Switching between neural networks based on analysis of localization scan images

Abstract

A method for processing medical images is provided, the method comprising receiving a first image and a second image different from the first image, wherein the second image has the same subject as the first image. The method further comprises identifying a plurality of anatomical structures in the first image and defining a plurality of image segments in the second image based on the location of the anatomical structures identified in the first image. The method then applies processing routines associated with the first anatomical structures to the first image segments in the second image and processing routines associated with the second anatomical structures to the second image segments in the second image. An imaging system for implementing the method and a non-transitory computer-readable medium storing programs for processing medical images are also provided.

Description

Technical Field

[0001] This disclosure generally relates to systems and methods for processing images such as medical images using trained neural networks. Background Technology

[0002] Images are typically acquired through standard imaging methods such as computed tomography (CT) scans, resulting in image artifacts and noise embedded in the images. Therefore, these images are usually processed using denoising algorithms. These denoising algorithms are often related to specific anatomical structures and are designed and trained, in the case of learning algorithms, to promote specific image features.

[0003] Therefore, when processing CT scans, reconstruction filters are used to enhance certain image features, such as sharpness or soft tissue contrast. As an example, when reconstructing head images, the system can use filters designed to enhance soft tissue contrast in brain tissue, while different filters can be used to reconstruct body images. In contrast, filters used to reconstruct body images may result in sharper images.

[0004] Reconstructed images are often noisy and may contain various artifacts from the reconstruction process. Various denoising algorithms, as well as other algorithms, can then be applied to improve image quality. Therefore, in scenarios where algorithms are learned for denoising, such as convolutional neural networks (CNNs) used to process CT scans, different filters may have already been used during image reconstruction to enhance image features (such as sharpness) or suppress different types of noise. The goal of denoising should be to preserve the image features enhanced by the filters used in the reconstruction. Therefore, learning algorithms such as CNNs can be trained for specific types of anatomical structures and specific reconstruction filters.

[0005] Due to the typically limited network capacity and resulting runtime, when training CNNs to denoise low-dose CT images, as an example, the current practice is to train different networks for different anatomical regions and reconstruction filters designed for those regions. This is particularly advantageous when the filters are designed to produce very specific image features within a given anatomical structure.

[0006] However, a single scan, such as a head and neck scan, may cover multiple different anatomical regions and corresponding structures. If such an image is reconstructed using a head filter, the natural choice would be to use a network trained on the same filter, but such a network typically does not encounter any anatomical structures other than the head during training. Training a single network on a single filter using various anatomical structures often leads to suboptimal performance in other regions, especially when the filter is designed for a specific anatomical structure.

[0007] Furthermore, due to the limited available network capacity and the resulting runtime, training a single network on various anatomical structures reconstructed using the same filters is generally not feasible. Summary of the Invention

[0008] A system and method for processing medical images are described, wherein multiple anatomical structures appear in a single image, and different processing routines are applied to each such anatomical structure. Thus, for example, different machine learning methods can be used to process different parts of a single image containing multiple such anatomical structures.

[0009] In addition to applying different processing routines to different anatomical structures appearing in a single image, the system and method can further determine locations within such images for switching between such processing techniques. This determination can therefore be based on available localization scans independent of the primary medical image being processed. Thus, multiple anatomical structures can be defined in preliminary images (such as localization scans) using machine learning or classical techniques. Such definition can then be applied to a primary image, different from the localization scan, where the primary image is of the same subject.

[0010] Therefore, a method for processing medical images is provided, wherein the method receives a first image and receives a second image different from the first image, wherein the second image includes the same subject as the first image. The method then identifies multiple different anatomical structures in the first image, and then defines multiple image segments in the second image based on the positions of the different anatomical structures identified in the first image, such that a first image segment among the multiple image segments contains a first anatomical structure among the multiple different anatomical structures, and a second image segment among the multiple image segments contains a second anatomical structure among the multiple different anatomical structures.

