Ultrasound contrast agent indicator

Abstract

Methods and systems for ultrasound imaging that provide an automatic assessment of intrinsic image quality that can be used, for example, to determine that intrinsic image quality is below a threshold and output an indication that: (i) it is possible to improve intrinsic image quality using an ultrasound enhancer; or (ii) it is impossible to improve the intrinsic image quality.

Description

Technical Field

[0001] The present invention generally relates to ultrasound imaging and, in particular, to automatically determining whether a contrast agent should be used when imaging a patient. Background Art

[0002] It is estimated that in adult patients, cardiac ultrasound examinations used to evaluate important cardiac function conditions (such as left ventricular volume, ejection fraction, regional wall motion (RWM), and other parameters) are non-diagnostic up to 20% of the time in routine scans and up to 30% of the time in the intensive care environment. This is because despite significant progress in ultrasound transducers and equipment quality, the increasing obesity of the population, often combined with lung and other diseases, makes it virtually impossible to obtain key diagnostic parameters in a very large number of patients. In such cases, ultrasound contrast agents are often used to allow the generation of diagnostic-grade images. The Heart Association has developed guidelines for when to indicate contrast agents. The American Society of Echocardiography 2018 Contrast Agent Use Guidelines state that: According to the 2008 ASE guidelines, for a routine resting echocardiogram study, when two or more LV segments cannot be adequately visualized for the assessment of LV function (LVEF and regional wall motion assessment) and / or in settings where the study indicates the need for an accurate analysis of RWM, a UEA should be used. The ASE guidelines for contrast agents for sonographers state that: As long as there are suboptimal images for quantifying cardiac chamber volume and ejection fraction and for assessing regional wall motion, ultrasound contrast agents should be used. Suboptimal images can be defined as the inability to detect two or more consecutive segments in any of the three apical windows. If the spectral signals for quantifying velocity and pressure gradients are inadequate, Doppler blood flow estimation using UCA should be performed during a resting or stress study. Applying these guidelines in practice is difficult because they involve many subjective interpretations and require extensive experience. Tracking whether the apical view fails to detect consecutive segments of the myocardium is affected by human error. Therefore, the decision to use these agents can be a complex one, and typically the person performing the scan is not a doctor but a sonographer or technician who does not have the knowledge and expertise to make the call as to whether a contrast agent should be used. As a result, diagnosis and care may be delayed, and the cost of patient management may increase. Therefore, there is a need for methods and systems that can assist users in making such determinations. Summary of the Invention

[0003] In one aspect, a method for ultrasound imaging is described herein.

[0004] In some aspects, the method includes using an ultrasound imaging system to acquire a plurality of ultrasound images of at least a portion of an organ of an object for performing a diagnostic procedure on the object. In some aspects, the method includes processing the acquired plurality of ultrasound images by a trained machine learning model to determine an inherent image quality of the plurality of ultrasound images. In some aspects, the method includes automatically determining that the inherent image quality of the plurality of ultrasound images is less than a required threshold quality associated with the diagnostic procedure, at least in part based on an output of the trained machine learning model.

[0005] In some aspects, the method includes outputting an indication to a user of the ultrasound system, at least in part based on the automatic determination.

[0006] In some aspects, the indication may include an indication that an ultrasound contrast agent is needed to improve the inherent image quality to a level required for successful completion of the diagnostic procedure. In some aspects, the indication may include an indication that it is expected that using an ultrasound contrast agent will not improve the inherent image quality to a level required for successful completion of the diagnostic procedure.

[0007] In some aspects, the indication is provided in real time during the diagnostic procedure. In some aspects, the trained machine learning model is trained using training data that includes a plurality of data points labeled to indicate that an expert has determined that an ultrasound contrast agent is needed. In some aspects, the training data includes a plurality of data points captured using an ultrasound contrast agent.

[0008] In some aspects, the required threshold is adjustable by the user. In some aspects, the automatic determination includes determining, by the machine learning model, that one or more views of the organ associated with the diagnostic procedure have been captured at a probe position expected to provide clinically acceptable image quality and / or have been captured in a mode expected to provide clinically acceptable image quality, while the inherent image quality remains below the required threshold.

[0009] In some aspects, determining that one or more views of the organ associated with the diagnostic procedure have been captured at a probe position expected to provide clinically acceptable image quality and / or have been captured in a mode expected to provide clinically acceptable image quality is performed using a trained probe guidance machine learning model. In some aspects, the one or more views include a plurality of views, and the inherent quality of the plurality of views is integrated to determine that the inherent image quality of the plurality of ultrasound images is less than the required threshold quality associated with the diagnostic procedure.

[0010] In some aspects, the organ is the heart. In some aspects, the plurality of acquired ultrasound images are two-dimensional ultrasound images. In some aspects, the diagnostic procedure includes a cardiac function measurement, and the required threshold intrinsic quality associated with the diagnostic procedure is the minimum intrinsic quality required for performing the cardiac function measurement. In some aspects, the cardiac function measurement includes a measurement of the ejection fraction and / or a measurement of the left ventricular function. In some aspects, the minimum intrinsic image quality includes the visibility of a minimum number of myocardial segments in the plurality of acquired ultrasound images.

[0011] In some aspects, the process of determining the intrinsic image quality of the plurality of ultrasound images includes analyzing the individual image quality of a plurality of myocardial segments of the subject's heart through a machine learning model. In some aspects, the analysis is performed to determine the individual image quality of a plurality of myocardial segments of the subject's heart across multiple views.

[0012] In some aspects, the plurality of acquired ultrasound images include Doppler ultrasound images. In some aspects, the method further includes detecting the presence or absence of valvular pathology in the plurality of acquired ultrasound images.

[0013] In some aspects, automatically determining that the intrinsic image quality of the plurality of ultrasound images is less than the required threshold intrinsic quality associated with the diagnostic procedure includes: estimating the expected blood flow parameters and comparing the expected blood flow parameters with the measured blood flow parameters obtained from the Doppler ultrasound images. In some aspects, the indication to the user includes an alert that the gap between the expected parameter and the measured parameter indicates impaired Doppler signal. In some aspects, both the expected blood flow parameter and the measured blood flow parameter are velocities.

[0014] In some aspects, the plurality of acquired ultrasound images are two-dimensional ultrasound images. In another aspect, described herein is a non-transitory computer-readable medium storing instructions that, when executed by a processor of a computer, cause the computer to perform any of the methods described herein.

[0015] In another aspect, described herein is an ultrasound imaging system. In some aspects, the ultrasound imaging system includes an ultrasound imaging probe and a computing system. In some aspects, any of the ultrasound systems described herein can be configured to perform any of the methods described herein. In some aspects, the ultrasound imaging system may further include any non-transitory computer-readable storage medium described herein.

[0016] Another aspect of the present disclosure provides a system that includes one or more computer processors and a computer memory coupled thereto. The computer memory includes machine-executable code that, when executed by the one or more computer processors, implements any of the methods above or elsewhere herein.

[0017] Through the following detailed description, additional aspects and advantages of the present disclosure will become readily apparent to those skilled in the art, in which only illustrative aspects of the present disclosure are shown and described. As will be recognized, the present disclosure is capable of having other aspects and different aspects, and several details thereof can be modified in various obvious aspects, all of which do not depart from the present disclosure. Accordingly, the drawings and description are to be regarded as illustrative in nature and not restrictive.

