Systems and methods for organ evaluation

Abstract

A system can include an imaging device, and a computing device in communication with the imaging device. The computing device can be configured to receive, from the imaging device, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ, isolate the organ in each image of the plurality of images to generate a plurality of organ images, generate a plurality of markers overlaid on each organ image of the plurality of organ images, track a position of each marker of the plurality of markers across the plurality of organ images, determine an organ function of the organ, based on the tracking of the plurality of markers, and provide an action for the organ, based on the organ function.

Description

SYSTEMS AND METHODS FOR ORGAN EVALUATIONCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Patent Application No. 63 / 706,516 filed October 11, 2024, and entitled, “System and Method for Cardiac Performance Assessment Using Ex Vivo Video Kinematics,” which is hereby incorporated by reference in its entirety.STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH

[0002] N / A.BACKGROUND

[0003] Organ transplantation is needed for a variety of reasons, but most importantly, organ transplantation is lifesaving (i.e., without the organ transplant the individual would likely die). At this point, organs that are candidates for organ transplantation are rigorously analyzed with extremely selective criteria. While this decreases the likelihood of the failure of a transplanted organ, this further limits the already small pool of transplantable organs. Therefore, individuals that need organ transplants, regardless of reason, far exceed the organs that are available for transplantation. Thus, it would be desirable to have improved systems and methods for organ evaluation.SUMMARY OF THE DISCLOSURE

[0004] Some non-limiting examples of the disclosure provide a system. The system can include an imaging device, and a computing device in communication with the imaging device. The computing device can be configured to: receive, from the imaging device, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ, isolate the organ in each image of the plurality of images to generate a plurality of organ images, generate a plurality of markers overlaid on each organ image of the plurality of organ images, track a position of each marker of the plurality of markers across the plurality of organ images, determine an organ function of the organ, based on the tracking of the plurality of markers, and provide an action for the organ, based on the organ function.

[0005] In some non-limiting examples, an organ can be a heart, the plurality of organ images can be a plurality of heart images, and an organ function can be of the heart.

[0006] In some non-limiting examples, isolating an organ for each image of a plurality of images to generate a plurality of organ images can include a computing device being further configured to for each image of the plurality of images, isolate the organ from a background.

[0007] In some non-limiting examples, a computing device can be further configured to provide each image to a machine learning model that outputs a corresponding organ image, the machine learning model being trained to identify an organ from an inputted image.

[0008] In some non-limiting examples, a plurality of markers can include a grid of markers.

[0009] In some non-limiting examples, a computing device can be further configured to generate tracking information, based on a tracking of each marker of a plurality of markers across a plurality of organ images.

[0010] In some non-limiting examples, a computing device can be further configured to generate tracking information by providing each organ image of a plurality of organ images with a plurality of markers overlaid to a machine learning model that outputs the tracking information. The machine learning model can be trained to track the same marker of a plurality of markers across images.

[0011] In some non-limiting examples, tracking information can include sub tracking information that can define a movement of a given marker of a plurality of markers across organ images.

[0012] In some non-limiting examples, an organ function can include at least one of a contractility of an organ; or a compliance of the organ.

[0013] In some non-limiting examples, determining an organ function of the organ, based on a tracking of a plurality of markers can include the computing device being further configured to at least one of account for a relative size of the organ in a plurality of images; or account for a contraction rate of the organ in the plurality of images.

[0014] In some non-limiting examples, accounting for a relative size of an organ or accounting for a contraction rate of the organ includes a computing device being further configured to standardize tracking information from tracking a position of each marker across a plurality of images.

[0015] In some non-limiting examples, a computing device can be further configured to compensate a movement of a plurality of markers, based on background movement.

[0016] In some non-limiting examples, a computing device can be further configured to for each image of a plurality of images, isolate a background to generate a plurality of background images.

[0017] In some non-limiting examples, a plurality of markers is a first plurality of markers, and a computing device can be further configured to generate a second plurality of markers overlaid on each background image of a plurality of organ images, track the position of each marker of the second plurality of markers across the plurality of background images; and determine an organ function of the organ, based on the tracking of the second plurality of markers.

[0018] In some non-limiting examples, a computing device can be further configured to adjust a movement of a first plurality of markers across a plurality of organ images, based on a movement of a second plurality of markers across a plurality of background images.

[0019] In some non-limiting examples, adjusting a movement of a first plurality of markers includes a computing device being further configured to average a movement of a second plurality of markers across a plurality of background images, and subtract a movement of each marker of the first plurality of markers from the average.

[0020] In some non-limiting examples, a computing device can be further configured to notify a user of an imaging device, during acquisition of a plurality of images by the imaging device, to at least one of maintain a position of the imaging device relative to the organ being imaged, avoid changing an imaging angle or field of view (FOV) of the imaging device; or steading the imaging device.

[0021] In some non-limiting examples, providing an action for an organ includes a computing device being further configured to adjust a loading on a organ of a heart-lung machine coupled to the organ, adjust an operation of a perfusion device that is coupled to the organ, notify to a practitioner, that the organ is healthy, notify to a practitioner, that the organ is abnormal, notify to a practitioner, that the organ is cleared to be transplanted; or notify to a practitioner, that the organ is not recommended to be transplanted.

[0022] In some non-limiting examples, a computing device can be further configured to compare an organ function of an organ to a threshold value, and based on the organ function being above the threshold, provide an action for the organ.

[0023] In some non-limiting examples, the organ function can be a contractility and a threshold value can include a threshold contractility. In some non-limiting examples, an organ function can be a compliance and a threshold value can include a threshold compliance.

[0024] In some non-limiting examples, a compliance of an organ can correspond to a distance traveled of a plurality of markers across a plurality of images. In some non-limiting examples, a contractility of an organ corresponds to a peak velocity of a plurality of markers across a plurality of images.

[0025] In some non-limiting examples, a plurality of images can be acquired while an organ is in an unloaded state.

[0026] In some non-limiting examples, a plurality of images can be acquired while an organ is in a partially loaded state or a fully loaded state.

[0027] In some non-limiting examples, a plurality of images can be synchronized.

[0028] In some non-limiting examples, each image of a plurality of images can be an ex vivo image of an organ.

[0029] In some non-limiting examples, each image of a plurality of images can be acquired during an open surgery of an organ.

[0030] In some non-limiting examples, a plurality of images can be a first plurality of images and a computing device can be further configured to receive a second plurality of images of an organ of the subject. In some non-limiting examples, determining an organ function can be based on the second plurality of images.

[0031] Some non-limiting examples of the disclosure provide a system. The system can include an imaging device and a computing device in communication with the imaging device. The computing device can be configured to receive, from the imaging device, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ, generate a plurality of markers overlaid on the organ of each image of the plurality of images, track the position of each marker of the plurality of markers across the plurality of images, determine a function of a portion of the portion of the organ, based on the tracking of the plurality of markers; and provide an action for the portion of the organ, based on the function.

[0032] In some non-limiting examples, a computing device can be further configured to isolate a portion of an organ for each image of a plurality of images to generate a plurality oforgan portion images. A plurality of markers can be overlaid on a portion of an organ. Determining a function of a portion of an organ can be based on a tracking of a plurality of organs of the portion of the organ.

[0033] In some non-limiting examples, generating a plurality of markers overlaid on an organ can include a computing device being further configured to generate the plurality of markers on a portion of the organ in each image of a plurality of images.

[0034] In some non-limiting examples, an organ can be a heart. A portion of the organ can be a left ventricle or a right ventricle.

[0035] In some non-limiting examples, a function can be contractility. A computing device can be further configured to determine a contractility value for each marker of a plurality of markers, and based on at least one contractility value for a marker of the plurality of markers exceeding a contractility threshold, determine a function of a portion of the organ.

[0036] In some non-limiting examples, a function can be contractility. A computing device can be further configured to determine a contractility value for each marker of a plurality of markers, and identify each marker of the plurality of markers having a contractility value being below a contractility threshold, and provide an action based on the identified markers having a contractility value below the contractility threshold.

[0037] In some non-limiting examples, provide an action can include a computing device being further configured to present, on a display, an image of an organ with identified markers.

[0038] In some non-limiting examples, identified markers can be distinguished from a plurality of markers that were not identified.