[0011] The method then applies a first processing routine associated with the first anatomical structure to the first image segment to obtain a processed first image segment, and applies a second processing routine, different from the first processing routine, associated with the second anatomical structure to the second image segment to obtain a processed second image segment, and then outputs a processed second image including both the processed first image segment and the processed second image segment.

[0012] In some embodiments, the first or second image is selected from imaging modes including computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), single-photon emission computed tomography (SPECT), X-ray imaging including digital X-ray radiometry (DXR), or image-guided therapy (IGT) imaging with fluoroscopy sequences.

[0013] For example, the second image can be a CT image or a master image in a different imaging modality, while the first image can be a localization scan used for the master image. Therefore, in the case where the second image is a CT image, a lower radiation dose can be used to acquire the localization scan image than the second image.

[0014] In some embodiments, multiple distinct anatomical structures in the first image may be identified before receiving the second image, such that these identification results are available upon receiving the second image. In some embodiments, multiple distinct anatomical structures in the first image are identified upon receiving the second image.

[0015] In some embodiments, the first image segment does not include the second anatomical structure and the second image segment does not include the first anatomical structure. In some embodiments, the first and second image segments are parsed linearly such that the first image segment includes the entire width of the upper portion of the second image and the second image segment includes the entire width of the lower portion of the second image.

[0016] The first or second anatomical structure can be selected from the head, neck, upper body, abdomen, pelvic region, lower body, and legs.

[0017] In some embodiments, the first processing routine is a first machine learning algorithm associated with a first anatomical structure, and the second processing routine is a second machine learning algorithm associated with a second anatomical structure.

[0018] An imaging system is also provided, comprising a memory storing multiple instructions, an imaging unit, and a processor circuit connected to the memory and configured to execute instructions to acquire an image and implement the methods described above. In such a system, a second image can be received from the imaging unit.

[0019] A non-transitory computer-readable medium is also provided for storing a program for processing medical images, the program including instructions for implementing the methods described above. For example, such a method can be implemented in the context of the system described. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of a system according to an embodiment of the present disclosure.

[0021] Figure 2 An imaging apparatus according to an embodiment of the present disclosure is shown.

[0022] Figure 3A and Figure 3B The same image is shown being processed using two different processing routines.

[0023] Figure 4 This is a flowchart illustrating a method for processing medical images according to the present disclosure.

[0024] Figure 5 The analysis of the image according to this disclosure is shown. Detailed Implementation

[0025] The description of illustrative embodiments based on the principles of the present invention is intended to be consistent with the appendix. Figure 1 From this reading, the accompanying drawings will be considered an integral part of the entire written description. In the description of embodiments of the invention disclosed herein, any references to direction or orientation are merely for descriptive convenience and are not intended to limit the scope of the invention in any way. Relative terms such as “lower,” “upper,” “horizontal,” “vertical,” “above,” “below,” “up,” “top,” and “bottom,” and their derivatives (such as “horizontally,” “downward,” “upward,” etc.) should be understood to refer to the orientation shown in the figures being described or discussed. These relative terms are merely for descriptive convenience and do not require the device to be constructed or operated in a specific orientation unless explicitly indicated otherwise. Terms such as “attach,” “adhere,” “connect,” “join,” “interconnected,” and similar terms refer to a relationship in which structures are directly or indirectly fixed or attached to each other by an intermediate structure, and both movable and rigid attachments or relationships, unless explicitly described otherwise. Furthermore, the features and advantages of the invention are illustrated by reference to exemplary embodiments. Therefore, the present invention should not be limited to such exemplary embodiments that illustrate combinations of feasible non-limiting features that may exist alone or in combination with other features; the scope of the invention is defined by the appended claims.

[0026] This disclosure describes the best mode of carrying out the invention as currently contemplated. This description is not intended to be construed in a limiting sense, but rather provides examples of the invention for illustrative purposes only, to inform those skilled in the art of the advantages and structure of the invention by referring to the accompanying drawings. In the various views of the drawings, similar reference numerals denote similar or related parts.