[0018] Incorporated by reference

[0019] All publications, patents, and patent applications mentioned in this specification are incorporated herein by reference to the extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference. To the extent that the publications and patents or patent applications incorporated by reference conflict with the disclosure contained herein, this specification is intended to supersede and / or take precedence over any such conflicting material. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the invention will be obtained by reference to the following detailed description that sets forth illustrative aspects of the invention, in conjunction with the accompanying drawings (also referred to herein as "FIGURES" and "FIG.") in which:

[0021] Figure 1 An example workflow according to the methods and systems described herein is shown.

[0022] Figure 2 An alternative example workflow according to the methods and systems described herein is shown.

[0023] Figure 3 A third example workflow according to the methods and systems described herein is shown.

[0024] Figure 4 A fourth example workflow according to the methods and systems described herein is shown.

[0025] Figure 5 A fifth example workflow according to the methods and systems described herein is shown.

[0026] Figure 6 A non-contrast apical two-chamber echocardiogram is shown, showing a poorly defined endocardium (arrow) and not showing more than two consecutive segments.

[0027] Figure 7 Improved visualization of the left ventricular wall with ultrasound contrast agent (UCA) in a 58-year-old obese male after UCA administration is shown.

[0028] Figure 8 An example segmentation of the heart useful in implementing the methods and systems described herein is shown. This shows how different echo views acoustically penetrate different myocardial segments. In terms of segments, the parasternal long axis and the apical 3-chamber (also known as the apical long axis) are the same. Segments 14 and 16 are shown in two of the three apical views. The apical 5 view is essentially the same as the apical 4, but the transducer is slightly tilted to show the aortic valve. AP5 is used more for aortic valve Doppler examination compared to the assessment of myocardial or wall motion. The names associated with the numbered segments are provided on the next slide.

[0029] Figure 9 An alternative example segmentation of the heart that can be used to implement the methods and systems described herein is shown.

[0030] Figure 10 A computer system programmed or otherwise configured to implement the methods provided herein is shown. Detailed Description

[0031] Although various aspects of the invention have been shown and described herein, it will be apparent to those skilled in the art that these aspects are provided by way of example only. Many variations, changes, and substitutions will occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the aspects of the invention described herein may be employed.

[0032] Whenever the terms "at least", "greater than", or "greater than or equal to" precede the first of a series of two or more numerical values, the terms "at least", "greater than", or "greater than or equal to" apply to each numerical value in the series. For example, greater than or equal to 1, 2, or 3 is equivalent to greater than or equal to 1, greater than or equal to 2, or greater than or equal to 3.

[0033] When the terms "not greater than", "less than", or "less than or equal to" precede the first of a series of two or more numerical values, the terms "not greater than", "less than", or "less than or equal to" apply to each numerical value in the series. For example, less than or equal to 3, 2, or 1 is equivalent to less than or equal to 3, less than or equal to 2, or less than or equal to 1.

[0034] As used herein, the terms "ultrasound enhancer", "UEA", "contrast agent", and "contrast medium" are used interchangeably and refer to a reagent that can be used to increase the contrast in the acquired ultrasound images when administered to the object being imaged. Examples of ultrasound enhancers include injecting sulfur hexafluoride lipid microspheres and / or perfluoropropane lipid microspheres into the object.

[0035] Certain inventive aspects of the present disclosure contemplate numerical ranges. When a range is present, the range includes the range endpoints. Additionally, each sub-range and the values within the range are present as if explicitly written out. The term "about" or "approximately" may indicate an acceptable error range around a particular value, which will depend in part on how the value is measured or determined, such as the limitations of the measurement system. For example, in accordance with the practice in the art, "about" may mean within one or more standard deviations. Alternatively, "about" may represent a range of up to 20%, up to 10%, up to 5%, or up to 1% of a given value. Where a particular value is described in the present application and the claims, unless otherwise stated, it may be assumed that the term "about" indicates an acceptable error range around the particular value.

[0036] The improvements disclosed herein provide methods and systems for image quality assessment in generating images during cardiac ultrasound scanning, where the need for an ultrasound contrast agent (UAE, commonly referred to as a "contrast agent") to be used is automatically evaluated. This evaluation can provide an objective determination that the image quality of the patient being scanned is non-diagnostic and thus the use of a contrast agent is needed and is consistent with practice guidelines. The prediction of the need for a contrast agent is generated using machine learning methods. These methods and systems can provide multiple operations that act independently or in coordination with each other.

[0037] Certain improvements disclosed herein capture and / or further extend the knowledge of multiple expert sonographers in a single method or system. Specifically, sonographers understand the anatomical structures or features they see in an ultrasound image and the diagnostic quality of the image, and how and where to move the ultrasound probe to acquire a desired image based on the current ultrasound image. The platforms, systems, and methods disclosed herein utilize machine learning techniques such as deep learning to capture and / or exceed this perceptual ability so that a wider range of clinical users, including non-experts, can access high-quality ultrasound imaging, particularly for echocardiography.

[0038] Deep learning is a form of machine learning based on artificial neural networks. Deep learning involves a variety of techniques, but common attributes include composing simple computational elements into a layer, composing many layers into a deep stack, and using supervised learning to adjust the parameters of the elements.

[0039] Diagnostic image quality

[0040] A particular challenge in ultrasonic medical imaging is to accurately determine what probe posture or movement will result in a clinical or diagnostic quality image. As used herein, image quality (e.g., diagnostic quality or clinical quality) can be used to refer to one or more aspects of the image quality. In some aspects, image quality refers to an image that can be viewed by a trained expert or a machine learning tool in a manner that enables identification of anatomical structures and allows for a diagnostic interpretation. In some aspects, image quality refers to an image in which the target is displayed in a clear and well-defined manner. For example, in the case of minimal external noise or clutter, the grayscale display shows subtle variations in tissue type and texture, the blood flow signal is clear and distinct, the frame rate is high, thus providing an accurate depiction of tissue or blood flow movement, the boundaries between tissue types or blood flow and blood vessels or other structures are well resolved, ultrasonic artifacts such as grating and sidelobes are minimized, there is no acoustic noise, the locations where measurements are made in the image are obvious and clear, or any combination thereof, depending on the nature of the ultrasound examination. In some aspects, image quality refers to an image that contains the necessary anatomical targets to represent a standard diagnostic view. For example, the apical four-chamber view of the heart should show the apex of the heart, the left and right ventricles, the myocardium, the mitral and tricuspid valves, the left and right atria, and the interatrial septum. As another example, the long-axis view of the carotid artery at the bifurcation should show the common carotid artery, the external carotid artery, and the carotid bulb. In some aspects, image quality refers to an image in which a disease condition, abnormality, or pathology is well visualized. For example, medical images can be labeled by cardiologists, radiologists, or other healthcare professionals as to whether they are considered to have a disease condition, abnormality, or pathology that is well visualized, and then used to train machine learning algorithms to distinguish between images based on image quality.