[0039] In some non-limiting examples, a function associated with an identified marker can be at least one of indicative of a previously damaged region of an organ that includes an identified marker, indicative of an infarction of the region of the organ that includes the identified marker, or indicative of ischemia of the region of the organ that surrounds the identified marker.

[0040] Some non-limiting examples of the disclosure provide a system. The system can include an imaging device and a computing device a computing device in communication with the imaging device. The computing device can be configured t receive, from the imaging device, a plurality of optical images of an organ of a subject, generate one or more markers overlaid on the organ of each image of the plurality of images, track the position of each markerof the one or more markers across the plurality of images, determine a function of the organ, based on the tracking of the one or more markers, and provide an action for the organ, based on the function.

[0041] In some non-limiting examples, a function of a portion of an organ can include a health score of the organ.

[0042] In some non-limiting examples, a computing device can be further configured to determine an action, based on a health score.

[0043] In some non-limiting examples, a computing device can be further configured to determine an action, based on a health score exceeding a threshold health score, or determine the action, based on the health score being below a threshold health score.

[0044] Some non-limiting examples of the disclosure provide a computer implemented method for evaluating organ health. The method can include receiving, using one or more computing devices, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ, generating, using the one or more computing devices, one or more markers overlaid on the organ of each image of the plurality of images, tracking, using the one or more computing devices, the position of each marker of the one or more markers across the plurality of images, determining, using the one or more computing devices, a function of at least a portion of the organ, based on the tracking of the one or more markers, and providing, using the one or more computing devices, an action for at least the portion of the organ, based on the function.

[0045] In some non-limiting examples, a method can include determining, using one or more computing devices, a contractility value of at least a portion of an organ, based on tracking of one or more markers; and determining, using the one or more computing devices, a function based on a contractility value exceeding or being below a contractility threshold.

[0046] In some non-limiting examples, a method can include determining, using one or more computing devices, a compliance value of at least a portion of an organ, based on tracking of one or more markers, and determining, using the one or more computing devices, a function based on a compliance value exceeding or being below a compliance threshold.

[0047] Some non-limiting examples of the disclosure provide a method of assessing cardiac performance using ex vivo video kinematics in accordance with any non-limiting example disclosed herein, alone or in combination with any other method.

[0048] Some non-limiting examples of the disclosure provide a system for assessing cardiac performance using ex vivo video kinematics in accordance with any non-limiting example disclosed herein, alone or in combination with any other system.

[0049] The foregoing and other aspects and advantages of the present disclosure will appear from the following description. In the description, reference is made to the accompanying drawings that form a part hereof, and in which there is shown by way of illustration one or more exemplary versions. These versions do not necessarily represent the full scope of the disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The following drawings are provided to help illustrate various features of nonlimiting examples of the disclosure, and are not intended to limit the scope of the disclosure or exclude alternative implementations.

[0051] FIG. 1 shows a system for evaluating an organ.

[0052] FIG. 2 shows an example of a system, which can be implemented in a similar way as the system of FIG.1.

[0053] FIG. 3 shows a schematic illustration of specific block diagrams of computing devices, and a server.

[0054] FIG. 4 shows an example of a machine learning model.

[0055] FIG 5A shows another example of a machine learning model.

[0056] FIG 5B shows another example of a machine learning model.

[0057] FIG. 6 shows an example of generating markers on an image and a machine learning model.

[0058] FIG. 7 shows an example of a graph of coordinates of the path taken by a marker across the organ images.

[0059] FIG. 8 shows a flowchart of a process for evaluating an organ (e.g., the entire organ, an anatomical feature of an organ, a sub anatomical feature of an anatomical feature, etc.).

[0060] FIG. 9 shows a schematic of a software pipeline in accordance with an non-limiting example of the invention.

[0061] FIGS. 10A-10F show video frame snapshots of a beating heart with overlaid pixels for tracking motion.

[0062] FIGS. 11 A-F show graphs of the left ventricular contractility versus ex vivo perfusion time and ventricular compliance for six hearts.

[0063] FIGS. 12A-F show graphs of the right ventricular contractility versus ex vivo perfusion time and ventricular compliance for six hearts.

[0064] FIG. 13 shows software developed for automated, non-invasive quantification of ex-vivo beating heart function using video kinematics.

[0065] FIG. 14 shows two graphs, a left hand graph of different hearts’ average distance traveled at the donor center compared to at the recipient center, and a right hand graph of different hearts’ median peak velocity at the donor center compared to at the recipient center.

[0066] FIG. 15 shows two graphs, a left hand graph of the average distance traveled of hearts that did not load versus hearts that tolerated loading, and a right hand graph of the mean peak velocity of hearts that did not load versus hearts that tolerated loading.

[0067] FIG. 16 shows an image of the setup including the camera, a tripod to support the camera, and the heart. FIG. 16 also shows the heart along with an output of one of the trackers.

[0068] FIG. 17A shows a graph of the path traveled for a marker and FIG. 17B shows a second graph of the y position traveled for the marker over time.

[0069] FIG. 18A shows a graph of the contractility of five virtual markers for two hearts, and FIG. 18B shows the compliance of five virtual markers for the two hearts.

[0070] FIG. 19 shows a graph of the true positive rate versus the false positive rate for average distance traveled to predict the likelihood of tolerating loading.

[0071] FIG. 20 shows a graph of the true positive rate versus the false positive rate for mean peak velocity to predict the likelihood of tolerating loading.

[0072] FIG. 21A-C shows video kinematics quantified from five different videos of the same heart taken at different angles to produce consistent metrics. Specifically FIG. 21A shows images of the five different angles of that video was taken at, FIG. 2 IB shows a graph of the average distance traveled versus the video angle, and FIG. 21C shows a graph oof the mean peak velocity versus the video angle.

[0073] FIG. 22A shows an illustration of the experimental process investigating the use of video kinematics during a porcine study with ESHP and FIG. 22B shows a human proof-of- concept study during current clinical workflow of DCD heart transplantation.

[0074] FIG. 23A-C show the ability of the techniques herein to separate virtual markers by different regions and calculate regional specific metrics such as peak velocity and distance traveled for each marker within the region. Specifically, FIG.23A shows an image of a heart with left markers and right markers. FIG. 23B shows a graph of the peak velocity for the left and right sides. FIG. 23 C shows a graph of the distance traveled for the left and right sides.DETAILED DESCRIPTION OF THE PRESENT DISCLOSURE

[0075] Organ transplantation is needed for a variety of reasons, but most importantly, organ transplantation is lifesaving (i.e., without the organ transplant the individual would likely die). At this point, organs that are candidates for organ transplantation are rigorously analyzed with extremely selective criteria. While this decreases the likelihood of the failure of a transplanted organ, this further limits the already small pool of transplantable organs. Therefore, individuals that need organ transplants, regardless of reason, far exceed the organs that are available for transplantation.

[0076] Over the last decade, there has been increasing adoption of a new technique for organ transplantation termed donation after circulatory death (“DCD”). In this method, the donor has suffered injury that is deemed too severe to ever achieve meaningful recovery, and the individual’s family decides that the individual would want to be an organ donor. Although the individual does not technically meet all criteria necessary for legal brain death, this new category provides an avenue for organ donation. The individual is brought into the operating room and life sustaining artificial support is removed (e.g., mechanical ventilation and medications). If the individual expires, the individual’ s organs are procured for transplantation according to their wishes. Unlike brain death donation, however, the organs experience the effects of true circulatory death during this process, which risks damage to the organs — especially the heart.

[0077] Heart transplantation in this DCD methodology has shown acceptable survival rates comparable to brain death donation, but immediately after the transplantation these DCD hearts more frequently require other forms of support including short-term support, such as mechanical circulatory support (“MCS”), extracorporeal membrane oxygenation (“ECMO”), etc. Most transplant recipient patients recover after a few days, but this process inherently increases risk of morbidity to the recipient. Better organ quantification of organs(e.g., at the hospital), especially hearts in this case, could lead to better transplant patient outcomes since hearts at an increased risk of poor performance could be flagged much earlier to either prevent future damage or provide the needed assistance (e.g., MCS). Further, many potential organ candidates from marginal donors (e.g., older age, high risk mechanism of death, unknown cardiac history, etc.) that could otherwise perform adequately if transplanted currently are not attempted because current methodologies cannot confirm that these organs function adequately to warrant a transplant. In other words, the highly selective criteria rejects these potential organs for transplantation, since no other methodologies can prove otherwise (e.g., that the rejected candidate is actually suitable for transplantation).