[0027] It should be noted that the disclosed embodiments are merely examples of the many advantageous uses of the inventive teachings herein. Generally, the statements in this specification are not intended to limit anything disclosed in the various claims. Furthermore, some statements may apply to certain inventive features but not to others. Generally, unless otherwise indicated, singular elements may be plural, and vice versa, without loss of generality.

[0028] Typically, a single image is reconstructed using a single filter, so it's crucial to select a filter appropriate for the subject matter of the image being reconstructed. Therefore, when reconstructing a head image, the system might use a filter designed to enhance the contrast of soft tissue within the brain. However, when reconstructing a body image, different, potentially sharper filters could be used.

[0029] Therefore, in the context of medical imaging based on computed tomography (CT), different image processors, such as machine learning algorithms in the form of convolutional neural networks (CNNs), can be used to process images. Then, in the case of machine learning algorithms, these image processors are trained on corresponding different anatomical regions and structures in the context of reconstruction filters.

[0030] In medical imaging other than CT, such as magnetic resonance imaging (MRI) or positron emission tomography (PET), methods other than using this reconstruction filter can be used to recreate and process the image. Instead, different reconstruction algorithms can be used depending on the type of scan or data acquisition and the desired image features for the specific scan. The reconstruction algorithm can then be tuned to produce certain image features. For example, in common iterative MRI or PET reconstruction, the number of iterations can be carefully selected, and regularization terms can be added.

[0031] However, a single scan, such as a head and neck scan, can cover different anatomical regions. Therefore, a system and method are provided for efficiently processing different anatomical regions in a single image.

[0032] Figure 1 This is a schematic diagram of a system 100 according to an embodiment of the present disclosure. As shown, system 100 generally includes a processing device 110 and an imaging device 120.

[0033] Processing device 110 can apply processing routines to received images. Processing device 110 may include memory 113 and processor circuitry 111. Memory 113 may store multiple instructions. Processor circuitry 111 may be coupled to memory 113 and configured to execute instructions. Instructions stored in memory 113 may include processing routines and data related to various machine learning algorithms, such as various convolutional neural networks for processing images.

[0034] The processing device 110 may further include an input device 115 and an output device 117. The input device 115 may receive information, such as an image, from the imaging device 120. The output device 117 may output information to a user or a user interface device. The output device may include a monitor or display.

[0035] In some embodiments, the processing device 110 may be directly associated with the imaging device 120. In alternative embodiments, the processing device 110 may be different from the imaging device 120, such that it receives images for processing at the input device 115 via a network or other interface.

[0036] In some embodiments, the imaging device 120 may include an image data processing device and a spectral or conventional CT scanning unit for generating CT projection data when scanning an object (e.g., a patient).

[0037] Figure 2 An exemplary imaging apparatus according to one embodiment of this disclosure is shown. It should be understood that although a CT imaging apparatus is shown, and the following discussion is in the context of CT images, similar methods can be applied in the context of other imaging apparatuses, and images to which these methods can be applied can be acquired in a variety of ways.

[0038] In an imaging apparatus according to embodiments of the present disclosure, a CT scanning unit may be configured to perform multiple axial and / or helical scans of an object to generate CT projection data. In an imaging apparatus according to embodiments of the present disclosure, the CT scanning unit may include an energy-resolved photon-counting image detector. The CT scanning unit may include a radiation source that emits radiation for passing through the object when acquiring projection data.

[0039] Furthermore, in the imaging apparatus according to embodiments of the present disclosure, the CT scanning unit may perform a localization scan different from the main scan, thereby generating different images related to the localization scan and the main scan, wherein these images are different but include the same subject.

[0040] exist Figure 2 In the example shown, the CT scanning unit 200 (e.g., a computed tomography (CT) scanner) may include a fixed gantry 202 and a rotating gantry 204, which may be rotatably supported by the fixed gantry 202. During the acquisition of projection data, the rotating gantry 204 may rotate about a longitudinal axis and about an examination area 206 for the object. The CT scanning unit 200 may include a support 207 to support the patient in the examination area 206 and is configured to allow the patient to pass through the examination area during imaging.