[0041] In some aspects, image quality means that there is some combination of these aforementioned characteristics. It will be necessary to provide effective navigation guidance to ensure that the captured ultrasound images meet the combination of these image quality characteristics necessary to produce an overall clinical or diagnostic quality image, because in ultrasound imaging, patient presentation may pose a challenge to obtaining high-resolution, low-noise images. For example, when trying to assess blood flow in the kidneys of an obese person, it may be particularly challenging to obtain a sufficiently strong blood flow Doppler signal because the kidneys are so deep below the adipose tissue. In patients who smoke for a long time, lung disease may make it difficult to obtain high-quality cardiac images. These conditions are very common, and in this case, image quality may mean that the image may be suboptimal in terms of noise and resolution, but still provide sufficient information for making a diagnosis. In a similar manner, patient presentation and pathology may make it impossible to obtain a view showing all anatomical components of a standard, normative image. For example, a technically difficult heart patient may make it impossible to obtain an apical four-chamber view in which all four chambers are well defined, but if some images, for example, show the left ventricle well, this can be considered a quality image because many key diagnostic conclusions can only be drawn from this image.

[0042] In some aspects, the anatomical views used in the present disclosure include one or more of a probe position or window, an imaging plane, and a visualized region or structure. Examples of probe positions or windows include parasternal, apical, subcostal, and suprasternal notches. Examples of imaging planes include long axis (LAX), short axis (SAX), and four chambers (4C). Examples of visualized regions or structures include two chambers, aortic valve, mitral valve, etc. For example, the anatomical views may include parasternal long axis (LV inflow / outflow), RV inflow + / - RV outflow, parasternal short axis (aortic valve level, mitral valve level, papillary muscle level, apical LV level), apical four chambers, apical five chambers, apical two chambers, apical three chambers, subcostal four chamber views, subcostal short axis and long axis, suprasternal long axis (aortic arch), and suprasternal short axis (aortic arch).

[0043] Intrinsic image quality

[0044] Intrinsic image quality generally refers to the cumulative effect of one or more factors on diagnostic quality, where the one or more factors contribute to or lack of diagnostic quality, which is not caused by operator error (e.g., such as poor probe positioning and / or use of inappropriate acquisition modes). Examples of factors that may contribute to poor intrinsic image quality can include obesity of the object being imaged, scarring due to long-term smoking. As described above regarding diagnostic image quality, the effects of these factors can vary from view to view and from organ feature to organ feature. When the intrinsic image quality of an organ feature or the views necessary for that particular procedure is too low, completion of the diagnostic procedure can thus be precluded. The use of ultrasound contrast agents (contrast agents) can improve the intrinsic image quality, thus allowing completion of diagnostic procedures where the intrinsic image quality would otherwise be too low. However, in some cases, the use of contrast agents cannot adequately compensate for the underlying factors contributing to poor intrinsic image quality.

[0045] The methods and systems described herein can provide even trained ultrasound clinicians with insight by providing an automated assessment of whether it is desirable to use contrast agents to improve the intrinsic image quality in a selected diagnostic procedure.

[0046] The methods and systems described herein can provide real-time ultrasound image quality determination, which can evaluate the diagnostic quality of ultrasound images (such as echocardiogram images), and can further determine the intrinsic image quality of one or more features or views necessary to perform the selected diagnostic procedure. In some cases, the image quality model is adapted and used to determine that the cardiac image specifically does not meet the criteria for diagnostic-level quality for left ventricular function assessment and / or Doppler assessment.

[0047] In some cases, the methods and systems described herein allow for automation of cardiac left ventricular (LV) function assessment, as well as image quality determination specific to LV function. An image quality score can be given using the methods described herein, and an additional level of quality assessment can be performed for individual measurements and parameters required for the selected diagnostic procedure.

[0048] For example, a first level of quality assessment can determine quality at a coarser or more general diagnostic level (e.g., overall diagnostic quality, which includes the effect of intrinsic image quality), and a second level can apply a quality assessment process specific to the prediction accuracy of the selected diagnostic procedure (e.g., measurements of cardiac volume, left ventricular ejection fraction measurement, and / or regional wall motion scoring).

[0049] In some cases, the general diagnostic level can include separating the intrinsic image quality from the overall diagnostic quality.

[0050] The methods and systems described herein can be implemented according to a number of alternative workflows. For example, Figure 1The workflow shown in, for example, using an imaging probe of an ultrasound imaging system (e.g., a transducer of a two-dimensional ultrasound imaging system) to acquire an ultrasound image 101; for example, using a machine learning model as described herein to determine the inherent image quality 103, the machine learning model being trained using images annotated with respect to whether a contrast agent is needed; and for example, outputting an indication of whether an ultrasound enhancer (contrast agent) will improve the inherent quality of subsequently acquired ultrasound images by providing an alert to the user of the ultrasound imaging system indicating that they use a contrast agent and / or re-acquire subsequent images using a contrast agent 105.

[0051] In Figure 2 An alternative example workflow shown in includes: selecting a diagnostic procedure 201, for example, a procedure that can include obtaining a series of views of an object's heart to evaluate valve function can be selected. The workflow can then include acquiring an ultrasound image 203 based on the selected diagnostic procedure, for example, to obtain the heart views necessary to evaluate valve function. The workflow can also include determining the inherent image quality of the acquired image based on the diagnostic procedure 205, for example, using a machine learning model as described herein, the machine learning model being trained using images annotated with respect to whether a contrast agent is needed, wherein the images can be annotated based at least in part on the quality and / or view requirements associated with the diagnostic procedure or a combination thereof (e.g., for the evaluation of valve function, the determination of inherent quality can be based in part on the relative inherent quality required to successfully distinguish a healthy valve from an unhealthy valve). In some cases, the workflow can include outputting an indication of whether an ultrasound enhancer (contrast agent) is expected to improve the inherent quality of subsequently acquired ultrasound images required to complete the diagnostic procedure 207, for example, when it is determined that the inherent quality is too low to distinguish an unhealthy valve from a healthy valve in the acquired images, a machine learning algorithm as described herein can be used to determine whether a contrast agent is expected to improve the inherent quality of the subsequently acquired images. If a contrast agent is expected to improve the inherent quality (e.g., when the machine learning model determines that the poor inherent quality is caused by obesity and / or tissue damage from smoking), the user of the ultrasound system can be instructed to re-acquire the image using a contrast agent. If a contrast agent is not expected to help, the user can be warned of the poor inherent image quality and instructed not to attempt to re-acquire with a contrast agent and / or to consult a cardiologist before doing so.

