the frequency space.

[0131] In some non-limiting embodiments, the radiometric machine learning system 102 may specify (e.g., automatically specify) a portion (e.g., a designated portion) of the examination region of an examination object in frequency space for each reference representation (e.g., each reference representation in a set of reference representations) among a plurality of examination objects to provide a simplified reference representation of the examination region of each examination object in frequency space. For example, the radiometric machine learning system 102 may specify a portion of the reference representation of the examination region in frequency space from the set of reference representations of the examination region in frequency space such that the portion contains the center of frequency space. In some non-limiting embodiments, the center of the portion may correspond to the center of frequency space. In some non-limiting embodiments, the portion may have a certain shape. For example, the designated portion may be circular, angular, concave, and / or convex. In some non-limiting embodiments, in the case of 3D frequency space in Cartesian coordinates, the portion may be cubic (e.g., cuboid). In some non-limiting embodiments, in the case of 2D frequency space in Cartesian coordinates, the portion may be rectangular (e.g., square). In some non-limiting embodiments, the size of the designated portion may be the same as that in frequency space. For example, in the case of a 2D representation in a 2D frequency space, the portion may include a plane. In another embodiment, in the case of a 3D representation in a 3D frequency space, the portion may include a volume.

[0132] like Figure 4 As shown, in step 406, process 400 may include reducing the reference representation. For example, the radiometric machine learning system 102 may reduce the reference representation of the inspected region of the object in the frequency space based on a specified portion (e.g., a designated portion) to provide a simplified reference representation of the inspected region of the object in the frequency space. In some non-limiting embodiments, the radiometric machine learning system 102 may reduce each reference representation by removing (e.g., discarding, cutting, etc.) any portion of the reference representation that is not in that portion of the reference representation. For example, the radiometric machine learning system 102 may reduce each reference representation by covering the reference representation with a mask (e.g., a mask corresponding to the designated portion) and removing any portions of the reference representation that are not covered by the mask. In some non-limiting embodiments, when covering the reference representation with a mask, the radiometric machine learning system 102 may set one or more color values ​​of image elements (e.g., pixels, voxels) that are not covered by the mask to zero (e.g., black).

[0133] like Figure 4As shown, in step 408, process 400 may include training a predictive machine learning model using multiple simplified reference representations. For example, the radiation machine learning system 102 may train the predictive machine learning model based on multiple simplified reference representations of the inspection region (e.g., the inspection region of an object being inspected) in the frequency space. In some non-limiting embodiments, the radiation machine learning system 102 may use a self-learning algorithm to train the predictive machine learning model during supervised machine learning. In some non-limiting embodiments, the radiation machine learning system 102 may use a self-learning algorithm during machine learning to generate a statistical model based on the training dataset.

[0134] In some non-limiting embodiments, the radiological machine learning system 102 may minimize the amount of error provided by an error function during training. In some non-limiting embodiments, the error function may quantify the deviation between a predicted representation (e.g., a predicted representation) of an examination region of a subject to which a specific amount of contrast agent has been applied during medical imaging techniques and a baseline true reference representation of the examination region in frequency space (e.g., a baseline true reference representation of the examination region of a subject to which a specific amount of contrast agent has been applied during medical imaging techniques). In some non-limiting embodiments, the error function quantifies the deviation between a predicted simplified representation (e.g., a predicted simplified representation) of an examination region of a subject to which a specific amount of contrast agent has been applied during medical imaging techniques and a baseline true simplified reference representation of the examination region in frequency space.

[0135] In some non-limiting embodiments, the predictive machine learning model may be configured to provide a prediction as output of the representation of the examination region in frequency space of the region to which a specific amount of contrast agent was applied during a medical imaging technique (e.g., the same medical imaging technique associated with a first representation of the examination region in frequency space and a second representation of the examination region in frequency space). In some non-limiting embodiments, this specific amount of contrast agent differs from a first amount of contrast agent applied during the medical imaging technique and associated with the first representation of the examination region in frequency space, and differs from a second amount of contrast agent applied during the medical imaging technique and associated with the second representation of the examination region in frequency space. For example, this specific amount of contrast agent is greater than both the first amount of contrast agent applied during the medical imaging technique and associated with the first representation of the examination region in frequency space, and the second amount of contrast agent applied during the medical imaging technique and associated with the second representation of the examination region in frequency space.

[0136] In some non-limiting embodiments, the radiological machine learning system 102 may train a predictive machine learning model based on k-space data. For example, each simplified reference representation of the examination region of an object in frequency space may include k-space data associated with an MRI examination (e.g., an MRI examination performed by the medical imaging system 104). The radiological machine learning system 102 may train a predictive machine learning model based on k-space data contained in each simplified reference representation of the examination region of an object in frequency space among multiple examination objects.