[0041] The CT scanning unit 200 may include a radiation source 208, such as an X-ray tube, which may be supported by a rotating gantry 204 and configured to rotate with the rotating gantry 204. The radiation source 208 may include an anode and a cathode. A source voltage applied to the anode and cathode may accelerate electrons from the cathode to the anode. The electron flow may provide a current flow from the cathode to the anode to generate radiation for passing through the examination area 206.

[0042] The CT scanning unit 200 may include a detector 210. The detector 210 may be aligned with an angular arc across the examination area 206 relative to the radiation source 208. The detector 210 may include a one-dimensional or two-dimensional array of pixels (e.g., direct conversion detector pixels). The detector 210 may be adjusted to detect radiation passing through the examination area and to generate a signal indicating its energy.

[0043] The CT scanning unit 200 may further include generators 211 and 213. Generator 211 may generate tomographic projection data 209 based on signals from detector 210. Generator 213 may receive the tomographic projection data 209 and generate a raw image 311 of the object based on the tomographic projection data 209. The raw image 311 may be a localization scan image or a main scan image and may be input to the input device 115 of the processing device 110.

[0044] Figure 3A and Figure 3B The same image is shown being processed using two different processing routines that result in different processed images. The original image 311 is typically received at the input device 115 of the processing device 110. The image 311 is then processed according to the processing routines. The processing routines may include, for example, reconstructing the image 311 using filters implementing machine learning algorithms such as convolutional neural networks (CNNs). Such machine learning algorithms are typically trained on sample images similar to the image on which they will ultimately be applied. Therefore, the processing routines may be specific to a particular anatomical region or a corresponding reconstruction filter.

[0045] Therefore, processing routines can be applications of reconstruction filters designed for a specific anatomical region and corresponding CNNs trained on that specific anatomical region (e.g., head networks trained on head filters). Head filters can be specifically designed to enhance the contrast of soft tissue in brain tissue and then trained using head images, while body filters can be designed to provide sharper results and paired with a CNN trained on body images.

[0046] However, in practice, the scan images obtained using imaging device 120 may contain multiple anatomical regions. With this in mind, Figure 3A The image shown is an image acquired during a head and neck scan, reconstructed using a head filter, and subsequently denoised using a network trained on head anatomy using that head filter. In contrast, Figure 3B It shows the relationship with Figure 3A The same image shown is reconstructed using a body filter and then denoised using a network trained on body anatomy.

[0047] Figure 3A The processed image shown is essentially more than Figure 3B The images shown are noisier and of lower quality, but in fact, each image was denoised using a CNN trained on the corresponding filters. This is because the head network typically does not see any anatomical structures other than the head during training. A similar phenomenon occurs when body networks and filters are applied to scans of the head region.

[0048] Therefore, when given an image containing multiple anatomical structures, existing systems and methods must determine which of several potential processing routines should be applied. According to this disclosure, a method is proposed in which different image segments within a single image can be processed using different processing routines. Thus, different filters and CNNs trained using such filters can be applied to different image segments associated with the corresponding anatomical structures.

[0049] The method according to one embodiment of the invention further determines the optimal location for switching between such different processing routines during processing by evaluating a first image, such as a positioning scan image, containing the same subject as the main image being processed.

[0050] Figure 4 This is a flowchart illustrating a method for processing medical images according to the present disclosure. As shown, the method includes applying different networks trained on different anatomical structures to different parts of the image.

[0051] The method includes initially receiving a first image from the imaging device 120 at input device 115 of processing device 110 (at 400). The first image may be, for example, a localization scan performed using the same imaging device 120 as the subsequent master image. The received localization scan may be, for example, a planning image used to plan the subsequent master image, and in the case of a CT scan, it may be obtained using a lower radiation dose than the subsequent master image. It should be understood that images can be received in a wide variety of formats, including image formats directly output by imaging device 120 and partially processed data associated with such images. In addition, the first image received (at 400) and any other images discussed below may take different forms and may include different file types. Thus, as discussed below, the first image may be received in the form of raw data that can be processed to identify anatomical structures, regardless of whether such data can be fully reconstructed to create the corresponding first image.