[0052] In Figure 3Another alternative example workflow shown in FIG. includes selecting a diagnostic procedure 301. For example, a procedure that includes obtaining a series of views of an object's heart to evaluate heart function (such as ejection fraction measurement) can be selected. The workflow can also include acquiring ultrasound images 303 at least partially based on the selected diagnostic procedure 301, for example, to obtain the heart views necessary for calculating the ejection fraction at a sufficient quality level for calculating the ejection fraction. The workflow can also include determining the intrinsic image quality 305 of the acquired images partially based on the diagnostic procedure, for example, using a machine learning model as described herein, which is trained using images annotated with regard to whether a contrast agent is needed, where the images can be annotated at least partially based on the quality and / or view requirements associated with the diagnostic procedure or a combination thereof (e.g., for the calculation of ejection fraction, the determination of intrinsic quality can be partially based on the minimum relative intrinsic quality required to successfully calculate an accurate ejection fraction). The workflow can also include determining an intrinsic image quality threshold 307 partially based on probe guidance and / or the diagnostic procedure. For example, a probe guidance machine learning model configured and / or trained to provide probe guidance can be used in combination with the method for determining intrinsic quality to isolate the effects of imaging probe position or other operator-derived factors (e.g., if the probe guidance machine learning model determines that the operator has held the probe in a clinically acceptable position for a given view, the poor clinical quality of the acquired image can be evaluated as more likely due to poor intrinsic image quality). The workflow can include outputting an indication 309 as to whether an ultrasound contrast agent (contrast agent) will improve the intrinsic image quality of subsequently acquired images. For example, when it is determined that the clinical quality is too low to calculate the ejection fraction from the acquired images, a machine learning algorithm as described herein can be used to determine whether this is due to positioning (or other user error), for example, based on the output of the probe guidance model, or whether the intrinsic quality of the acquired images is too low. If the poor quality is due to probe position or other operator error, instructions can be provided to guide the user to an improved probe position. If the intrinsic quality is too low (meaning the probe is correctly positioned and any applicable operator errors have been resolved), the machine learning model described herein can be used to determine whether a contrast agent is expected to improve the intrinsic quality of subsequently acquired images. If a contrast agent is expected to improve the intrinsic quality (e.g., when the machine learning model determines that the poor intrinsic quality is caused by obesity and / or tissue damage from smoking rather than operator error), the user of the ultrasound system can be instructed to re-acquire the image using the contrast agent. If a contrast agent is not expected to help, the user can be warned of the poor intrinsic image quality and instructed not to attempt to re-acquire with the contrast agent and / or to consult a cardiologist before doing so.

[0053] Figure 4Another example workflow is shown, which includes: selecting a diagnostic procedure 401, for example, a procedure that can include obtaining a series of views of an object's heart to evaluate a pathology such as aortic valve stenosis. The workflow can also include acquiring ultrasound images 403 at least in part based on the selected diagnostic procedure 401, for example, to obtain the heart views necessary for a successful diagnosis of the pathology. The workflow can also include: determining the inherent image quality 405 of the acquired images in part based on the diagnostic procedure, for example, using a machine learning model as described herein, which is trained using images annotated with respect to whether a contrast agent is required, where the images can be annotated at least in part based on the quality and / or view requirements associated with the diagnostic procedure or a combination thereof (e.g., for the presence and / or severity of the selected pathology, the determination of the inherent quality can be based in part on the minimum relative inherent quality required to successfully classify the severity and / or presence). The workflow can also include determining an inherent image quality threshold 407 in part based on a localization model and / or the diagnostic procedure, for example, a probe position machine learning model configured and / or trained to determine the ideal clinical position of an imaging probe can be used in combination with the method for determining the inherent quality to isolate the effects of the imaging probe position or other operator-derived factors (e.g., if the probe position machine learning model determines that the operator has acquired an image at the clinically ideal probe position, it can be determined that the poor clinical quality of the acquired image is due to poor inherent image quality). The workflow can include outputting an indication 409 as to whether an ultrasound contrast agent (contrast agent) will improve the inherent image quality of subsequently acquired images, for example, when it is determined that the clinical quality is too low to classify the pathology based on the acquired images, a machine learning algorithm as described herein can be used to determine whether this is due to a positioning error (or other user error), for example, based on the output of the probe position machine learning model. Alternatively, the machine learning algorithm described herein can determine whether the inherent quality value of the acquired image is too low. If the poor quality is due to the probe position or other user error, the user can be provided with instructions to continue imaging until a sufficient number of images are captured at the ideal probe position. If the inherent quality is still too low (meaning the probe is correctly positioned and any applicable user errors have been resolved), the machine learning model described herein can be used to determine whether a contrast agent is expected to improve the inherent quality of subsequently acquired images. If a contrast agent is expected to improve the inherent quality (e.g., when the machine learning model determines that the poor inherent quality is caused by obesity and / or tissue damage from smoking rather than user error), the user of the ultrasound system can be instructed to re-acquire the image using the contrast agent. If a contrast agent is not expected to help, the user can be warned of the poor inherent image quality and instructed not to attempt to re-acquire with the contrast agent and / or to consult a cardiologist before doing so.

[0054] AtFigure 5 Another example workflow is shown in Figure 5 , which includes: selecting a diagnostic procedure 501. For example, a procedure can be selected that includes obtaining a series of views of an object's heart to evaluate a pathology such as aortic valve stenosis. The workflow can also include acquiring ultrasound images 503 at least partially based on the selected diagnostic procedure 501. For example, to obtain the heart views necessary for a successful diagnosis of the pathology. The workflow can also include, for example, using Figure 3 the probe guidance model described in the workflow of Figure 3 , Figure 4 the probe position model described in the workflow of Figure 4 , or a combination thereof to determine whether the contemporaneous probe position is expected to produce a clinically quality image 505. If not, the workflow can include acquiring more ultrasound images 503, or if so, determining whether the intrinsic image quality of the acquired images is below a threshold 507. For example, based on a threshold of the minimum intrinsic quality of the selected diagnostic procedure (e.g., as described in the workflow of Figures 2 to 4 Figures 2 to 4 ), or a minimum absolute intrinsic threshold below which little or no diagnostic information can be obtained from the acquired images. If the expected probe position produces a clinically quality image and the intrinsic image quality of the acquired images is above the threshold, the diagnostic procedure 509A can be completed without a contrast agent. For example, allowing for a direct classification of the pathology, calculation of the ejection fraction, determination of valve function, and / or other diagnostic of interest. If the intrinsic image quality is below the threshold, the workflow includes, for example, using a machine learning model as described herein to determine whether an ultrasound contrast agent (contrast agent) is expected to improve the intrinsic quality 509B. The machine learning model is trained using images annotated with regard to whether a contrast agent is needed. If not, the user is warned of the poor image quality and / or that the expected contrast agent does not allow for the completion of the diagnostic procedure 511A, for example, by an alert provided in the user interface of the ultrasound imaging system. If the contrast agent is expected to improve the intrinsic image quality, the user is instructed to use an ultrasound contrast agent (contrast agent) to complete the diagnostic procedure 511B.

[0055] In some cases, an algorithm for continuous segment detection of a target organ (e.g., an object's heart) is used. Such an algorithm can, in some cases, use local image segmentation methods (e.g., by segmenting training images and / or acquired images to be analyzed into multiple segments, such as as shown in Figure 8 or Figure 9 Figure 9 ). The segmentation-based algorithm can be used to determine whether specific, continuous, or discontinuous segments of the myocardium are clearly visible in an image with diagnostic image quality. Such specific segment detection can be performed for each view, thereby identifying one or more parts of the heart that are not adequately imaged. The methods and systems described herein can co-track segments across multiple views to give an overall picture of which segments fail during a complete examination.

[0056] In some cases, the methods and systems described herein predict whether an image will cause an expert to use a contrast agent. Components of the method or system can be algorithms trained on a dataset from an expert laboratory where decisions to use contrast agents are made. The input can be images and labels that have used a contrast agent. In some cases, the methods and systems described herein may not require detailed segmentation analysis. This can provide another level of confidence indicating the contrast agent.