[0137] In some non-limiting embodiments, the radial machine learning system 102 can train a predictive machine learning model using a backpropagation method. In this way, the radial machine learning system 102 can train the predictive machine learning model to have maximum reliability in mapping inputs to corresponding outputs. The mapping quality can be described by an error function (e.g., a loss function). During the backpropagation method, the objective is to minimize the error function. In the case of the backpropagation method, the predictive machine learning model can be updated (e.g., learned, taught, etc.) by changing the connection weights of nodes. During training, the connection weights between the processing elements of the predictive machine learning model can contain information about the relationship between a first reference representation and a second reference representation of the examined region in the frequency space, as well as information about a third reference representation. This information can be used to predict the representation of the examined region in the frequency space based on the first and second representations. In some non-limiting embodiments, the radial machine learning system 102 can use a cross-validation method to divide the data into a training data set and a validation data set. The training data set can be used for backpropagation training of the connection weights of the nodes. The validation data set can be used to verify the accuracy of predictions based on representations provided by the trained predictive machine learning model using data not used during training.

[0138] In some non-limiting embodiments, the radiological machine learning system 102 may generate a predictive machine learning model based on a training dataset. For example, the radiological machine learning system 102 may generate a predictive machine learning model to predict the representation (e.g., representation in frequency space) of an examination area where a specific amount of contrast agent has been applied during a medical imaging technique (e.g., a medical imaging technique performed during a radiological examination).

[0139] In some non-limiting embodiments, the predictive machine learning model may include a machine learning model designed to receive data as input associated with a frequency-space representation of an examination region to which a certain amount (e.g., some amount, no amount, etc.) of contrast agent was applied during medical imaging, and to provide as output a prediction of the frequency-space representation of an examination region to which a specific amount of contrast agent was applied during medical imaging. For example, the predictive machine learning model may be designed to receive data associated with a first representation of an examination region in frequency space and data associated with a second representation of an examination region in frequency space, the first representation including representations of examination regions to which no amount of contrast agent was applied or to which a first amount of contrast agent was applied during medical imaging, the second representation including representations of examination regions to which a second amount of contrast agent was applied during medical imaging, and to provide an output including a prediction of the frequency-space representation of an examination region to which a third amount of contrast agent was applied during medical imaging. In some non-limiting embodiments, the radiological machine learning system 102 may store the predictive machine learning model (e.g., for later use).

[0140] In some non-limiting embodiments, as described herein, the radiographic machine learning system 102 can process data associated with the representation of the inspected region of an object in frequency space to obtain training data (e.g., a training dataset) for a predictive machine learning model. For example, the radiographic machine learning system 102 can process data to change it into a format that can be analyzed (e.g., by the radiographic machine learning system 102) to generate a predictive machine learning model. The modified data (e.g., the data obtained from the modification) can be referred to as training data. In some non-limiting embodiments, the radiographic machine learning system 102 can process data associated with the representation of the inspected region of an object in frequency space during a time interval to obtain training data based on the received data. Additionally or alternatively, based on an instruction received by the radiographic machine learning system 102 from a user of the radiographic machine learning system 102 (e.g., a user associated with user device 106) that the radiographic machine learning system 102 will process data to obtain training data, such as when the radiographic machine learning system 102 receives an instruction to generate a predictive machine learning model for a time interval corresponding to the data.

[0141] In some non-limiting embodiments, the radiological machine learning system 102 can process data associated with the representation of the examination area of ​​the subject in frequency space by determining predictor variables based on the data. Predictor variables may include measures associated with the examination area or the subject, which may be derived based on data associated with the representation of the examination area of ​​the subject in frequency space. Predictor variables can be analyzed to generate a predictive machine learning model. For example, predictor variables may include variables associated with the subject, such as variables associated with the subject's sex, age, weight, height, medical history, nature and / or duration and / or amount of medications ingested, blood pressure, central venous pressure, respiratory rate, serum albumin, total bilirubin, blood glucose, iron content, respiratory capacity, and / or similar information. Additionally or alternatively, predictor variables may also include variables associated with the examination area, such as variables associated with the pre-existing condition of the examination area, medical procedures associated with the examination area (such as partial resection of the examination area), organ conditions (such as whether a liver transplant has been performed, whether iron liver is present, whether fatty liver is present), and / or similar conditions. Additionally or alternatively, predictor variables may also include variables associated with the radiological examination performed on the subject of examination, such as variables associated with the type of radiological examination performed on the subject of examination.