[0078] Having an accurate and reproducible method of quantifying the function of organs including DCD hearts before implantation (e.g., when they are undergoing ex vivo reperfusion) can improve outcomes for current transplant recipients and save many more lives, as the pool of potential organ donors can be significantly increased.

[0079] Some non-limiting examples of the disclosure provide advantages to these issues (and others) by providing improved systems and methods for organ evaluation. Specifically, some non-limiting examples of the disclosure provide a system for evaluating a function of an organ to be transplanted (e.g., prior to transplantation). This system can include an imaging device and a computing device in communication with the imaging device. The imaging device can acquire a plurality of images of the organ, which are used to determine an organ function of the organ (e.g., by the computing device). These plurality of images can be acquired by light reflected off the organ, such as ambient light or light from the imaging device (e.g., each image can be an optical image, and therefore each image may not be an acoustic image, an attenuation image, etc.). In this way, the organ can be imaged while the organ is visible through a cavity, such as an open cavity (e.g., during an open surgery, such as an open heart surgery). In other cases, such as after the organ has been taken out of the donor, the organ can be coupled to a perfusion device to, for example, perform ex vivo perfusion during travel of the organ to another site, a room, or to prolong the life of the organ until the organ can be transplanted into the transplant recipient (e.g., while the transplant recipient is being prepped in the operating room). In this way, the organ can be imaged (e.g., the plurality of images can be acquired) while the organ is coupled to the ex vivo perfusion device to evaluate an organ function during perfusion of the organ, and each image of the plurality of images canbe an ex vivo image. Accordingly, the organ, connected to the perfusion device, can be continuously evaluated, indicating the health of the organ, and whether the organ is actually suitable for transplantation.

[0080] In some cases, the computing device can analyze the plurality of images to determine an action to be taken for the organ. For example, this can include the computing device, for each image, isolating the organ from the background to generate organ images and background images. In some cases, the computing device can segment the organ from the background, by for example, providing the image to a machine learning model trained to identify the organ. The computing device can generate a grid of markers on each organ image and each background image. The computing device can then track the position of each marker across the organ images, and separately, across the background images. For example, the computing device can provide each marker overlaid image (e.g., the marker organ images and the marker background images) to a machine learning model trained to detect movement of markers across images. This machine learning model can then, output tracking information for the organ (e.g., from the marker organ images) and the background (e.g., from the marker background images). The computing device can then use this tracking information from the organ to determine an organ function of the organ (e.g., contractility of the organ or compliance of the organ). This organ function can advantageously be a quantifiable value. That is, for example, contractility values and compliance values are quantitative indicators of organ function, rather than, say a qualitative view-based evaluation. Then, the computing device can provide an action for the organ, based on the organ function (e.g., with the computing device first determining whether the organ function exceeds a threshold value, is below a threshold value, etc.). The action for the organ can be, for example, notifying a practitioner about the organ (e.g., presenting a notification on a display, transmitting a notification or other corresponding message to a practitioner, etc.), such as for example, with the notification indicating the organ is healthy, abnormal, cleared to be transplanted, not recommended to be transplanted, etc., controlling a heart-lung machine for the organ (e g., adjusting a loading on an organ), controlling an operation of a perfusion device that is coupled to the organ, etc.

[0081] Further, some non-limiting examples of this disclosure can include evaluation of an organ while the organ is in an unloaded state (e.g., an ex vivo organ, a heart that is not pumping blood, etc ). These processes herein can accurately predict future performance (e.g.loading performance) of an unloaded organ in a non-invasive manner. For example, in theory, an organ could be tested in a loaded state, but that would likely damage the organ — particularly one that is already prone to be stressed easily.

[0082] FIG. 1 shows a system 100 for evaluating an organ. The system 100 can include imaging devices 102,104, 106, and a computing device 108 in communication with the imaging devices 102, 104, 106. Each imaging device 102, 104, 106 can include an imaging sensor (or more than one image sensor), such as a charge-coupled device (“CCD”) an active pixel sensor (e.g., a complementary metal-oxide-semiconductor “CMOS”), etc., to acquire one or more images of a field of view (FOV) for the respective imaging device 102, 104, 106. For example, each imaging device 102, 104, 106 can define a FOV (e.g., that defines an optical axis of the respective imaging device, which can be an axis perpendicular to the image sensor). More specifically, the imaging device 102 can define a FOV 110, the imaging device 104 can define a FOV 112, and the imaging device 106 can define a FOV 112. As shown in FIG. 1, each FOV 110, 112, 114 can be different, and at least a portion of each FOV 110, 112, 114 can overlap with a portion of a different FOV 110, 112, 114. For example, the FOV 110 can overlap with the FOV 112, the FOV 112 can overlap with the 114, etc. In some cases, all the FOVs 110, 112, 114 can at least partially overlap. In some cases, each imaging device 102, 104, 106 can define an imaging angle, which can bisect the corresponding FOV. For example, the imaging device 102 can define an imaging angle 116 that can bisect the FOV 110, the imaging device 104 can define an imaging angle 118 that can bisect the FOV 112, and the imaging device 106 can define an imaging angle 120 that can bisect the FOV 114.

[0083] As shown in FIG. 1, each imaging angle 116, 118, 120 can be different, such that each imaging angle 116, 118, 120 is angled different relative to the organ 122. In this way, the image(s) acquired at different imaging angles can advantageously acquire different perspectives of the organ 122. For example, portions of organ 122 may be obscured while imaging at the imaging angle 116, but otherwise captured by imaging at the imaging angle 120. In some cases, at least one of the imaging angles 116, 118, 120, can be substantially perpendicular (or perpendicular) to the organ 122. More specifically, at least one of the imaging angles 116, 118, 120 can extend above and below the organ 122 (e.g., with the corresponding FOV being a top view of the organ 122). In some cases, an imaging angle (e.g., one or more of the imaging angles 116, 118, 120) can intersect with a contraction axis of theorgan (e.g., the axis that intersects the heart when the organ 122 is a heart). Therefore, the imaging angle can be substantially parallel to the contraction axis of the organ 122.

[0084] Each imaging device 102, 104, 106 can acquire one or more images of the organ 122 (e.g., a plurality of images). For example, each FOV 110, 112, 114 can at least partially overlap with the organ 122, such that the images acquired therefrom include at least a portion (or the entire) organ 122 within the image. In some cases, the organ 122 can be visible during an open surgery (e.g., and exposed to the ambient environment), such that the ambient light can reflect off the organ 122 and be sensed by the imaging devices 102, 104, 106 (e.g., by an imaging senor). Further, in some cases, each imaging device 102, 104, 106 can include a light source to emit light to illuminate the organ 122. In this way, the light from an imaging device can reflect off the organ 122 and can be sensed by each imaging device 102, 104, 106 while acquiring the one or more images therefrom. In some cases, the light source can be separate from the imaging device 102, 104, 106 (e.g., in the operating room where the organ 122 is located). Regardless, therefore, the one or more images can be acquired by light reflecting off the organ (e.g., and then sensed by the respective imaging device). In some cases, the organ 122 can be coupled to an ex vivo perfusion device and therefore the one or more images can be ex vivo images.

[0085] The one or more images from each imaging device 102, 104, 106 can be a plurality of images. Therefore, the imaging device 102 can acquire a plurality of images of the organ 122, the imaging device 104 can acquire a plurality of images of the organ 122, the imaging device 106 can acquire a plurality of images of the organ 122. Each plurality of images can be a time series of images and can be acquired at a specific frequency, such as in frames per second (fps). Therefore, each image in the plurality can be time-stamped or otherwise associated with an acquisition time. This can be useful, for example, to determine a heart rate of a heart using the images, which as described below, can be used to standardize a heart function. In some non-limiting examples, the plurality of images can be acquired at a frequency where the frequency is less than or equal to 240 fps, 60 fps, 30 fps, etc. In some specific cases, the frequency can advantageously be less than 240 fps (e.g., “slow motion” acquisition), where the processes herein can determine an organ function without being in slow motion acquisition or even increased fps. This significantly expands the imaging devices capable of determining organ function using the processes herein, such as, for example,commercially available cell phones, rather than expensive, high frame per second cameras. In some cases, each plurality of images can be a video stream.