[0057] The methods and systems described herein can provide cross - view tracking of image quality related to contrast agent indication. The method and system can include tracking individual image views, such as various apical views, and marking the presence of sub - optimal images that will require a contrast agent. For example, the device can note that the apical 4 - chamber view has been lost in a myocardial segment, making the image discontinuous. Then, if the same condition is seen in, for example, the apical 2 - chamber view and the apical 3 - chamber view, the device keeps track of this condition for the user. In some cases, the inherent quality is at least partially based on multiple segments of the target organ that are not adequately imaged in multiple views.

[0058] The methods and systems described herein can provide an alert or other indication to the user that a contrast agent is needed. The device can generate an output to the user to warn them that a contrast agent is required to complete the selected diagnostic study. In some aspects, such an alert or other indication can be presented in real - time during scan acquisition. In some aspects, such an alert or other indication can be provided for a single image clip. In some aspects, such an alert or other indication can be provided for a single view. In some aspects, such an alert or other indication can be provided for a group of multiple views. In some aspects, such an alert or other indication can be provided for two - dimensional or Doppler modes.

[0059] In some cases, the methods and systems described herein can determine that a contrast agent is not needed because the inherent quality is sufficient and enough continuous segments are visualized to perform the selected diagnosis with confidence. This aspect can be used as a form of reassurance to the user, and / or as an aid in making a clinical decision to avoid over - use of contrast agents. According to one aspect, when it is determined that a contrast agent is not needed, there can be no output as no procedural change is required. The methods and systems described herein can also operate on images that have been previously acquired.

[0060] User-selectable parameter

[0061] Because individual users can have legitimate differences in the image quality thresholds used to indicate the need for contrast agent, depending on the diagnostic procedures selected and the individual requirements of the user, the methods and systems described herein can allow the user to configure and / or customize settings. For example, the settings can be configured for a different number of non-visual segments used to create a contrast agent indication alert or the number of views required for such sub-optimal images to configure the settings.

[0062] For example, the American Society of Echocardiography (ASE) guidelines for determining contrast agent for 2D images can be subjective. The determination that a segment is not well visualized is typically an arbitrary determination by an expert user. However, in many cases, it is obvious that a segment is not well visualized, especially the endocardial border depicting the myocardial edge at the blood pool in the ventricle. The assessment of inadequate Doppler and the possible need for contrast agent can be even more difficult.

[0063] The Doppler mode can be quantitative, displaying blood flow velocity in real time. In valvular pathology, especially valvular stenosis, the blockage at the valve outlet impedes flow. Blood moves through the occluded valve at a higher velocity and pressure. The abnormal high velocity can be used to detect and quantify the stenosis. If no significant and high velocity flow is seen on spectral Doppler, the user can infer that the valve is not stenotic. But this may be because the Doppler image quality is impaired. Figure 6 The illustration in shows a dramatic example of this situation. Different from 2D, when the user can view the apical view and knows that all myocardial segments should be clearly seen in the optimal image, in abnormal Doppler conditions, there is no such simple and reliable method to evaluate the quality. The methods and systems described herein can be used to automatically detect poor inherent image quality, such as Figure 6 shown in, so as to warn the user whether it is expected that ultrasound contrast agent can improve the inherent quality of the image. In the case of indicating ultrasound contrast agent, the user administers the ultrasound contrast agent to the subject, and subsequent images showing improved inherent image quality are collected, for example, as Figure 7 shown.

[0064] The methods and systems described herein can address this problem by detecting the presence of valvular stenosis from 2D images. In some cases, a classification of the stenosis severity can be generated. In some aspects, an estimate of the expected flow velocity can be generated from 2D images. In some cases, if the Doppler flow velocity does not match the flow velocity predicted from the 2D method, the user can be warned that the difference may be due to the Doppler image quality, where a contrast agent Doppler study can be indicated.

[0065] In some aspects, real-time suggestions are made to the user to automatically use contrast agents in the echo. In these aspects, clinical errors can be reduced by providing insights to the users of the systems and methods described herein. Such aids can reduce the amount of time and effort they spend during scanning by helping sonographers and / or physicians decide more quickly and with less effort when to use contrast agents.

[0066] The methods and systems described herein can reduce the need for human expertise in depth in the diagnosis of subjects by ultrasound imaging by reducing the difficulty for sonographers and even many physicians to make an accurate determination of the need for contrast agents. In some cases, the methods and systems described herein can allow less experienced users to determine whether contrast agents should be used.

[0067] The methods and systems described herein can combine multiple algorithmic or methodological features, such as multi-image quality assessment methods, LV function-specific quality assessment methods, measurement-specific image quality assessment methods, training using a dataset with contrast agents to improve the reliability of one or more automated suggestions. The methods and systems described herein can track specific myocardial segments, including across multiple views, and monitor compliance of the images with practice guidelines for contrast agent use.

[0068] In some cases, the ability to customize parameters and thresholds for contrast agent indication conditions based on algorithmic output and individual clinical practice can be provided to users of the methods and systems described herein.

[0069] In some cases, for 2D or 3D imaging, algorithms can be used to notify the user that the image meets the requirements for using contrast agents. In some aspects, multiple views are integrated for evaluation. In some cases, the determination of quality is performed from the perspective of a specific, selected diagnostic procedure (e.g., a specific measurement). In some aspects, the analysis of the image quality of individual segments of the myocardium is used in contrast agent need prediction. In some cases, the prediction is performed at the single view level and / or at the multiple view level. In some cases, one or more machine learning algorithms used in the prediction are trained using training data labeled from studies where experts have determined that contrast agents are needed.

[0070] In some aspects, for Doppler imaging, 2D-based algorithms are used that detect valve pathologies from the operation, estimate and expect blood flow parameters such as velocity, and then compare the estimated velocity with the actually obtained Doppler velocity. In some aspects, the user is provided with a warning or other indication that the discrepancy indicates an impaired Doppler signal and that contrast agents should be used.

[0071] In some aspects, during the study, individual myocardial segments and / or multiple views are tracked in image classification to determine whether guideline compliance indicates a need for contrast agent. In some aspects, a quality assessment of non-contrast images is created to indicate the need for contrast.

[0072] It should also be noted that the echocardiography literature includes different thresholds for indicating the need for contrast agent. Some suggest including all three major apical views before making a decision. Others have found that contrast agent is beneficial when used in patients with only two segments that are insufficiently evaluated in the apical four-chamber or apical two-chamber views. Other studies have defined suboptimal echocardiography as the inadequate depiction of at least two of the six segments of the left ventricular endocardial border in only the apical four-chamber view. Some suggest that the segments should be considered contiguous, while others do not require this. This variation also creates a need for customization addressed by the methods and systems described herein.

[0073] The methods and systems described herein can detect a suspected apical thrombus or LV mass in an image, where the thrombus is not clearly visualized. In some aspects, the methods and systems described herein can notify the user that the contrast agent will make the thrombus or mass clearly visible.