[0142] In some non-limiting embodiments, the radiology machine learning system 102 can analyze training data to generate a predictive machine learning model. For example, the radiology machine learning system 102 can use machine learning techniques to analyze training data to generate a predictive machine learning model. In some non-limiting embodiments, generating a predictive machine learning model (e.g., based on training data obtained from historical data associated with the representation of the examination region of the examined object in frequency space and / or entity space) can be referred to as training a predictive machine learning model. For example, machine learning techniques can include supervised and / or unsupervised techniques such as decision trees, random forests, logistic regression, linear regression, gradient boosting, support vector machines, additional trees (e.g., extensions of random forests), Bayesian statistics, learning automata, hidden Markov modeling, linear classifiers, quadratic classifiers, association rule learning, and / or similar techniques. In some non-limiting embodiments, the predictive machine learning model can include a feature-specific model, such as a model specific to a particular examination region, a particular examination object, a model specific to a specific amount of contrast agent administered during medical imaging techniques, etc. Additionally or alternatively, the predictive machine learning model can be specific to a particular type of examination object (e.g., mammalian type, such as human) that contains a particular type of examination region. In some non-limiting embodiments, the radiometric machine learning system 102 can generate one or more predictive machine learning models for one or more entities, a specific group of entities, and / or one or more users of one or more entities.

[0143] Additionally or alternatively, when analyzing training data, the radiology machine learning system 102 may identify one or more variables (e.g., one or more independent variables) as predictor variables (e.g., features) that can be used for prediction when analyzing the training data. In some non-limiting embodiments, the values ​​of the predictor variables may be inputs to a predictive machine learning model. For example, the radiology machine learning system 102 may identify a subset of variables (e.g., an appropriate subset) as predictor variables that can be used to accurately predict the representation of the examined region in frequency space. In some non-limiting embodiments, the predictor variables may include one or more of the examined region variables and / or examined object variables discussed above, which have a significant impact (e.g., a threshold-satisfying effect) on the prediction of the representation of the examined region in frequency space for which a specific amount of contrast agent has been applied during medical imaging techniques, as determined by the radiology machine learning system 102.

[0144] In some non-limiting embodiments, the radial machine learning system 102 can validate the predictive machine learning model. For example, the radial machine learning system 102 can validate the predictive machine learning model after it has been generated. In some non-limiting embodiments, the radial machine learning system 102 can validate the predictive machine learning model based on a portion of the training data to be used for validation. For example, the radial machine learning system 102 can divide the training data into a first portion and a second portion, wherein the first portion can be used to generate the predictive machine learning model, as described above. In this embodiment, the second portion of the training data (e.g., validation data) can be used to validate the predictive machine learning model.

[0145] In some non-limiting embodiments, the radiological machine learning system 102 may validate the predictive machine learning model by providing validation data (e.g., data on the representation of the examination area of ​​the subject in frequency space) associated with the representation of the examination area in frequency space as input to the predictive machine learning model, and determining, based on the output of the predictive machine learning model, whether the predictive machine learning model correctly or incorrectly predicts the representation of the examination area in frequency space for which a specific amount of contrast agent was applied during medical imaging techniques. In some non-limiting embodiments, the radiological machine learning system 102 may validate the predictive machine learning model based on a validation threshold. For example, the radiological machine learning system 102 can be configured to validate the predictive machine learning model when the predictive machine learning model correctly predicts (as identified by validation data) the representation in frequency space of the examination area where a specific amount of contrast agent was applied during medical imaging techniques (e.g., when the predictive machine learning model correctly predicts 50% of the representation in frequency space of the examination area where a specific amount of contrast agent was applied during medical imaging techniques, 70% of the representation in frequency space of the examination area where a specific amount of contrast agent was applied during medical imaging techniques, or other threshold quality of the representation in frequency space of the examination area where a specific amount of contrast agent was applied during medical imaging techniques, and / or similar cases).

[0146] In some non-limiting embodiments, if the radiographic machine learning system 102 fails to validate the predictive machine learning model (e.g., when the percentage of the examination area in frequency space correctly predicted during a medical imaging technique when a specific amount of contrast agent was applied does not meet a validation threshold), the radiographic machine learning system 102 may generate one or more additional predictive machine learning models.

[0147] In some non-limiting embodiments, once the predictive machine learning model has been validated, the radiology machine learning system 102 can further train the predictive machine learning model and / or generate new predictive machine learning models based on newly received training data. The new training data may include additional data associated with one or more representations of the examined region of the subject in the frequency space. In some non-limiting embodiments, the new training data may include data associated with the representation of the examined region of the subject in the frequency space. The radiology machine learning system 102 can use the predictive machine learning model to predict the representation of the examined region in the frequency space for which a specific amount of contrast agent has been applied during medical imaging techniques, and compare the output of the predictive machine learning model with new training data containing data associated with the representation of the examined region of the subject in the frequency space. In such embodiments, the radiology machine learning system 102 can update one or more predictive machine learning models based on the new training data.