[0086] As shown in FIG. 1, the computing device 108 can be in communication with each imaging device 102, 104, 106, to for example receive the one or more images acquired. The computing device 108 can implement some or all of the processes herein. For example, as described below, some imaging devices lack computing resources necessary for high computation tasks, such as implementing machine learning models. Accordingly, the computing device 108 can receive the one or more images and can implement the computational analysis (e.g., where the computing device 108 includes the necessary computational components). For example, each imaging device 102, 104, 106 can transmit the respective one or more images to the computing device 108, via wired or wireless communication. In other cases, the computing device 108 can be part of one of the imaging devices 102, 104, 106. In this case, the other imaging devices can each include a computing device, which can be similar to the computing device 108.

[0087] Each imaging device 102, 104, 106 can include one or more optical components, such as lenses, waveguides, etc., to facilitate image acquisition. Although FIG. 1 shows three imaging device 102, 104, 106, in other implementations, the system 100 can include one imaging device (e.g., the imaging device 102, a single imaging device). In this case, the imaging device can acquire a number of sets of a plurality of images, with each set of a plurality of images taken at a different imaging angle (e.g., the imaging angles 116, 118, 120). In this way, the organ evaluation can be advantageously accomplished with less components (e.g., less imaging devices). In some cases, it can be helpful to acquire the plurality of images with a relatively constant imaging angle and FOV. This can minimize any noise introduced, where movement is captured other than the organ 122 to be imaged. Therefore, in some cases, the computing device 108 can transmit a message to an imaging device instructing the user to keep the imaging device still during image acquisition. In other cases, the imaging device itself can provide a notification to the user to keep the imaging device still during image acquisition. Regardless, these notifications can be displayed prior to image acquisition. In some cases, each imaging device 102, 104, 106 can include a support structure that substantially maintains the respective imaging angle during image acquisition. For example, each imaging device 102, 104, 106 can include a stabilizer. The stabilizer can be a tripod, a monopod, etc.

[0088] In other configurations, the computing device 108 can reject the plurality of images (e.g., including prompting the user, via a notification to re-acquire the images, etc ), based on a determined movement of the imaging device. For example, the computing device 108 can receive sensor data from a sensor (e.g., a gyroscope, an accelerometer, etc.) of the imaging device and can determine that the images are to be rejected based on the sensor data. Specifically, the computing device can determine a position of the imaging device (e.g., using the accelerometer) and based on the position deviating greater than a threshold amount, the computing device can reject the images and prompt the user to re-acquire the images. Similarly, the computing device can determine that the images are to be analyzed, based on the sensed position of the imaging device being below a threshold amount. In other cases, the computing device can reject the images, based on the tracking information of the one or more markers. For example, if the total movement of the one or more markers (e.g., averaged) exceeds a threshold value, the computing device can reject the images (and then re-prompt the user to re-acquire the images), since large total movements of the markers can indicate undesirable movement of the imaging device. Similarly, if the total movement of the one or more markers is below a threshold value, the computing device can accept the images, since smaller total movements of the tackers can indicate a lack of undesirable movement of the imaging device.

[0089] In some non-limiting examples, the entire system 100 including the organ 122 can be positioned within a room (e.g., an operating room) or can be positioned within a container (e.g., that receives the organ 122, such as an perfusion device, an ex vivo perfusion device). In other cases, some components of the system 100 can be separate from each other. For example, the computing device 108 can be remote (e.g., with an intermediary computing device relaying the data), such as a server (e.g., a remote server), the computing device 108 can be positioned outside of the container. Still further, the imaging devices 102, 104, 106 can be positioned outside of the container, but can acquire images of the organ 122 positioned within the container.

[0090] FIG. 2 shows an example of a system 150, which can be implemented in a similar way as the system 100. For example, the system 150 can include some or all of the components of the system 100 and vice versa. Accordingly, the description of the system 100 applies to the description of the system 150 and vice versa. The system 150 can include imaging devices152, 154, 156, a computing device 158, a perfusion device 160, and a sensor 162. The imaging devices 152, 154, 156 can be implemented in a similar manner as the imaging devices 102, 104, 106. For example, the imaging device 152 defines a FOV 164 and an imaging angle 166 to acquire a plurality of images of the organ 180, the imaging device 154 defines a FOV 168 and an imaging angle 170 to acquire a plurality of images of the organ 180, the imaging device 156 defines a FOV 172 and an imaging angle 174 to acquire a plurality of images of the organ 180. Similarly, the computing device 158 can be implemented in a similar manner as the computing device 108.

[0091] As shown in FIG. 2, the perfusion device 160 can be coupled to the organ 180 and can deliver a perfusate to the organ 180, and receive a perfusate from the organ 180. For example, the perfusion device 160 can deliver a perfusate 176 (e.g., an unused perfusate, a perfusate loaded with nutrients including oxygen, etc.) to the organ 122 and can receive a perfusate 178 after the perfusate 176 as delivered nutrients to the organ 122 (e.g., the perfusate176 delivering oxygen, glucose, etc., to the organ 122). In some cases, therefore, the perfusate178 can be used. The perfusion device 160 can replenish the perfusate 178 to be directed back to the organ 180. Accordingly, the perfusion device 160 can recirculate perfusate to keep the organ 180 alive, In some cases, the perfusate can be a non-biological liquid or can be a biological liquid. For example, the perfusate can be blood.

[0092] The perfusion device 160 can include a perfusate reservoir 182, a nutrient reservoir 184, a pump 186, a cooler 188, a heater 190, an oxygenator 192, and a computing device 194.The pump 186 can direct the perfusate 176 to the organ 180 and the perfusate 178 can be directed back to the perfusate reservoir 182. The nutrient reservoir 184 can include nutrients (e.g., glucose) that can be introduced (e.g., by a pump) into the perfusate reservoir 182 (and thus added to the perfusate). The cooler 188 can cool the perfusate after being received (e.g. the perfusate 178) so as to maintain a constant cool temperature of the organ 180. The cooler 188 can be a refrigeration-based cooler, a heat exchanger, a thermoelectric heat pump, passive heat diffusing devices (e.g., fins, such as coupled to a heat sink), a fan, etc. In some cases, the cooler 188 can be thermally coupled to the perfusate reservoir. In some cases, the cooler 188 can cool the organ 180 to prolong the life of the organ 180, such as during travel (e.g., a helicopter flight). The heater 190 can heat the perfusate to maintain a specific temperature of the organ 180 (e.g., 37°C, body temperature, etc ). The heater 190 can be implemented indifferent ways. For example, the heater 190 can be an electrical heater (e.g., a resistive heater), a light-based heater (e.g., an infrared heater), etc. In some cases, the heater 190 can be thermally coupled to the perfusate reservoir. The heater 190 can heat the organ 180 to maintain the life of the organ 180, for example, just prior to transplantation in a recipient, such as when the organ 180 is in the operating room, hospital, etc. (e.g., because the organ 180 should be at or near body temperature when actually transplanted).

[0093] The oxygenator 192 can add oxygen to the perfusate prior to the perfusate 176 being delivered to the organ 180. In this way, oxygen can be continuously delivered to the organ 180, via the perfusate. In some cases, the oxygenator 192 can be positioned downstream of the perfusate reservoir 182 and upstream of the organ 180. The oxygenator 192 can be implemented in different ways. For example, the oxygenator 192 can be a membrane oxygenator (e.g.,. including a source of oxygen). In some cases, the oxygenator 192 (or a different component of the perfusion device 160) can remove carbon dioxide from the perfusate 178 (or the perfusate 176). The computing device 194 can be implemented in a similar manner as the computing device 158, and the computing device 196 can control the functioning of the perfusion device 160. For example, the computing device 196 can cause the pump 186 to direct the perfusate 176 from the perfusate reservoir 182 and to the organ 180 and can similarly adjust a flow rate of the perfusate 176 (e.g., by increasing or decreasing the speed of the pump 186). As another example, the computing device 196 can cause the cooler 188 to increase or decrease cooling of the perfusate (e.g., in the perfusate reservoir 182), the computing device 196 can cause the heater 190 to increase or decrease heating of the perfusate.