[0052] The method then identifies multiple distinct anatomical structures (at 410) in the first image. These anatomical structures can be, for example, the head, neck, upper body, abdomen, pelvic region, lower body, and legs. The first image may contain two such anatomical structures, or it may contain several such structures, for example, in a whole-body scan scenario.

[0053] After receiving a first image (at 400), the method receives a second image (at 420) that is different from the first image. The second image may be the primary image, and although the second image is different from the first image, it typically includes the same subject as the first image. Therefore, when the first image obtained at 400 includes multiple anatomical structures identified at 410, the second image obtained at 420 typically includes the same multiple anatomical structures.

[0054] The first or second image can be obtained using various imaging modalities. Such an image can then be obtained by computed tomography (CT), for example by the imaging device 120 discussed above, or by magnetic resonance imaging (MRI), positron emission tomography (PET), single-photon emission computed tomography (SPECT), X-ray imaging including digital radiometry (DXR), or fluorescence fluoroscopy sequences in image-guided therapy (IGT) imaging. When using any of these imaging modalities, both the first and second images can be obtained using the same imaging modality, or the first image can be obtained using a different imaging modality than the second image.

[0055] For example, the first image may be a localization image from a CT scan and can be obtained using CT. However, the first image may be acquired using a lower radiation dose than the second image and can then be used as a planning image for a CT scan, which will then be performed using a higher radiation dose. In other embodiments, different imaging modalities may be used to obtain the localization image, making the localization scan faster, easier, less invasive, less costly, or otherwise more convenient. For example, the localization image may be obtained using an X-ray procedure, followed by a CT scan to obtain the main image. Similarly, a CT scan may be used as a localization image, with a subsequent PET or SPECT as the main image. In some such embodiments, in addition to using such CT images to define different anatomical structures in the image, CT scans may be used to improve the reconstruction of PET or SPECT. The first or second image may each be a three-dimensional image, or it may be a two-dimensional image. Furthermore, the first image may be captured at a lower resolution or with lower contrast or color density compared to the second image.

[0056] Additionally, in some embodiments, the identification of multiple distinct anatomical structures in the first image (at 410) occurs before the second image is received from the imaging device 120 (at 420). In such embodiments, the method can then define the anatomical structures before receiving the second image, thereby allowing for more efficient processing of the corresponding second image. Similarly, in some embodiments, multiple distinct anatomical structures in the first image are identified upon receiving the second image (at 420). These methods allow the system 100 implementing the method to improve system processing efficiency by completing all processing related to the identification of distinct anatomical structures in the first image before or simultaneously with receiving the second image. Thus, once the second image is received partially or completely, it can be processed immediately based on the identification results obtained from the first image.

[0057] In some embodiments, the processing of the first image does not need to be completed before processing the second image begins. Therefore, once the anatomical structure is identified in the first image and the corresponding portion of the second image has been received, the relevant portion of the second image can be processed according to subsequent steps.

[0058] The method then defines multiple image segments (at 430) in the second image based on the location of the different anatomical structures identified in the first image. Thus, if the first image is determined to contain a first anatomical structure (e.g., the head) and a second anatomical structure (e.g., the neck) at 410, the method defines a first image segment containing the first anatomical structure and a second image segment containing the second anatomical structure.

[0059] In embodiments where the anatomical structure is defined (at 410) before the second image is received (at 420), the image segment may be defined immediately after the second image is received (at 430). In some embodiments, where an image is received at the processing device 110 (at 420) as the image is acquired by the imaging device 120, such an anatomical structure may be identified and the corresponding image segment defined before or simultaneously with receiving the entire second image. For example, in the case of performing a whole-body scan after a localization scan, the image segment corresponding to the first anatomical structure to be scanned may be retrieved and processed before the scan for generating the second image is completed, as described below.