[0148] In some non-limiting embodiments, the radiographic machine learning system 102 may store predictive machine learning models. For example, the radiographic machine learning system 102 may store predictive machine learning models in a data structure (e.g., a database, linked list, tree, and / or similar structure). The data structure may be located inside or outside the radiographic machine learning system 102 (e.g., away from the radiographic machine learning system 102).

[0149] Now for reference Figures 5A to 5C , Figures 5A to 5C This is a schematic diagram of an implementation 500 of a process (e.g., process 300) for providing a prediction of an examination area, the prediction being generated using medical imaging techniques involving contrast agents. See below for... Figures 5A to 5C As described, implementation 500 may include a radiographic machine learning system 102 that performs the steps of the process. In some non-limiting embodiments, one or more steps of the process may (e.g., wholly, partially, etc.) be performed by another device or set of devices, independent of and / or including the radiographic machine learning system 102, such as a medical imaging system 104 and / or a user device 106.

[0150] like Figure 5A As shown, the radiation machine learning system 102 can generate a representation of the examination region of the object in frequency space. For example... Figure 5A As further shown, at three different time points t1, t2, and t3, the radiology machine learning system 102 can generate three representations of the examination area of ​​the subject. For example, the radiology machine learning system 102 can receive the results of a radiological examination involving medical imaging techniques performed on the examination area of ​​the subject by the medical imaging system 104. The examination area may be the lungs of a human patient. Based on the medical imaging techniques performed on the examination area by the medical imaging system 104, the results may include three representations of the examination area of ​​the subject in physical space.

[0151] The result may include three radiographic images, including a first radiographic image O1, a second radiographic image O2, and a third radiographic image O3. The first radiographic image O1 may include an image of the examination area formed when no contrast agent was applied during the medical imaging technique or when a first amount of contrast agent was applied during the medical imaging technique. The second radiographic image O2 may include an image of the examination area formed when a second amount of contrast agent was applied during the medical imaging technique. The third radiographic image O3 may include an image of the examination area formed when a third amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the second amount of contrast agent differs from the first amount of contrast agent. In some non-limiting embodiments, the third amount of contrast agent differs from the first amount of contrast agent and the second amount of contrast agent. In one embodiment, the first amount of contrast agent is greater than or equal to zero, the second amount of contrast agent is greater than the first amount of contrast agent, and the third amount of contrast agent is greater than the second amount of contrast agent.

[0152] The first representation F1 may include a representation of an examination area in which no contrast agent was applied during medical imaging or in which a first amount of contrast agent was applied during medical imaging. The second representation F2 may include a representation of an examination area in which a second amount of contrast agent was applied during medical imaging. The third representation F3 may include a representation of an examination area in which a third amount of contrast agent was applied during medical imaging.

[0153] In some non-limiting embodiments, based on the results of a radiological examination performed by the medical imaging system 104, the radiology machine learning system 102 can generate three representations F1, F2, and F3 of the examined object in frequency space. For example, based on the representations O1, O2, and O3 of the examined region in solid space, the radiology machine learning system 102 can use a Fourier transform (FT) to generate three representations F1, F2, and F3 of the examination in frequency space. The radiology machine learning system 102 can use a Fourier transform to convert the representations O1, O2, and O3 of the examined region in solid space into representations F1, F2, and F3 of the examined region in frequency space. The radiology machine learning system 102 can use an inverse Fourier transform (iFT) to convert the representations F1, F2, and F3 of the examined region in frequency space into representations O1, O2, and O3 of the examined region in solid space. In some non-limiting embodiments, the radiology machine learning system 102 can use transforms other than the Fourier transform to convert the representations in solid space into representations in frequency space. Three key characteristics of this transformation can include the existence of an explicit inverse transformation (e.g., an explicit relationship between the representation in entity space and the representation in frequency space), the localization of contrast information, and robustness to defective image registration.

[0154] like Figure 5BAs shown, the frequency-space representations F1, F2, and F3 of the examined regions can be used to train a predictive machine learning model. These frequency-space representations can be included in the training dataset of the examined objects. The radiographic machine learning system 102 can use multiple training datasets of multiple examined objects to train a predictive machine learning model.

[0155] In some non-limiting embodiments, based on a first representation F1 and a second representation F2, a predictive machine learning model is trained as output to provide a prediction of the representation in frequency space of the examined region to which a specific amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the specific amount of contrast agent corresponds to a third amount of contrast agent applied during the medical imaging technique.