[0094] In some cases, the perfusion device 160 can perform machine perfusion on the organ 180, the perfusion device 160 can be an ex vivo perfusion device to perform ex vivo perfusion. In other cases, the perfusion device 160 can be an in vivo perfusion device, such that the organ 180 is in a patient during perfusion of the organ 180 with the perfusate. For example, the perfusion device 160 can be a heart-lung machine or a cardiopulmonary bypass (“CPB”) machine. In this case, the perfusate 176, 178 can be blood, in which the perfusate 176 (e.g., oxygenated) is delivered to the aorta of the organ 180 that is a heart, and the perfusate 178 is received from the vena cava of the organ 180 that is a heart. Prior to perfusion, the major blood vessels of the heart (including the cardiac artery) can be clamped so that blood is refrained from obscuring the surgical field of the heart (e.g., during a cardiac bypassprocedure). In some cases, as described below, evaluation of the heart by the processes herein can indicate that the heart needs assistance (e.g., to reduce clamping of the cardiac artery to provide blood and nourishment to the heart muscle, to increase clamping of the aorta, or to increase clamping of the vena cava), or the heart is functioning normally at which point the heart can be loaded (e.g., slightly) by reducing a clamping of the aorta and the vena cava to begin loading the heart. As described below, advantageously, the processes herein can provide evaluation of an unloaded heart (e.g., a heart in which no or substantially no blood is being pumped by the heart). Regardless, the perfusion device 160 can artificially preserve the organ 180 to prevent the organ 180 from dying during travel, during a procedure (e.g., open heart surgery), etc.

[0095] As shown in FIG. 2, the system 100 can include a sensor 162. The sensor 162 can be physiologically coupled to the organ 180 to sense a feature of the organ 180. For example, the sensor 162 can receive sensor information from the organ 180. The computing device 158 (or other computing device) can use the sensor information to determine the organ function (e.g., along with the plurality of images). For example, the sensor 162 can be an electrocardiogram (“ECG”), a plethysmography sensor (e.g., light-based photoplethysmography (“PPG”), etc., to detect a heart rate of the heart when the organ 180 is a heart. As described below, the computing device 158 can compensate the tracking information of the heart using the heart rate (e.g., since faster heart rates indicate more movement of pixels over time, and vice versa).

[0096] Although not shown, the organ 180 can be positioned within a container of the perfusion device 160 and sealed from the ambient environment (e.g., hermetically sealed). The container can be translucent or transparent so as to allow light therethrough. The imaging devices 152, 154, 156, and the computing device 158 can be positioned outside of the container. In this way, the imaging devices 152, 154, 156 can each acquire the plurality of images while the organ 180 secured within the container so as to prevent contaminants from reaching the organ 180.

[0097] FIG. 3 shows a schematic illustration of specific block diagrams of computing devices 200, 202, and a server 206. The computing device 200 can be implemented in a similar way as the other computing devices, such as the computing devices 108, 158, 194. Therefore,the description of the computing device 200 pertains to the other computing devices (e.g., the computing devices 108,158, 194) and vice versa.

[0098] The computing device 200 can include a processor(s) 210, a power source(s) 212, input(s) / output(s) 214, a display(s) 216, a communication system 218, and memory 220. The processor 210 can be any suitable hardware processor or combination of processors, such as a central processing unit (“CPU”), a graphics processing unit (“GPU”), etc., which can execute a program (e.g., retrieved from memory 220) that can include the processes described below or otherwise herein.

[0099] The power source 212 can supply power to all components within the computing device 200 (or other components of the broader system, such as, for example, the systems 100, 150). In some specific cases, the power source 212 can be a non-rechargeable batter or a rechargeable battery. In other cases, the power source 212 can be a hardwired connection (e.g., a Universal Serial Bus connection), or the power source 212 can be an electrical storage device (e.g., a battery).

[0100] The input(s) / output(s) 214 of the computing device 200 can be indicators, sensors, actuated buttons, etc. For example, the inputs(s) / output(s) 214 can include a user interface device (e.g., a mouse, touchscreen, etc.) so as to receive a user input from the user (e.g., to adjust how markers are distributed on an image). As another example, the input(s) / output(s) 214 can include indicators, sensors, actuatable buttons, a keyboard, a mouse, a graphical user interface, a touch-screen display, etc., so as to interact with images presented on the display 216 (e.g., the graphical user interface).

[0101] The display 216 can be implemented in different ways. For example, the display 216 can be part of the computing device 200 (e.g., within the same housing), or separate from the computing device 200. In some non-limiting examples, the display 216 can present a graphical user interface. In some non-limiting examples, the display 216 can be implemented using any suitable display devices, such as a monitor, a touchscreen, a television, etc.

[0102] The communication system 218 of the computing device 200 can include any suitable hardware, firmware, and / or software for communicating with other components of the devices and specifically devices of the systems herein (e.g., the systems 100, 150). For example, the communications system 218 can include one or more transceivers, one or morecommunication chips and / or chip sets, etc. In a more particular example, communications system 218 can include hardware, firmware and / or software that can be used to establish a coaxial connection, a fiber optic connection, an Ethernet connection, a USB connection, a WiFi connection, a Bluetooth connection, a cellular connection, etc. In some cases, the computing device 200 can communicate with the computing device 202, the server 206, and other computing devices via a communication network 208.

[0103] The memory 220 can include a non-transitory computer-readable medium including any suitable volatile memory, non-volatile memory, storage, or any suitable combination thereof. For example, the memory 220 can include random-access memory (“RAM”), static random-access memory (“SRAM”), read-only memory (“ROM”), electrically erasable programmable read-only memory (“EEPROM”), one or more flash drives, one or more hard disks, one or more solid state drives, one or more optical drives, etc. In some nonlimiting examples, the memory 220 can have encoded thereon a computer program for controlling operation of the processor 210.

[0104] The memory 220 can store data and instructions for implementing some or all of the processes herein. For example, the memory 220 can store a plurality of images (e.g., acquired from an imaging device) to, for example, analyze the images. As another example the memory 220 can store instructions for generating one or more markers (e.g., virtual markers) on each image. In some cases, this can be a grid of markers on each organ image. As yet another example, the memory 220 can store machine learning models 222, 224. Each machine learning model 222, 224 can be a trained model each of which can be trained to implement a task. For example, the machine learning model 222 can isolate the organ (e.g., a portion of the organ or the entire organ) from each image of a plurality of images, and can be trained accordingly. Similarly, for each image of the plurality of images, the machine learning model 222 can isolate the background of each image (e.g., from the isolation of the organ ). For example, the computing device 200 can subtract the organ image (extract by the machine learning model 222) from the image (e.g., raw image, inputted image, provided image, etc.) to generate a background image. As yet another example, the machine learning model 224 can track the markers across the organ images (e.g., track each marker across different images of the plurality of organ images). This tracking can result in the machine learning model 224 providing tracking information (e.g., for each marker). This tracking information can be in theform of kinematics information, that is, the path the particular marker took across the images including the coordinates (e.g., X-Y coordinates) including over time, the velocity of the marker (e.g., pixels per second), etc. This tracking information can be used for determining movement of the organ. Similarly, the machine learning model 224 can track the markers across the background images (e.g., track each marker across different images of the plurality of background images). This tracking of the background can be used to normalize the movement of the organ.

[0105] As shown in FIG. 3, the computing device 200 can be in communication with the computing device 202, the server 206, and with other computing devices. In this way, each computing device (and the server 206) can perform some or all of the steps of the processes described herein. This can be advantageous in that computation heavy tasks including analyzing with machine learning models can be distributed among multiple different computing devices (e.g., the computing device 200 performing machine learning related tasks).

[0106] FIG. 4 shows an example of a machine learning model 250, which can be a specific implementation of the machine learning model 222 (e.g., the machine learning model 250 can be stored in a computing device). The machine learning model 250 can isolate an organ from a provided (or inputted image) and can correspondingly output an organ image and a background image. For example, an image 252 (e.g., an optical image) is provided to the machine learning model 250, and the machine learning model 250 outputs an organ image 254 and a background image 256. As shown in FIG. 4, the organ image 254 is the entire or substantially the entire organ of the image 252. In some cases, the background image 256 can be generated by subtracting the organ image 256 from the image 252 (e.g., the inputted image).