[0060] Once multiple image segments are defined (at 430), different processing routines are applied to each image segment. Thus, a first processing routine is applied to the first image segment (at 440), wherein the first processing routine is associated with a first anatomical structure contained in the first image segment.

[0061] Specifically, a second processing routine is applied to the second image segment (at 450), wherein the second processing routine is associated with a second anatomical structure contained in the second image segment. The first and second anatomical structures are different anatomical structures, and the processing routines applied to them are different from each other.

[0062] As described above, although the illustrated embodiments discuss the initial definition of multiple image segments (at 430) followed by the application of processing routines (at 440, 450), in some embodiments, one of the multiple image segments may be defined before the complete second image is received. In such a scenario, the first image segment may be defined (at 430), and processing may begin before the complete second image is received (at 440).

[0063] The processing routines are typically software routines designed to improve the quality or sharpness of the second image. Such routines can be, for example, denoising routines, or can otherwise remove artifacts from the imaging modality used. The processing routines can be machine learning algorithms, each of which can be trained on a corresponding anatomical structure. Therefore, the first processing routine applied at 440 could be a first machine learning algorithm, such as a CNN trained on the first anatomical structure. Then, the second processing routine applied at 450 could be a second machine learning algorithm, such as a CNN trained on the second anatomical structure.

[0064] The processing routine may further include reconstructing the relevant image segment using an appropriate filter designed for the corresponding anatomical structure. Thus, in the case where the first anatomical structure is the patient's head, the first processing routine applied at 440 could be the reconstruction of the corresponding first image segment using a head filter, followed by denoising using a corresponding head CNN trained on the head filter. Similarly, in the case where the second anatomical structure is the patient's neck or torso, the second processing routine applied at 450 could be the reconstruction of the corresponding second image segment using a body filter, followed by denoising using a CNN trained on the same body filter associated with the neck or torso.

[0065] In some embodiments, the same reconstruction filter may be used for multiple image segments, but different CNNs trained on the corresponding anatomical structures are applied to different image segments. It should be understood that although CNNs have been described, alternative algorithmic architectures, including various machine learning algorithms, may also be utilized.

[0066] After processing the defined first and second image segments (correspondingly at 440 and 450), the processed second image is output (at 460). The processed second image contains both the processed first and second image segments. If the first and second image segments are parsed into different images for processing purposes, they are typically recombined into a single image before outputting the aforementioned image.

[0067] In the illustrated embodiment, the method identifies multiple anatomical structures (at 410) in the first image. This identification can be achieved using processing routines, such as the application of machine learning algorithms. For example, anatomical structures can be identified by recognizing landmarks in the image associated with these structures. Alternatively, such anatomical structures can be manually identified before the image is resolved into corresponding image segments. This manual identification can be performed by a technician identifying and locating structures in the scan image on a user interface before, during, or after obtaining the primary scan image.

[0068] In some embodiments, the identification of anatomical structures can be performed in the same image as the image being processed. In such embodiments, instead of identifying anatomical structures in a first image and then processing a second image, a first analysis is performed on that image in which anatomical structures and corresponding image segments are identified, followed by a second analysis in which the image segments are processed differently.

[0069] Figure 5 The analysis of image 500 according to this disclosure is illustrated. As shown, image 500 includes the head 510, neck 520, and upper torso 530 of a scanned patient. Typically, the method will identify different anatomical structures associated with each of the head 510, neck 520, and upper torso 530, and then define a first image segment 540 associated with a first anatomical structure (such as head 510) and a second image segment 550 associated with a second anatomical structure (such as upper torso 530).

[0070] Although two image segments 540 and 550 are shown, a different third image segment may also be defined to cover the neck 520. Alternatively, in the illustrated embodiment, the second image segment 550 may be defined to include both the neck 520 and the upper torso 530, and both may be processed using a single processing routine suitable for both.

[0071] It should be noted that, for ease of understanding of the method according to an embodiment of the present invention, Figure 5 Examples of positioning scan images are provided. Although image segments 540 and 550 are shown on the illustrated positioning scan image 500, it should be understood that during use, such image segments will be defined in the main image received after the positioning scan image is received.