[0074] Machine learning algorithm

[0075] The present disclosure provides platforms, systems, and methods for using machine learning algorithms to provide ultrasound image classification. In particular, in some aspects, the machine learning algorithms include deep learning neural networks configured to evaluate ultrasound images. The algorithms can include one or more of a localization algorithm, a scoring algorithm, a probe guidance algorithm, and an intrinsic image quality algorithm. The localization algorithm can include one or more neural networks that estimate the probe position relative to an ideal anatomical view or perspective and / or the distance or deviation of the current probe position from the ideal probe position. The intrinsic image quality algorithm can determine that the intrinsic image quality is below a threshold, at least in part, based on the determination by the localization algorithm that one or more images have been acquired at a probe position where clinical quality images are expected to be obtained.

[0076] The development of each machine learning algorithm spans three phases: (1) dataset creation and management, (2) algorithm training, and (3) design elements necessary to tune product performance and usability. The dataset used to train the algorithm can be generated by obtaining ultrasound images, which are then curated and labeled by expert radiologists, for example, according to localization, scoring, and other metrics. Each algorithm is then trained using the training dataset, which can include one or more different target organs and / or one or more different views of a given target organ. The training dataset for the localization algorithm can be labeled according to known probe pose deviations from the optimal probe pose. A non-limiting description of the training and application of the localization algorithm or estimator can be found in U.S. Patent Application No. 15 / 831,375, the entire text of which is incorporated herein by reference. Another non-limiting description of the localization algorithm and the probe guidance algorithm can be found in U.S. Patent Application No. 16 / 264,310, the entire text of which is incorporated herein by reference. The design elements can include a user interface that includes an omnidirectional guidance feature.

[0077] The machine learning model can include a supervised, semi-supervised, unsupervised, or self-supervised machine learning model. In some cases, one or more ML methods perform classification or clustering of MS data. In some examples, the machine learning methods include classical machine learning methods such as, but not limited to, support vector machines (SVMs) (e.g., one-class SVM, linear or radial kernel, etc.), K-nearest neighbors (KNN), isolation forest, random forest, logistic regression, AdaBoost classifier, extra trees classifier, extreme gradient boosting, Gaussian process classifier, gradient boosting classifier, light gradient boosting, linear discriminant analysis, naive Bayes, quadratic discriminant analysis, ridge classifier, or any combination thereof. In some examples, the machine learning methods include deep learning methods (e.g., deep neural networks (DNNs)) such as, but not limited to, fully connected networks, convolutional neural networks (CNNs) (e.g., one-class CNN), recurrent neural networks (RNNs), transformers, graph neural networks (GNNs), convolutional graph neural networks (CGNNs), multi-layer perceptrons (MLPs), or any combination thereof.

[0078] In some aspects, classical ML methods include one or more algorithms that learn from existing observations (i.e., known features) to predict an output. In some aspects, one or more algorithms perform clustering of data. In some examples, classical ML algorithms for clustering include K-means clustering, mean-shift clustering, density-based spatial clustering of applications with noise (DBSCAN), expectation maximization (EM) clustering (e.g., using a Gaussian mixture model (GMM)), agglomerative hierarchical clustering, or any combination thereof. In some aspects, one or more algorithms perform classification of data. In some examples, classical ML algorithms for classification include logistic regression, naive Bayes, KNN, random forest, isolation forest, decision tree, gradient boosting, support vector machine (SVM), or any combination thereof. In some examples, SVM includes one-class SMV or multi-class SVM.

[0079] In some aspects, deep learning methods include one or more algorithms that learn by extracting new features to predict an output. In some aspects, deep learning methods include one or more layers. In some aspects, deep learning methods include neural networks (e.g., a DNN including more than one layer). A neural network generally includes connected nodes in the network, which can perform functions such as transforming or translating input data. In some aspects, the output from a given node is passed as input to another node. Nodes in the network generally include input units in an input layer, hidden units in one or more hidden layers, output units in an output layer, or a combination thereof. In some aspects, input nodes are connected to one or more hidden units. In some aspects, one or more hidden units are connected to output units. Nodes can generally receive input through input units and generate output from output units using an activation function. In some aspects, the input or output includes a tensor, matrix, vector, array, or scalar. In some aspects, the activation function is a rectified linear unit (ReLU) activation function, sigmoid activation function, hyperbolic tangent activation function, or Softmax activation function.

[0080] The connections between nodes also include weights for adjusting the input data to a given node (i.e., activating or deactivating the input data). In some aspects, the weights are learned by a neural network. In some aspects, the neural network is trained to learn the weights using gradient-based optimization. In some aspects, the gradient-based optimization includes one or more loss functions. In some aspects, the gradient-based optimization is gradient descent, conjugate gradient descent, stochastic gradient descent, or any of its variants (e.g., Adaptive Moment Estimation (Adam)). In some further aspects, backpropagation is used to compute the gradients in the gradient-based optimization. In some aspects, the nodes are organized into a graph to generate a network (e.g., a graph neural network). In some aspects, the nodes are organized into one or more layers to generate a network (e.g., a feedforward neural network, a convolutional neural network (CNN), a recurrent neural network (RNN), etc.). In some aspects, the CNN includes one type of CNN or multiple types of CNNs.

[0081] In some aspects, the neural network includes one or more recurrent layers. In some aspects, the one or more recurrent layers are one or more Long Short-Term Memory (LSTM) layers or Gated Recurrent Units (GRU). In some aspects, the one or more recurrent layers perform sequential data classification and clustering, where data ordering is considered (e.g., time series data). In these aspects, future predictions are made by the one or more recurrent layers based on a sequence of past events. In some aspects, the recurrent layers retain or “remember” important information while selectively “forgetting” information that is not necessary for classification.

[0082] In some aspects, the neural network includes one or more convolutional layers. In some aspects, the input and output are tensors representing variables or attributes (e.g., features) in a dataset, which may be referred to as feature maps (or activation maps). In these aspects, the one or more convolutional layers are referred to as the feature extraction stage. In some aspects, the convolution is one-dimensional (1D) convolution, two-dimensional (2D) convolution, three-dimensional (3D) convolution, or any combination thereof. In other aspects, the convolution is 1D transposed convolution, 2D transposed convolution, 3D transposed convolution, or any combination thereof.

[0083] The layers in a neural network can also include one or more pooling layers before or after the convolutional layer. In some aspects, one or more pooling layers use filters that summarize regions of a matrix to reduce the dimension of the feature map. In some aspects, this down-samples the number of outputs and thus reduces the parameters and computational resources required by the neural network. In some aspects, one or more pooling layers include max pooling, min pooling, average pooling, global pooling, standard pooling, or a combination thereof. In some aspects, max pooling reduces the dimension of the data by taking only the maximum value in the matrix region. In some aspects, this helps capture one or more of the most important features. In some aspects, one or more pooling layers are one-dimensional (1D), two-dimensional (2D), three-dimensional (3D), or any combination thereof.

[0084] The neural network can also include one or more flattening layers that can flatten the input to be passed to the next layer. In some aspects, the input (e.g., the feature map) is flattened by reducing it to a one-dimensional array. In some aspects, the flattened input can be used for the classification of output objects. In some aspects, the classification includes binary classification or multi-class classification of visual data (e.g., images, videos, etc.) or non-visual data (e.g., measurements, audio, text, etc.). In some aspects, the classification includes binary classification of images (e.g., requires contrast agent or does not require contrast agent). In some aspects, the classification includes multi-class classification of text (e.g., recognizing handwritten digits). In some aspects, the classification includes binary classification of measurements. In some examples, the binary classification of measurements includes the classification of the system performance using the physical measurements described herein (e.g., normal or abnormal, normal or abnormal).