[0156] like Figure 5B As further shown, the radiographic machine learning system 102 can provide input to a predictive machine learning model, and the radiographic machine learning system 102 can receive the output of the predictive machine learning model based on the input. In some non-limiting embodiments, a first representation F1 and a second representation F2 can be provided as input to the predictive machine learning model. The output of the predictive machine learning model can include a prediction of the representation in frequency space of the examination region to which a third amount of contrast agent has been applied during medical imaging techniques. As part of the training process, the radiology machine learning system 102 can predict the representation in frequency space of the examination area where a third amount of contrast agent has been applied during medical imaging techniques. The representation F3 in frequency space is compared with that of the examination area where a third amount of contrast agent was applied during medical imaging techniques. The predicted representation... The deviation from the third representation F3 can be used in backpropagation to train a predictive machine learning model, thereby reducing the deviation to a defined minimum. This is possible if the predictive machine learning model has been trained on multiple training datasets for multiple examination subjects, and if the predicted representation of the examination area in frequency space is obtained after a third amount of contrast agent has been applied during medical imaging. Once the required accuracy has been achieved, the predictive machine learning model can be considered to have been trained and is ready for prediction.

[0157] like Figure 5C As shown, the radiology machine learning system 102 can use a predictive machine learning model to provide a prediction of the representation in frequency space of an examination region of a subject who has been examined and for which a specific amount of contrast agent has been applied during medical imaging techniques. In some non-limiting embodiments, the radiology machine learning system 102 can receive a first representation in frequency space of the examination region of the subject. The first representation in frequency space This may include a representation of the examination area where no contrast agent was applied during the medical imaging technique or where a first amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the radiographic machine learning system 102 may receive a second representation of the examination area of ​​the subject in the frequency space. In some non-limiting embodiments, the second representation in the frequency space It may include a representation of the examination area where a second amount of contrast agent was applied during medical imaging techniques, and the second amount of contrast agent may be greater than the first amount of contrast agent.

[0158] like Figure 5C As further shown, the radiometric machine learning system 102 can provide input to a predictive machine learning model. The input to the predictive machine learning model may include a first representation in the frequency space. and the second representation in frequency space Predictive machine learning models can include trained machine learning models configured to provide predictions of the representation in frequency space of the examined region where a third amount of contrast agent has been applied during medical imaging techniques. As an output, in some non-limiting embodiments, the third quantity of contrast agent is greater than the first quantity and the second quantity of contrast agent.

[0159] like Figure 5C As further shown, the radiation machine learning system 102 can receive the output of a predictive machine learning model based on this input. The iFT is used to convert the output of the predictive machine learning model into a predicted representation of the inspected region of the inspected object in the entity space. In some non-limiting embodiments, the radiographic machine learning system 102 can provide a predictive representation of the inspection area of ​​the object in the physical space. .

[0160] Now for reference Figures 6A to 6B , Figures 6A to 6B This is a schematic diagram of an implementation 600 of a process (e.g., process 300) for providing a prediction of an examination area, the prediction being generated using medical imaging techniques involving contrast agents. See below for... Figures 6A to 6B As described, implementation 600 may include a radiographic machine learning system 102 that performs the steps of the process. In some non-limiting embodiments, one or more steps of the process may (e.g., wholly, partially, etc.) be performed by another device or set of devices, independent of and / or including the radiographic machine learning system 102, such as a medical imaging system 104 and / or a user device 106.

[0161] like Figure 6AAs shown, in the same or similar manner as in embodiment 500, the frequency space representations F1, F2, and F3 of the examined regions can be used to train a predictive machine learning model. The frequency space representations F1, F2, and F3 of the examined regions can be included in the training dataset of the examined objects. The radiometric machine learning system 102 can use multiple training datasets of multiple examined objects to train a predictive machine learning model.

[0162] Similar to embodiment 500, the first representation F1 may include a representation of an examination area in which no contrast agent was applied during medical imaging or in which a first amount of contrast agent was applied during medical imaging; the second representation F2 may include a representation of an examination area in which a second amount of contrast agent was applied during medical imaging; and the third representation F3 may include a representation of an examination area in which a third amount of contrast agent was applied during medical imaging.

[0163] In some non-limiting embodiments, in the same or similar manner as embodiment 500, based on the results of a radiological examination performed by the medical imaging system 104, the radiological machine learning system 102 can generate three representations F1, F2, and F3 of the examined area of ​​the subject in the frequency space. In some non-limiting embodiments, based on a portion of the first representation and a portion of the second representation, a predictive machine learning model is trained as output to provide a prediction of a portion of the representation in the frequency space of the examined area for which a specific amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the specific amount of contrast agent corresponds to a third amount of contrast agent applied during the medical imaging technique.

[0164] like Figure 6A As further shown, the radiation machine learning system 102 can specify a portion A of the first representation F1 that includes the center of the frequency space to provide a first simplified representation of the examination region of the object in the frequency space. Furthermore, the radiographic machine learning system 102 can specify a portion A of the second representation F2 that includes the center of the frequency space to provide a second simplified representation of the examination region of the object in the frequency space. Furthermore, the radiographic machine learning system 102 can specify a portion A of the frequency space center of the third representation F3 to provide a third simplified representation of the examination region of the object in the frequency space. .