[0072] As shown in the figure, the first image segment 540 and the second image segment 550 can be linearly parsed such that the first image segment 540 includes the entire width of the upper part of the image 500, and the second image segment 550 includes the entire width of the lower part of the image.

[0073] Additionally, as shown in the figure, the first image segment 540 can be defined to exclude the second anatomical structure, namely the upper torso 530, and the second image segment 550 can be defined to exclude the first anatomical structure, namely the head 510. This method prevents individual portions of the image from being processed separately using the first processing routine and the second processing routine.

[0074] Alternatively, a portion of the image can be defined as a part of both the first image segment 540 and the second image segment 550. In such an embodiment, the results of the first and second processing routines can be merged before outputting the processed second image (at 460). Such merging can be, for example, by averaging the resulting overlapping image segments.

[0075] It should be understood that although the methods described herein are primarily in the context of CT scan images, various imaging techniques, including a wide range of medical imaging technologies, are envisioned, and images generated using these diverse imaging techniques can be effectively denoised or otherwise processed using the methods described herein.

[0076] The methods according to this disclosure can be implemented on a computer as a computer-implemented method, or in dedicated hardware, or in a combination of both. Executable code for the methods according to this disclosure can be stored on a computer program product. Examples of computer program products include memory devices, optical storage devices, integrated circuits, servers, online software, etc. Preferably, the computer program product may include non-transitory program code stored on a computer-readable medium for performing the methods according to this disclosure when the program product is executed on a computer. In embodiments, the computer program may include computer program code adapted to perform all steps of the methods according to this disclosure when the computer program is run on a computer. The computer program may be embodied on a computer-readable medium.

[0077] Although this disclosure has been described with respect to several described embodiments in a certain length and with a certain specificity, it is not intended to limit it to any such specific embodiment or any particular embodiment, but rather to be interpreted with reference to the appended claims so as to provide the broadest possible interpretation of such claims in consideration of the prior art, and thus effectively cover the intended scope of this disclosure.

[0078] All examples and conditional language described herein are intended for pedagogical purposes to aid the reader in understanding the principles of this disclosure and the ideas contributed by the inventors to advance the art, and should be interpreted as not being limited to these specifically described examples and conditions. Furthermore, all statements of principles, aspects, and embodiments of the disclosures herein, along with specific examples thereof, are intended to cover both structural and functional equivalents. Alternatively, such equivalents also include those currently known and those developed in the future, i.e., any elements developed to perform the same function, regardless of their structure.

Claims

1. A method for processing medical images, comprising: Receive the first image; Receive a second image that is different from the first image, wherein the second image includes the same subject as the first image; Several different anatomical structures were identified in the first image; Based on the location of the different anatomical structures identified in the first image, multiple image segments are defined in the second image, such that a first image segment among the multiple image segments contains a first anatomical structure among the multiple different anatomical structures, and a second image segment among the multiple image segments contains a second anatomical structure among the multiple different anatomical structures; A first processing routine associated with the first anatomical structure is applied to the first image segment to obtain a processed first image segment reconstructed by the first processing routine; A second processing routine, different from the first processing routine, associated with the second anatomical structure, is applied to the second image segment to obtain a processed second image segment reconstructed by the second processing routine; as well as Output a processed second image, which includes both the processed first image segment and the processed second image segment.

2. The method of claim 1, wherein, The first image or the second image is selected from an imaging mode including at least one of computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), single-photon emission computed tomography (SPECT), X-ray imaging including digital X-ray radiometry (DXR), and fluorescence fluoroscopy sequences in image-guided therapy (IGT) imaging.

3. The method of claim 1, wherein, The first image corresponds to a localization scan image obtained using a lower radiation dose than the second image.

4. The method of claim 1, wherein, The various anatomical structures in the first image were identified before the second image was received.

5. The method according to claim 1, wherein, The various anatomical structures in the first image are identified when the second image is received.