[0085] The neural network can also include one or more dropout layers. In some aspects, dropout layers are used during the training of the neural network (e.g., to perform binary or multi-class classification). In some aspects, one or more dropout layers randomly set some weights to 0 (e.g., approximately 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80% of the weights). In some aspects, setting some weights to 0 also sets the corresponding elements in the feature map to 0. In some aspects, one or more dropout layers can be used to avoid overfitting of the neural network.

[0086] The neural network may also include one or more dense layers, which include a fully connected network. In some aspects, information is passed through the fully connected network to generate a predicted classification of an object. In some aspects, an error associated with the predicted classification of the object is also calculated. In some aspects, the error is backpropagated to improve the prediction. In some aspects, one or more dense layers include a Softmax activation function. In some aspects, the Softmax activation function converts a vector of numbers into a vector of probabilities. In some aspects, these probabilities are then used in classification, such as classification of the need for an ultrasound contrast agent, in order to acquire an image with a minimum inherent quality.

[0087] Computer system

[0088] The present disclosure provides a computer system programmed to implement the methods of the present disclosure. Figure 10 A computer system 1001 is shown that is programmed or otherwise configured to evaluate whether an ultrasound contrast agent is expected to improve image quality according to any of the methods described herein. The computer system 1001 may adjust various aspects of the present disclosure. The computer system 1001 may be an electronic device of a user or a computer system remotely located relative to the electronic device. The electronic device may be a mobile electronic device.

[0089] The computer system 1001 includes a central processing unit (CPU, also referred to herein as a “processor” and a “computer processor”) 1005, which may be a single-core or multi-core processor, or multiple processors for parallel processing. The computer system 1001 also includes a memory or memory location 1010 (e.g., random access memory, read only memory, flash memory), an electronic storage unit 1015 (e.g., a hard disk), a communication interface 1020 for communicating with one or more other systems (e.g., a network adapter), and peripheral devices 1025 such as a cache, other memory, data storage, and / or an electronic display adapter. The memory 1010, storage unit 1015, interface 1020, and peripheral devices 1025 communicate with the CPU 1005 via a communication bus (solid lines), such as a motherboard. The storage unit 1015 may be a data storage unit (or data repository) for storing data. The computer system 1001 may be operably coupled to a computer network (“network”) 1030 with the help of the communication interface 1020. The network 1030 may be the Internet, the Internet and / or an extranet, or an intranet and / or extranet that communicates with the Internet. In some cases, the network 1030 is a telecommunications and / or data network. The network 1030 may include one or more computer servers that may implement distributed computing, such as cloud computing. In some cases, with the help of the computer system 1001, the network 1030 may implement a peer-to-peer network, which may enable devices coupled to the computer system 1001 to act as clients or servers.

[0090] The CPU 1005 can execute a series of machine-readable instructions, which can be embodied in a program or software. The instructions can be stored in a memory location, such as the memory 1010. The instructions can be directed to the CPU 1005, and the CPU 1005 can then program or otherwise configure the CPU 1005 to implement the methods of the present disclosure. Examples of operations performed by the CPU 1005 can include fetching, decoding, executing, and writing back.

[0091] The CPU 1005 can be part of a circuit, such as an integrated circuit. One or more other components of the system 1001 can be included in the circuit. In some cases, the circuit is an application specific integrated circuit (ASIC).

[0092] The storage unit 1015 can store files, such as drivers, libraries, and saved programs. The storage unit 1015 can store user data, such as user preferences and user programs. In some cases, the computer system 1001 can include one or more additional data storage units external to the computer system 1001, such as located on a remote server that communicates with the computer system 1001 via an intranet or the Internet.

[0093] The computer system 1001 can communicate with one or more remote computer systems via the network 1030. For example, the computer system 1001 can communicate with a remote computer system of a user (e.g., a professional sonographer or an untrained technician). Examples of remote computer systems include personal computers (e.g., portable PCs), tablets or tablet PCs (e.g., iPad, Galaxy Tab), telephones, smart phones (e.g., iPhone, Android-enabled devices, ) or personal digital assistants. The user can access the computer system 1001 via the network 1030.

[0094] The methods described herein can be implemented by machine (e.g., computer processor) executable code stored in an electronic storage location of the computer system 1001, such as, for example, the memory 1010 or the electronic storage unit 1015. The machine executable code or machine readable code can be provided in the form of software. During use, the code can be executed by the processor 1005. In some cases, the code can be retrieved from the storage unit 1015 and stored on the memory 1010 for ready access by the processor 1005. In some cases, the electronic storage unit 1015 can be excluded and the machine executable instructions are stored on the memory 1010.

[0095] The code can be pre-compiled and configured to be used with a machine having a processor suitable for executing the code, or can be compiled during run-time. The code can be provided in a programming language that can be selected such that the code can be executed in a pre-compiled or compiled manner.

[0096] Aspects of the systems and methods provided herein, such as computer system 1001, may be embodied in programming. Aspects of the present technology may be regarded as a "product" or "article" typically in the form of machine (or processor) executable code and / or associated data carried or embodied on a type of machine-readable medium. The machine executable code may be stored on an electronic storage unit such as a memory (e.g., read only memory, random access memory, flash memory) or a hard disk. A "storage" type medium may include any or all of the tangible memories or their associated modules of a computer, processor, etc., such as various semiconductor memories, tape drives, disk drives, etc., which may provide non-transitory storage for software programming at any time. All or part of the software can sometimes be communicated via the Internet or various other telecommunications networks. Such communication can, for example, enable the loading of software from one computer or processor to another, such as from an administrative server or a host computer to the computer platform of an application server. Thus, another type of medium that can carry software elements includes optical, electrical, and electromagnetic waves such as used across physical interfaces between local devices via wired and optical landlines networks as well as in various air links. Physical elements that carry such waves, such as wired or wireless links, optical links, etc., may also be regarded as media that carry software. As used herein, unless restricted to non-transitory, tangible "storage" media, terms such as computer or machine "readable media" refer to any medium that participates in providing instructions to a processor for execution.

[0097] Thus, a machine-readable medium such as computer-executable code can take many forms, including but not limited to tangible storage media, carrier media, or physical transmission media. Non-volatile storage media includes, for example, optical discs or magnetic disks, such as any storage device in any computer, such as may be used to implement a database shown in the accompanying drawings. Volatile storage media includes dynamic memory, such as the main memory of such a computer platform. Tangible transmission media includes coaxial cables; copper wire and fiber optics, including the wires that make up a bus within a computer system. Carrier transmission media can take the form of electrical signals or electromagnetic signals, or acoustic or light waves, such as those generated during radio frequency (RF) and infrared (IR) data communications. Thus, common forms of computer-readable media include, for example: floppy disks, flexible disks, hard disks, magnetic tapes, any other magnetic media, CD-ROMs, DVDs or DVD-ROMs, any other optical media, punched cards, paper tapes, any other physical storage media with hole patterns, RAM, ROM, PROM, and EPROM, FLASH-EPROM, any other memory chip or cartridge, a carrier wave that transports data or instructions, a cable or link that transports such a carrier wave, or any other medium from which a computer can read programming code and / or data. Many of these forms of computer-readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution.