[0165] like Figure 6A As further shown, the radiographic machine learning system 102 can represent the first simplified representation of the examination area of ​​the object in the frequency space. The second simplified representation of the inspection area of ​​the inspection object in frequency space. The data is provided as input to a predictive machine learning model, and the radiological machine learning system 102 can receive the output of the predictive machine learning model based on this input. In some non-limiting embodiments, the output may include a prediction of a portion of the representation in frequency space of the examined region to which a third amount of contrast agent was applied during medical imaging techniques. .

[0166] As part of the training process, the radiological machine learning system 102 can predict a portion of the representation in frequency space of the examination area where a third amount of contrast agent was applied during medical imaging techniques. The third simplified representation of the inspection area of ​​the inspection object in frequency space Comparison. Prediction of a portion of the representation in the frequency space. With the third simplified representation in frequency space The deviation can be used in the backpropagation method to train a predictive machine learning model, thereby reducing the deviation to a defined minimum. This is possible if the predictive machine learning model has been trained on multiple training datasets for multiple examination subjects, and if a third amount of contrast agent is applied during medical imaging, the prediction is based on a portion of the frequency space representation. Once the required accuracy has been achieved, the predictive machine learning model can be considered to have been trained and is ready for prediction.

[0167] like Figure 6B As shown, the radiology machine learning system 102 can use a predictive machine learning model to provide a prediction of the representation in frequency space of an examination region of a subject who has been examined and for which a specific amount of contrast agent has been applied during medical imaging techniques. In some non-limiting embodiments, the radiology machine learning system 102 can receive a first representation in frequency space of the examination region of the subject. The first representation in frequency space This may include a representation of the examination area where no contrast agent was applied during the medical imaging technique or where a first amount of contrast agent was applied during the medical imaging technique. In some non-limiting embodiments, the radiographic machine learning system 102 may receive a second representation of the examination area of ​​the subject in the frequency space. In some non-limiting embodiments, the second representation in the frequency space It may include a representation of the examination area where a second amount of contrast agent was applied during medical imaging techniques, and the second amount of contrast agent may be greater than the first amount of contrast agent.

[0168] like Figure 6B As further shown, the radiographic machine learning system 102 can specify a first representation. A portion A, including the center of the frequency space, is used to provide a first simplified representation of the inspection area of ​​the object in the frequency space. Furthermore, the radiometric machine learning system 102 can specify a second representation. A portion A, including the center of the frequency space, is used to provide a second simplified representation of the inspection area of ​​the object in the frequency space. .

[0169] like Figure 6B As further shown, the radiometric machine learning system 102 can provide input to a predictive machine learning model. The input to the predictive machine learning model may include a first simplified representation in the frequency space. and the second simplified representation in frequency space The predictive machine learning model may include a trained machine learning model configured to provide a simplified representation in frequency space of the examined region where a third amount of contrast agent has been applied during medical imaging techniques. As an output, in some non-limiting embodiments, the third quantity of contrast agent is greater than the first quantity and the second quantity of contrast agent.

[0170] like Figure 6B As further shown, the radiation machine learning system 102 can receive the output of a predictive machine learning model based on the input. Furthermore, the radiometric machine learning system 102 can use a first simplified representation in frequency space that does not contain a frequency space center and is not specified to provide an examination region of the examination object. Part of To supplement the output of predictive machine learning models To provide supplementary outputs to predictive machine learning models. In some non-limiting embodiments, the radiographic machine learning system 102 may use iFT to convert the supplementary output of the predictive machine learning model into a predictive representation of the inspection region of the inspected object in the entity space. In some non-limiting embodiments, the radiographic machine learning system 102 can provide a predictive representation of the inspection area of ​​the object in the physical space. .

[0171] Although the above-described systems, methods, and computer program products have been described in detail for illustrative purposes based on embodiments and aspects currently considered to be the most practical and preferred, it should be understood that such detail is for that purpose only, and that this disclosure is not limited to the described embodiments or aspects; rather, it is intended to cover modifications and equivalent arrangements within the nature and scope of the appended claims. For example, it should be understood that, to the extent possible, this disclosure considers at least one feature of any embodiment or aspect to be possible in combination with at least one feature of any other embodiment or aspect.

Claims

1. A system for providing a prediction of an examination region, the prediction being generated using a medical imaging technique involving contrast agents, the system comprising at least one processor programmed or configured to perform the following operations: A first representation of the examination area of ​​the object being examined in the frequency space, wherein the first representation includes a representation of the examination area in which no amount of contrast agent is applied during the medical imaging technique or in which a first amount of contrast agent is applied during the medical imaging technique; A second representation of the examination area of ​​the object being examined in the frequency space, wherein the second representation includes a representation of the examination area during medical imaging techniques when a second amount of contrast agent is applied, wherein the second amount of contrast agent is different from a first amount of contrast agent; The system provides input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent is applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent. Receive the output of the predictive machine learning model based on the input; The output of the predictive machine learning model is converted into a predicted representation of the inspection region of the inspected object in the entity space. as well as Provides a predicted representation of the inspection area of ​​the inspection object in the physical space.