6. The method according to claim 1, wherein, The first image segment does not include the second anatomical structure, and the second image segment does not include the first anatomical structure.

7. The method according to claim 1, wherein, The first processing routine is a first machine learning algorithm associated with the first anatomical structure, and the second processing routine is a second machine learning algorithm associated with the second anatomical structure.

8. The method according to claim 1, wherein, The first image segment and the second image segment are parsed linearly such that the first image segment includes the entire width of the upper part of the second image, and the second image segment includes the entire width of the lower part of the second image.

9. The method according to claim 1, wherein, The first anatomical structure or the second anatomical structure is selected from the head, neck, upper body, abdomen, pelvic region, lower body, and legs.

10. An imaging system, comprising: A memory that stores multiple instructions; Imaging unit; as well as Processor circuitry, which is connected to the memory and configured to execute the instructions to: Obtain the first image; A second image, different from the first image, is obtained from the imaging unit, wherein the second image includes the same subject as the first image; Several different anatomical structures were identified in the first image; Based on the location of the different anatomical structures identified in the first image, multiple image segments are defined in the second image, such that a first image segment among the multiple image segments contains a first anatomical structure among the multiple different anatomical structures, and a second image segment among the multiple image segments contains a second anatomical structure among the multiple different anatomical structures; A first processing routine associated with the first anatomical structure is applied to the first image segment to obtain a processed first image segment reconstructed by the first processing routine; A second processing routine, different from the first processing routine, associated with the second anatomical structure, is applied to the second image segment to obtain a processed second image segment reconstructed by the second processing routine; as well as Output a processed second image, which includes the processed first image segment and the processed second image segment.

11. The system according to claim 10, wherein, The imaging unit implements an imaging modality including at least one of computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), single-photon emission computed tomography (SPECT), X-ray imaging including digital X-ray radiometry (DXR), and fluorescence fluoroscopy sequences in image-guided therapy (IGT) imaging.

12. The system according to claim 10, wherein, The first image corresponds to a localization scan image obtained using a lower radiation dose than the second image.

13. The system according to claim 10, wherein, The various anatomical structures in the first image were identified before the second image was received.

14. The system according to claim 10, wherein, The various anatomical structures in the first image are identified when the second image is received.

15. The system according to claim 10, wherein, The first image segment does not include the second anatomical structure, and the second image segment does not include the first anatomical structure.

16. The system according to claim 10, wherein, The first processing routine is a first machine learning algorithm associated with the first anatomical structure, and the second processing routine is a second machine learning algorithm associated with the second anatomical structure.

17. The system according to claim 10, wherein, The first image segment and the second image segment are parsed linearly such that the first image segment includes the entire width of the upper part of the second image, and the second image segment includes the entire width of the lower part of the second image.

18. The system according to claim 10, wherein, The first anatomical structure or the second anatomical structure is selected from the head, neck, upper body, abdomen, pelvic region, lower body, and legs.

19. A non-transitory computer-readable medium storing a program for processing medical images, the program comprising instructions to: Obtain the first image; Obtain a second image that is different from the first image, wherein the second image includes the same subject as the first image; Several different anatomical structures were identified in the first image; Based on the location of the different anatomical structures identified in the first image, multiple image segments are defined in the second image, such that a first image segment among the multiple image segments contains a first anatomical structure among the multiple different anatomical structures, and a second image segment among the multiple image segments contains a second anatomical structure among the multiple different anatomical structures; A first processing routine associated with the first anatomical structure is applied to the first image segment to obtain a processed first image segment reconstructed by the first processing routine; A second processing routine, different from the first processing routine, associated with the second anatomical structure, is applied to the second image segment to obtain a processed second image segment reconstructed by the second processing routine; as well as Output a processed second image, which includes the processed first image segment and the processed second image segment.

20. The non-transitory computer-readable medium according to claim 19, wherein, The instruction provides to identify the plurality of different anatomical structures in the first image before or simultaneously with receiving the second image.

Images