[0098] Computer system 1001 may include or communicate with an electronic display 1035 that includes a user interface (UI) 1040 for providing, for example, an indication to a user as to whether an ultrasound contrast agent is needed to improve the inherent image quality. Examples of UIs include but are not limited to graphical user interfaces (GUIs) and web-based user interfaces.

[0099] The methods and systems of the present disclosure may be implemented by one or more algorithms. The algorithms may be implemented by software when executed by a central processing unit 1005, and the algorithms may be configured, for example, to perform any of the methods described herein.

[0100] Example

[0101] The following illustrative examples represent aspects of the software applications, systems, and methods described herein and are not meant to be limiting in any way.

[0102] Example 1: Automatic contrast agent indication during an ultrasound imaging procedure

[0103] An ultrasound physician acquires ultrasound images of an obese patient according to the methods described herein and / or using the systems described herein. The ultrasound physician selects a diagnostic procedure suitable for determining the left ventricular function of the patient's heart. The ultrasound imaging system guides the user to correctly position the imaging probe of the system in a manner that is expected to acquire diagnostic-quality images. However, during image acquisition, the ultrasound imaging system determines that the inherent image quality is below the threshold required for diagnosis according to the selected diagnostic procedure using the machine learning models described herein (e.g., a neural network trained using images annotated regarding whether a contrast agent is indicated by a professional ultrasound physician). The system then determines whether a contrast agent is likely to improve the inherent quality based on the image and provides an indication of this determination to the user according to the methods described herein, for example, using any of the workflows described in Figures 1 to 5 . When a contrast agent is indicated (e.g., when an image similar to the example shown in Figure 6 is obtained), the user then administers the contrast agent to the obese patient and restarts the imaging procedure. The images collected subsequently (e.g., as shown in Figure 7 ) show improved inherent image quality for the same view relative to the previous images.

[0104] Although the preferred aspects of the invention have been shown and described herein, it will be apparent to those skilled in the art that these aspects are provided by way of example only. This is not meant to limit the invention to the specific examples provided in the specification. While the invention has been described with reference to the foregoing specification, the description and illustration of the aspects herein are not meant to be construed in a limiting sense. Many variations, changes, and substitutions will occur to those skilled in the art without departing from the invention. In addition, it should be understood that all aspects of the invention are not limited to the specific descriptions, configurations, or relative proportions set forth herein, which depend on a variety of conditions and variables. It should be understood that various alternatives of the aspects of the invention described herein may be employed in practicing the invention. Accordingly, it is contemplated that the invention should also cover any such alternatives, modifications, variations, or equivalents. The following claims are intended to define the scope of the invention, and the methods and structures within the scope of these claims, as well as their equivalents, are hereby covered.

Claims

1. A method for ultrasonic imaging, the method comprising: automatically determining whether the inherent image quality of a plurality of ultrasonic images is less than a required threshold inherent quality associated with a diagnostic procedure; and based on determining that the inherent image quality is below the threshold, outputting an indication that: (i) using an ultrasonic contrast agent may improve the inherent image quality; or (ii) it is not possible to improve the inherent image quality.

2. The method according to claim 1, wherein The indication is provided in real time during the diagnostic procedure.

3. The method according to claim 1, wherein, The automatic determination is performed using a trained machine learning model, and the trained machine learning model is trained using training data, the training data including a plurality of data points labeled to indicate that an expert determined that the ultrasonic contrast agent was needed.

4. The method according to claim 3, wherein The training data includes a plurality of data points captured using the ultrasonic contrast agent.

5. The method according to claim 1, wherein The required threshold is at least the minimum inherent quality required to successfully complete the diagnostic procedure.

6. The method according to claim 1, wherein The automatic determination includes determining, by a machine learning model, that one or more views of an organ associated with the diagnostic procedure have been captured at a probe position that will provide clinically acceptable image quality while satisfying the inherent quality threshold, and / or have been captured in a mode that will provide clinically acceptable image quality while satisfying the inherent quality threshold, while the inherent image quality remains below the required threshold.

7. The method according to claim 6, wherein, Using a machine learning model to perform the following: determining that one or more views of the organ associated with the diagnostic procedure have been captured at the probe position that will provide clinically acceptable image quality while satisfying the inherent quality threshold, and / or have been captured in a mode that will provide clinically acceptable image quality while satisfying the inherent quality threshold, the machine learning model being trained to determine the expected clinical quality of the acquired images based in part on the probe position and / or the acquisition mode.

8. The method according to claim 6, wherein, The one or more views include a plurality of views, and the inherent quality of the plurality of views is integrated together to determine that the inherent image quality of the plurality of ultrasonic images is less than the required threshold inherent quality associated with the diagnostic procedure.

9. The method according to claim 6, wherein The organ is the heart, and the plurality of acquired ultrasonic images are two-dimensional ultrasonic images.

10. The method according to claim 9, wherein, The diagnostic procedure includes cardiac function measurement, and the required threshold inherent quality associated with the diagnostic procedure is the minimum inherent quality required for performing the cardiac function measurement.

11. The method according to claim 10, wherein The cardiac function measurement includes measurement of ejection fraction and / or measurement of left ventricular function.

12. The method according to claim 10, wherein, The minimum inherent image quality includes the visibility of a minimum number of myocardial segments in the plurality of acquired ultrasonic images.

13. The method according to claim 9, wherein The process of determining the inherent image quality of the plurality of ultrasonic images includes analyzing, by the machine learning model, the individual image quality of a plurality of myocardial segments of an object's heart.

14. The method according to claim 13, wherein, The analysis is performed to determine the individual image quality of the plurality of myocardial segments of the object's heart across a plurality of views.

15. The method according to claim 9, wherein, The plurality of acquired ultrasonic images include Doppler ultrasonic images.

16. The method according to claim 15, wherein, The method further includes detecting the presence or absence of valvular pathology in the plurality of acquired ultrasonic images.

17. The method according to claim 16, wherein, Automatically determining that the inherent image quality of the plurality of ultrasound images is less than a required threshold associated with the diagnostic procedure includes: estimating an expected blood flow parameter; and comparing the expected blood flow parameter with a measured blood flow parameter obtained from the Doppler ultrasound image, and the indication to the user includes an indication of the gap between the expected parameter and the measured parameter to alert that the Doppler signal is impaired.

18. The method according to claim 17, wherein, Both the expected blood flow parameter and the measured blood flow parameter are velocities.

19. An ultrasound imaging system configured to perform a diagnostic procedure on an object, the system comprising: an ultrasound imaging probe for acquiring a plurality of ultrasound images of at least a portion of an organ of the object; a computing system; and a non-transitory computer-readable storage medium storing instructions that, when executed by a processor of the computing system, cause the ultrasound imaging system to perform the method according to any one of claims 1 to 18.

20. A non-transitory computer-readable medium storing instructions that, when executed by a processor of a computer, cause the computer to perform the method according to any one of claims 1 to 18.

Images