[0107] As shown in FIG. 5 A, the image 252 is an optical image of a heart acquired during open heart surgery or is an ex vivo image of the heart (e.g., during machine perfusion of the heart). The machine learning model 250 isolates the heart from the image (and the background). In other words, the machine learning model 250 can identify the heart from nonheart objects including the background. The outputted image (e.g., the organ image) is an image of the heart with the background removed. Similarly, the background image 256 includes the background without the heart. In some cases, the images 254, 256 can be time stamped with the same time stamp as the image 252 (e.g., since the images 254, 256 are resultant from the image 252) or otherwise ordered such that the images are analyzed accordingto the time taken. In some specific cases, the machine learning model 250 can be a segmentation machine learning model.

[0108] FIG 5 A shows an example of a machine learning model 260, which can be a specific implementation of the machine learning model 220 (e.g., the machine learning model 260 can be stored in a computing device). The machine learning model 260 can isolate a portion of an organ from a provided (or inputted image) and can correspondingly output a partial organ image. In some cases, a portion of an organ can be a feature of an organ (e.g., an anatomical feature of the organ), an anatomical structure of the organ, etc. In some cases, an image 262 (e.g., an optical image, which can be the same as the image 252) is provided to the machine learning model 260, and the machine learning model 260 outputs a partial organ image 264. As shown in FIG. 5A, the partial organ image 264 (or stated differently an organ feature image 264) can be a specific anatomical feature of the organ, where the machine learning model 260 can be trained to identify the specific anatomical feature of the organ. In some cases, the organ feature image 264 can be the entire or substantially the entire organ feature from the image 262.

[0109] As shown in FIG. 5 A, the image 262 is an optical image of a heart acquired during open heart surgery or is an ex vivo image of the heart (e.g., during machine perfusion of the heart). The machine learning model 260 isolates an anatomical feature of the heart from the image 262. In other words, the machine learning model 260 can identify the anatomical feature of the heart from non-heart objects, and other non-anatomical features (e.g., different anatomical features of the heart, such as the aorta). The outputted image (e.g., the organ feature image) is an image of the organ feature with other structures and the background removed. In this case, the organ feature image 264 is a right ventricle of a heart. In some cases, as described below, right ventricles of the heart can be more likely to fail and may necessitate specific analysis. Similarly, although not shown, a computing device can isolate other anatomical features from the heart for specific analysis, such as for example, the left ventricle, the right atria, the left atria, etc. In some cases, the image 264 can be time stamped with the same time as the image 252 (e.g., since the image 264 is resultant from the image 262) or otherwise ordered such that the images are analyzed according to the time taken. In some specific cases, the machine learning model 260 can be a segmentation machine learning model.

[0110] FIG 5B shows an example of a machine learning model 270, which can be a specific implementation of the machine learning model 260 (e.g., the machine learning model 270 can be stored in a computing device). The machine learning model 270 can also isolate a portion of an organ from a provided (or inputted image) and can correspondingly output a partial organ image, similarly to the machine learning model 260. In this case, however, an organ image 272 (e.g., the organ image 254) is inputted to the machine learning model 270 that is trained to isolate an anatomical feature from the inputted organ image. In some cases, therefore, the machine learning model 270 can be the same as the machine learning model 260. In some configurations, the machine learning model 270 may perform better with inputted organ images than raw images, since the background has already been removed.

[0111] FIG. 6 shows an example of generating markers on an image and a machine learning model 302. The machine learning model 302 can be a specific implementation of the machine learning model 224. A computing device can generate one or more markers at block 300 on an image (e.g., virtual markers). For example, an image 301 (e.g., an optical image) is provided to the computing device, and the computing device overlays a plurality of markers 306 on the image 301 (e.g., with a plurality of rows of markers and a plurality of columns of markers). The plurality of markers 306 can be a grid of markers, where adjacent markers are separated uniformly from each other. For example, each marker in a row of markers can be separated uniformly from each other (e.g., the exact same amount), such as, for example, 20 pixels. As another example, each marker in a column of markers can be spaced uniformly from each other (e.g., the exact same amount), such as, for example, 10 pixels. In some cases, adjacent columns of markers can be separated from each other by a first distance and adjacent rows of markers can be separated from each other by a second distance. In some cases, the first distance is greater than the second distance. Although one image is shown in FIG. 6, a plurality of organ images can be inputted to overlay a gird of markers on each image of the organ images. In some cases, the number of markers is substantially the same (or exactly the same) across the overlaid organ images. In this way, the computing device (e.g., the machine learning model 302) can track each individual marker and the path taken across images.

[0112] The machine learning model 302 can track the one or more markers across the organ images having the one or more markers. More specifically, the organ image 304 with the grid of markers 306 can be inputted into the machine learning model 302 along with images308 which are different organ images isolated from respective images (e.g., raw images) and similarly overlaid with a grid of markers. The machine learning model 302 can be trained to track each marker throughout a plurality of images. For example, the marker 310 appears in the grid of markers 306 and also the grid of markers in the images 307. The machine learning model 302 can track the movement of this marker across the images 307 to determine the coordinates of the marker (e.g., the coordinates of the tracker, such as its x-y position) across the images. In addition, the velocity of each marker can be determined, since there is temporal information tied to the organ images, such as, for example, time stamping of the images, an acquisition speed or frequency (e.g., in frames per second), etc. An example of the path taken by a marker across the organ images is shown in FIG. 7. Referring back to FIG. 6, the machine learning model 302 can complete this process for each marker and can thus output tracking information 312. This tracking information, then, can include for each marker, coordinates of the marker taken, a velocity of the marker across the images. In some cases, this can include a total path length of the marker (e.g., from the coordinates), a peak velocity of the marker, etc. In some cases, this can be kinematic data for each marker. In some cases, the tracking information can define sub tracking information that is the tracking information for one marker. Accordingly, the tracking information can include the same numbers of sub tracking data as there are numbers of markers. In some cases, the tracking information can be consolidated, such as averaging the total distance traveled for all the markers (e.g., where distance traveled is in pixels, or pixels per beat), averaging the peak velocity for all the markers (e.g., taking the peak velocity for all the markers and dividing by the number of markers, in which the peak velocity can be in pixels per second). As described below, the distance traveled and the peak velocity of the markers are useful in determining organ function.

[0113] Although FIG. 6 has been shown with organ images, the process can be used to track specific anatomical features of an organ (e.g., the right ventricle of the heart, the left ventricle of the heart, etc.). Accordingly, the tracking information can be for an anatomical feature of an organ (and not for the entire organ itself). In this way, the processes herein can advantageously evaluate anatomical features specifically, such as, for example, the right ventricle and the left ventricle. In some cases, this can include tracking sub -anatomical features. For example, the images can be partial organ feature images, each of which are overlaid with a grid of markers, and tracked.

[0114] In some non-limiting examples, isolating an organ, an anatomical feature of an organ, or a sub anatomical feature of an anatomical feature, can include generating a plurality of markers (e.g., a grid of markers), such that the plurality of markers are on the organ, anatomical feature, sub anatomical feature, etc. (e.g., and not on the other portions of the image, the plurality of markers are only on the organ, anatomical feature, sub anatomical feature, etc.). For example, a user can adjust the grid of markers, based on a user input (e.g., from a computing device and interacting with the image presented on a display).