2. The system of claim 1, wherein the at least one processor is further programmed or configured to perform the following operations: Specify that the first representation includes a portion of the frequency space center, the second representation includes a portion of the frequency space center, or the first representation and the second representation each include a portion of the frequency space center, to provide a simplified representation of the inspection area of ​​the inspection object in the frequency space. The input to the predictive machine learning model includes a simplified representation of the inspection region of the object in the frequency space; and wherein When input to a predictive machine learning model is provided, the at least one processor is programmed or configured to perform the following operations: The simplified representation of the inspection region of the inspection object in the frequency space is provided as input to the predictive machine learning model.

3. The system of claim 2, wherein the at least one processor is further programmed or configured to perform the following operations: Supplement the output of the predictive machine learning model with the following: The first representation does not include a frequency space center and is not specified as a portion of the simplified representation of the inspection area of ​​the object to be inspected in frequency space. The second representation does not include a frequency space center and is not specified as a portion of the simplified frequency space representation of the inspection area of ​​the object being inspected, or The first and second representations do not include a frequency space center and are not specified as a portion of the simplified frequency space representation of the inspection area of ​​the object being inspected. To provide supplementary outputs to the predictive machine learning model; wherein When the output of the predictive machine learning model is converted into a predicted representation of the inspected region of the inspected object in entity space, the at least one processor is programmed or configured to perform the following operations: The supplementary output of the predictive machine learning model is converted into a predicted representation of the inspected region of the inspected object in the entity space.

4. The system according to any one of claims 1-3, wherein the first quantity of contrast agent applied during the medical imaging technique is greater than zero, and wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

5. The system according to any one of claims 1-3, wherein the first quantity of contrast agent applied during the medical imaging technique is zero, and wherein the second quantity of contrast agent applied during the medical imaging technique is greater than the first quantity of contrast agent.

6. The system according to any one of claims 1-3, wherein the at least one processor is further programmed or configured to perform the following operations: The first result of the radiological examination generates a first representation of the examination area of ​​the radiological examination subject in frequency space; and The second result of the radiological examination generates a second representation of the examination area of ​​the radiological examination subject in the frequency space.

7. The system of claim 6, wherein the radiological examination is a magnetic resonance imaging examination, a computed tomography examination, or an ultrasound examination.

8. The system of claim 6, wherein the radiological examination is a magnetic resonance imaging examination, wherein, When the first representation of the inspection area of ​​the inspection object in the frequency space is received in the frequency space, the at least one processor is programmed or configured to perform the following operations: Receive first k-space data associated with magnetic resonance imaging examination of the examination area of ​​the object being examined; and When receiving a second representation of the inspection region of the inspection object in the frequency space, the at least one processor is programmed or configured to perform the following operations: Receive second k-space data associated with magnetic resonance imaging examination of the examination area of ​​the object being examined.

9. The system according to any one of claims 1-3, wherein, When the first representation of the inspection area of ​​the inspection object in the frequency space is received in the frequency space, the at least one processor is programmed or configured to perform the following operations: A first representation of the examination area of ​​the receiving examination object in the physical space, wherein the first representation in the physical space includes a representation of the examination area in which no amount of contrast agent is applied during the medical imaging technique or in which a first amount of contrast agent is applied during the medical imaging technique; The first representation in the entity space is converted into the first representation of the inspection region of the inspection object in the frequency space; When receiving a second representation of the inspection region of the inspection object in the frequency space, the at least one processor is programmed or configured to perform the following operations: A second representation of the examination area of ​​the object being examined in physical space, wherein the second representation in physical space includes a representation of the examination area to which a second amount of contrast agent is applied during medical imaging techniques, wherein the second amount of contrast agent is different from the first amount of contrast agent; The first representation in the entity space is converted into the first representation of the inspection region of the inspection object in the frequency space.

10. The system according to any one of claims 1-3, wherein the at least one processor is further programmed or configured to perform the following operations: The predictive machine learning model is trained based on a training dataset, wherein the training dataset includes: A set of reference representations of the inspection region of each of the multiple inspection objects in the frequency space, wherein each set of reference representations of the inspection region of the inspection object includes: The first reference representation of the inspection area of ​​the inspection object in frequency space; The second reference representation of the inspection area of ​​the inspection object in frequency space; and The third reference representation of the inspection area of ​​the inspection object in frequency space; and The first reference representation includes a reference representation in frequency space of the examination area during medical imaging techniques where no amount of contrast agent is applied or where a first amount of contrast agent is applied during medical imaging techniques. The second reference representation includes a reference representation in frequency space of the examination area during medical imaging techniques when a second amount of contrast agent is applied. The third reference representation includes a reference representation in frequency space of the examination area where a third amount of contrast agent is applied during medical imaging techniques.