[0115] Although FIG. 6 is focused on the organ image 301 and tracking movement of markers of the organ, the machine learning model 302 can also track the movement of the background. For example, each corresponding background image (e.g., a complementary background image to the organ image 301) can be used by a computing device at the block 300 to generate a grid of markers on the background (e.g., overlaid on the background). Then, similarly, each background image with a grid of markers can be inputted into the machine learning model 302 to track the position of each marker across the images. Accordingly, the machine learning model 302 can output tracking information for the markers (e.g., with sub tracking information for each marker). This tracking information can be used to compensate the movement of the markers of the organ (or anatomical feature, or sub anatomical feature). For example, the tracking information of the organ (or anatomical feature, or sub anatomical feature) can be compensated by the tracking information of the background (e.g., the background images). Specifically, the sub tracking information from each marker of the background can be averaged to generate a background value, and this background value can be subtracted from the sub tracking data for each marker of the organ (or anatomical feature, or sub anatomical feature) to generate compensated background information. In this way, the movement or noise of the background (e.g., slight movements of the imaging device during acquisition) can be removed. In some cases, the tracking information of the background can be calculated for each pair of adjacent background images (or background tracked images) in the series to compensate the tracking information for the complementary organ images (or anatomical feature images, sub anatomical feature images, etc.). For example, the average background movement can be calculated for background images XI and X2 (e.g., across all the markers in those images), with complementary organ images Y1 and Y2 (e.g., XI and Y1 being acquired at the same time, X2 and Y2 being acquired at the same time), and the averagebackground movement can be subtracted from each sub tracking information for each marker between Y 1 and Y2. In this way, the background subtraction can be further tailored to between image frames. This subtraction can be applied across all the plurality of images. FIG. 8 shows a flowchart of a process 400 for evaluating an organ (e.g., the entire organ, an anatomical feature of an organ, a sub anatomical feature of an anatomical feature, etc.). The process 400 can be implemented using any of the systems, components, etc., described herein, as appropriate. For example, the process 400 can be a computer implemented method that can be implemented using one or more computing devices (e.g., the computing device 108, 158, 200, 202, the server 206, etc.).

[0116] At block 402, the process 400 can include one or more computing devices receiving a plurality of images (e.g., acquired from one or more imaging devices). As described above, each image of the plurality of images can be an optical image. In some cases, each image of the plurality of images can be acquired by light reflected off the organ (e.g., each image is not an acoustic image, not an attenuation image, such as from x-rays passing through a tissue, etc.). Each image of the plurality of images can include an organ (e.g., a heart), an anatomical feature of an organ (e.g., the right ventricle of the heart), or a sub anatomical feature of an organ (e.g., a region of heart tissue), depending, for example, on what is desired to be tracked.

[0117] In some cases, the block 402 can include the one or more computing devices receiving a plurality of sets of images, with each set of images (e g., a plurality of images) being for an organ, an anatomical feature of an organ, or an anatomical feature of an organ. For example, each set of images can be for the same organ, anatomical feature, or sub anatomical feature (e.g., each image in each set has the same organ, anatomical feature, or sub anatomical feature). As another example, each set of images can be for different anatomical features or different sub anatomical features for the same organ. In this way, multiple sets of a plurality of images can cover the desired number of anatomical or sub anatomical structures (e.g., where one FOV of an imaging device does not capture an anatomical feature, but the other FOV of the same or different imaging device does capture the anatomical feature and vice versa). As a specific example, a first set of a plurality of images includes a right ventricle (an anatomical feature), but does not include a left ventricle (or does not include the entire left ventricle or the left ventricle is partially or fully obscured from view). Correspondingly, a second set of a plurality of images includes a left ventricle (a different anatomical feature), butdoes not include the right ventricle (or does not include the entire right ventricle or the right ventricle is partially or fully obscured from view).

[0118] In some cases, each image of the plurality of images can be an ex vivo image of an organ (e.g., or anatomical feature, sub anatomical feature, etc.) while the organ is undergoing ex vivo machine perfusion. In other cases, each image of the plurality of images can be an en vivo image while the organ is undergoing a surgical procedure, or otherwise visible (e.g., during an open surgical procedure where the organ is visible).

[0119] In some cases, each image of the plurality of images (or sets of plurality of images) has been acquired while the organ is unloaded (e.g., the heart is not pumping any blood). In other cases, each image has been acquired while the organ is partially loaded (e.g., the heart is pumping some blood). In still further cases, each image has been acquired while the organ is fully loaded (e.g., the heart is pumping at 100% capacity). As described herein, analyzing the functioning of organs in an unloaded state (e.g., in an ex vivo environment) using images can yield accurate predictors of future performance during loading.

[0120] At the block 404, the process 400 can include the one or more computing devices isolating an organ (or other structure) from each image of the plurality of images. This can include isolating each organ, each anatomical feature of an organ, or each sub anatomical feature of an organ (e.g., each of which can be the same organ, such as the heart). For example, in some cases, a user may desire multiple analyses, specifically, the organ to be tracked (e.g., the heart), one or more anatomical features of the organ to be tracked (e.g., a left ventricle and a right ventricle), one or more sub anatomical features of the organ to be tracked (e.g., visually suspect regions that a practitioner may want to further investigate), etc. In this case, the one or more computing devices can determine each organ, each anatomical feature of the organ, and each sub anatomical feature of the organ (e.g., a sub anatomical feature of a specific anatomical feature, such as a smaller region of a right ventricle of the heart) to be tracked. For example, the one or more computing devices can receive one or more user inputs indicative of the desired organ, anatomical feature of the organ, sub anatomical feature of the organ, etc., to be tracked. Accordingly, at the block 404, the one or more computing devices can isolate each organ, each anatomical feature of the organ, and each sub anatomical feature of the organ, based on the respective user input.

[0121] In some cases, isolating the organ, the anatomical feature of the organ, or the sub anatomical feature of the organ can include the one or more computing devices providing each image of the plurality of images to a respective machine learning model trained to isolate that particular organ, anatomical feature of the organ, or sub anatomical feature. For example, the one or more computing devices can generate, for each image, an organ image that identifies the organ by inputting the image (e.g., raw image) into the machine learning model trained to identify the organ. In this case, and as described above, each image can provide an organ image and a corresponding background image. In some cases, where an organ is desired to be tracked and anatomical features of the organ (or sub anatomical features of the organ) are also desired to be tracked, the one or more computing devices can, for each image, isolate each anatomical feature and each sub anatomical feature, which can be based on a corresponding user input. In some cases, this can include inputting each image (or organ image) into a corresponding machine learning model trained to identify the specific anatomical feature or sub anatomical feature. To aid the machine learning model to identify the specific anatomical feature or sub anatomical feature, a user input can be provided to the one or more computing devices indicating where on the inputted image the specific anatomical feature or sub anatomical feature is located. This can include a number of seed locations, which can aid the machine learning model to correctly identify the location. In other cases, the user can manually isolate each organ, each anatomical feature, each sub anatomical feature from each organ by, for example, tracing the boundaries on each image. Accordingly, these boundary tracings can be examples of user inputs, and the one or more computing devices can define each organ image, each anatomical feature image of the organ, and each sub anatomical feature of the organ by these boundaries. In some cases, isolating the organ or other substructure of the organ can be segmenting the organ or other substructure from each image of the plurality of images.

[0122] In some cases, the one or more computing devices can refine each actual isolated image (e.g., an organ image, an anatomical feature image, etc.) or proposed isolated image (e g., prior to generating that image), for example, based on a user input. In some cases, a machine learning model may miss, misidentify, or fail to locate a region, below a desired standard, and therefore, a user can, via a user input received by the one or more computing devices, adjust each actual isolated or proposed isolated image to capture the desired organ, sub structure of the organ, etc.

[0123] Although the block 404 can include manual isolation of images, and manual refinement of images (e.g., after being outputted or indicated by the respective machine learning model), in some cases, the block 404 proceeds automatically without any user inputs. This can be advantageous, in that the underlying analysis (and overlaying of the one or more markers) can be done much faster (e.g., this avoids the time needed for a user to manually isolate or adjust the isolation of images). Further, in the case of global organ analysis, especially with averaging information, slight deviations (e.g., areas missed by the machine learning model) may not impact the broader analysis or determination.

[0124] At block 406, the process 400 can include the one or more computing devices generating one or more markers (e.g., for each image of the plurality of images). In some cases, each marker of the one or more markers can be virtual markers. This can be advantageous, since virtual tracking can be done non-invasively and avoids invasive intervention (e.g., cannula insertions). The one or more markers can be a plurality of markers, such as a grid of markers (e.g., see FIG. 6.).