11. The system according to claim 10, wherein, When training the predictive machine learning model, the at least one processor is programmed or configured to perform the following operations: Minimize the amount of error provided by the error function, which quantifies the deviation between the predicted representation of the examination area of ​​the subject in frequency space during medical imaging techniques when a third amount of contrast agent is applied and the third reference representation of the examination area of ​​the subject in frequency space.

12. The system according to any one of claims 1-3, wherein the at least one processor is further programmed or configured to perform the following operations: The predictive machine learning model is trained based on a training dataset, wherein the training dataset includes: A set of simplified reference representations of the inspection region of each of the multiple inspection objects in the frequency space, wherein each set of simplified reference representations of the inspection region of the inspection object includes: The first simplified reference representation of the inspection area of ​​the inspection object in frequency space; The second simplified reference representation of the inspection area of ​​the inspection object in frequency space; and The third simplified reference representation of the inspection area of ​​the inspection object in frequency space; and The first simplified reference representation includes a reference representation of the examination region in the frequency space that includes a portion of the center of the frequency space, wherein the first reference representation includes a reference representation of the examination region in which no amount of contrast agent is applied during medical imaging or in which a first amount of contrast agent is applied during medical imaging. The second simplified reference representation includes a reference representation of the examination region in the frequency space that includes a portion of the center of the frequency space, wherein the second reference representation includes a reference representation of the examination region during the medical imaging technique when a second amount of contrast agent is applied; The third simplified reference representation includes a reference representation of the examination region in the frequency space that includes a portion of the center of the frequency space, and the third reference representation includes a reference representation of the examination region during the medical imaging technique when a third amount of contrast agent is applied.

13. The system of claim 12, wherein, When training the predictive machine learning model, the at least one processor is programmed or configured to perform the following operations: Minimize the amount of error provided by the error function, which quantifies the deviation between the predicted simplified representation of the examination area of ​​the subject in frequency space during medical imaging techniques when a third amount of contrast agent is applied and the third simplified reference representation of the examination area of ​​the subject in frequency space.

14. A computer program product for providing a prediction of a representation of an examination area, the prediction being generated using a medical imaging technique involving contrast agents, the computer program product comprising at least one non-transitory computer-readable medium, the at least one non-transitory computer-readable medium comprising one or more instructions, the one or more instructions causing the at least one processor to perform the following operations when executed by at least one processor: A first representation of the examination area of ​​the object being examined in the frequency space, wherein the first representation includes a representation of the examination area in which no amount of contrast agent is applied during the medical imaging technique or in which a first amount of contrast agent is applied during the medical imaging technique; A second representation of the examination area of ​​the object being examined in the frequency space, wherein the second representation includes a representation of the examination area during medical imaging techniques when a second amount of contrast agent is applied, wherein the second amount of contrast agent is different from a first amount of contrast agent; The system provides input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent is applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent. Receive the output of the predictive machine learning model based on the input; The output of the predictive machine learning model is converted into a predicted representation of the inspection region of the inspected object in the entity space. as well as Provides a predicted representation of the inspection area of ​​the inspection object in the physical space.

15. A method for providing a prediction of an examination region, the prediction being generated using a medical imaging technique involving contrast agents, the method comprising: At least one processor receives a first representation of the examination area of ​​the examination object in the frequency space, wherein the first representation includes a representation of the examination area during medical imaging techniques in which no amount of contrast agent is applied or in which a first amount of contrast agent is applied during medical imaging techniques. The at least one processor receives a second representation of the examination area of ​​the examination object in the frequency space, wherein the second representation includes a representation of the examination area during medical imaging techniques when a second amount of contrast agent is applied, wherein the second amount of contrast agent is different from a first amount of contrast agent; The at least one processor provides input to a predictive machine learning model, wherein the input to the predictive machine learning model includes at least a portion of a first representation and at least a portion of a second representation, wherein the predictive machine learning model includes a trained machine learning model configured to provide a prediction as output of a representation in frequency space of an examination region to which a third amount of contrast agent is applied during medical imaging techniques, wherein the third amount of contrast agent is greater than the first amount of contrast agent and the second amount of contrast agent. The at least one processor receives the output of the predictive machine learning model based on the input; The at least one processor is used to convert the output of the predictive machine learning model into a predicted representation of the inspected region of the inspected object in the entity space. as well as The at least one processor provides a predictive representation of the inspection area of ​​the inspection object in the physical space.

Images