[0125] In some cases, the one or more computing devices can generate the one or more markers on each isolated image (e.g., outputted at the block 404) including each organ image, each anatomical feature of the organ, each sub anatomical feature of the organ, etc. For example, the one or more markers can span substantially the entire isolated image. In some cases, this can advantageously be an automated process, in which the one or more markers are placed on each isolated image using a predefined configuration (e.g., a preset grid size) by the one or more computing devices. In this way, since the desired region (e.g., organ, anatomical feature, or sub anatomical feature) has already been identified in the isolation step (e.g., at the block 404), the block 406 can proceed according to a predefined configuration of the one or more markers. In other cases, however, including when potions of the block 404 (or the block 404 is not at all implemented), the one or more computing devices can generate the one or more markers to span each organ, each anatomical feature of the organ, each sub anatomical feature of the organ, etc., of each image (e.g., of the plurality of images at the block 402, which can be the raw images) each of which can be based on a user input. For example, a user can place a grid of markers on each organ, each anatomical feature of the organ, and each sub anatomical feature of the organ. Therefore, in some cases, each image can have multiple sets of one or more markers (e.g., a plurality of markers, a grid of markers, etc.) each for trackinga different organ, anatomical feature of the organ, and sub anatomical feature of the organ, etc., as desired. As a specific example, the one or more computing devices can generate a first grid of markers on an organ of each image of the plurality of images to generate a plurality of marked organ images. The one or more computing devices can then generate a second grid of markers on an anatomical feature of the organ of each image of the plurality of images to generate a second plurality of marked anatomical feature images (e.g., where the anatomical feature is a right ventricle of the heart). The one or more computing devices can then generate a third grid of markers on a different anatomical feature of the organ of each image of the plurality of images to generate a third plurality of marked anatomical feature images (e.g., where the anatomical feature is a left ventricle of the heart). The one or more computing devices can then generate a fourth grid of markers on a sub anatomical feature of the organ of each image of the plurality of images to generate a plurality of marked sub anatomical feature images. These generated images can be grouped according to region to be analyzed (e.g., specific organ, specific anatomical feature, specific sub anatomical feature, etc.). In this way, not only can the overall organ function be analyzed, but also the individual structures of the organ can also be analyzed (e.g., which may help when the overall organ function is passable, but an anatomical feature function is not passable, specifically the heart passes, but the right ventricle does not).

[0126] In some cases, the number of the one or more markers, layout, etc., can be determined based on a user input, a size of the input image, etc. For example, a user can select the total number of markers, the density of the markers, etc. Further, the size of the input image can dictate the number of markers as larger images may require larger numbers of markers and vice versa.

[0127] In some non-limiting examples, as described above, the one or more computing devices can generate the one or more markers on each background image (e.g., a complementary image of the isolated organ image), or the background of each image of the plurality of images (e.g., in a similar manner as generating markers for organs, anatomical features, etc., on the images from the block 402). Similarly to cases where multiple regions are to be tracked (e.g., organ and a sub region of the organ), the background images or images with the one or more markers placed on the background can be grouped together.

[0128] At the block 408, the one or more computing devices track the position of each marker of the one or more markers. In some cases, this can include inputting each image of a set of images each having the one or more markers into a machine learning model (e.g., the machine learning model 302) that outputs tracking information (e.g., which can include sub tracking information for each marker). As an example, marked organ images can be inputted into the machine learning model and the machine learning model can output sub tracking information of each marker, marked anatomical feature images can be inputted into the machine learning model and the machine learning model can output sub tracking information of each marker, and so on. Although one machine learning model (e.g., the machine learning model 302) can serially analyze sets of images with markers thereon, there can be multiple machine learning models (e.g., each a tracking machine learning model that can be the same as the machine learning model 302). In this way, tracking information for each organ, each anatomical feature, each sub anatomical feature can be analyzed in parallel, which may help with real-time implementations

[0129] The sub tracking information of the tracking information can include, for each marker, the path traveled by the marker (e.g., in x-y coordinates) and the velocities of the marker across the images (e.g., from one image to the next in the series). In some non-limiting examples, therefore, the tracking information can include kinematics information for each marker.

[0130] At the block 410, the one or more computing devices can determine a function for the organ (or the specific structure being tracked, such as an anatomical feature, or a sub anatomical feature), which can be based on the tracking at the block 408 (and specifically the tracking information generated at the block 408). In some cases, the function can be advantageously quantitative. For example, the one or more computing devices can determine a contractility value (e.g., a the number of pixels traveled, for example, in a cardiac cycle, such as during systole) from the sub tracking information for each marker (e.g., using the path traveled across the images). As another example, the one or more computing devices can determine a compliance value (e.g., a the velocity of the marker in pixels traveled per second during, for example, a cardiac cycle, such as during systole) from the sub tracking information for each marker (e.g., using the path traveled across the images, coordinates and the time values in the form of time stamps, acquisition frequency, etc ). In some cases, the compliance valuecan be an average compliance value across all markers (e.g., the total path taken by each marker divided by the number of markers). In some cases, the contractility value can be the peak velocity of a marker between adjacent images (e.g., in the series). Similarly, then, the contractility value can be an average of the peak contractility values for each tracker.

[0131] In some non-limiting examples, the one or more computing devices can normalize a function, for example, before being analyzed (e.g., at the block 412). In some cases, this can be based on a size of an image, a resolution of an image, a size of an organ of the image, a dimension of an organ of the image, a zoom of an organ of an image, etc. For example, in some cases, a zoomed in image (e.g., a large relative organ) would correspondingly increase the total path length taken by a marker (and vice versa). In this case, then, the one or more computing devices can divide the compliance value (e.g., the total distance in pixels traveled for a given marker or a plurality of markers) by a dimension of an organ (e.g., a width, a height, a diagonal, an area such as the total area of the organ in the image, etc.) to generate a normalized compliance value. Further, this normalized compliance value can be further divided by the heart rate of the individual (e.g., during acquisition of the plurality of images) to generate a further normalized compliance value (e.g., the total distance in pixels traveled for a given marker accounting for the size of the organ, per heart beat). In some cases, the one or more computing devices can receive a heart rate (e.g., from a sensor, such as the sensor 162), in other cases, the heart rate can be determined from the tracking information (or sub tracking information). For example, the one or more computing devices can determine a heart rate by the movement in one dimension (e.g., the X dimension, or the Y dimension), which results in a uniform, heart rate signal (see, e.g., FIG. 17).

[0132] In some non-limiting examples, the one or more computing devices can normalize a contractility value (e.g., the velocity of a marker, an average velocity of markers, etc.). In this case the one or more computing devices can divide the contractility value (e.g., the pixels traveled per second for a given marker or a plurality of markers) by a dimension of an organ (e g., a width, a height, a diagonal, an area such as the total area of the organ in the image, etc.) to generate a normalized compliance value.

[0133] In some non-limiting examples, the function can be a health score of an organ. For example, the one or more computing devices can determine the health score, based on the specific values determined at the block 410 (e.g., in a look up table). For example, the healthscore can be based on a combination of a compliance value, a contractility value, etc., Further, the health score can be based on a plurality of compliance values (e.g., for each organ, each anatomical feature, each sub anatomical feature), a plurality of contractility values (e.g., for each organ, each anatomical feature, each sub anatomical feature), etc. In this case, therefore, the one or more thresholds at the block 414 can be one or more health score thresholds.

[0134] In some non-limiting examples, and as described above, the function can be determined at the block 410, for each set of a plurality of images (e.g., with each set of images acquired at a different imaging angle).

[0135] At the blocks 412, 414, 416 the one or more computing devices can provide an action for the organ based on the one or more functions (e.g., determined at the block 410). For example, at the block 412, the one or more computing devices can compare each function (e.g., the function for the organ, the function for each anatomical feature, the function for each sub anatomical feature) to a threshold value (e.g., the same or different threshold value). If at the block 412, one or more of the functions exceeds a threshold (e.g., the same or different thresholds), the process 400 can proceed to the block 414. If, however, if one or of the functions (e.g., all the functions) do not exceed a threshold (e.g., is below the threshold), the process 400 can proceed to the block 416. In some cases, the threshold can be a compliance threshold, which can be substantially or exactly 105 pixels (e.g., 105 pixels per beat). In some cases, the threshold can be a contractility threshold, which can be substantially or exactly 340 pixels per second (e.g., during contraction, such as a systole of the heartbeat). This compliance threshold has been found to accurately predict a heart of tolerating future loading (e.g., values above this threshold). Similarly, the contractility threshold has been found to accurately predict a heart of tolerating future loading (e.g., again, values above this threshold.

[0136] At the block 414, the one or more computing devices can provide an action. At this block, by passing all or at least some of the thresholds, the organ is predicted to function well (e.g., the organ passing a threshold, an anatomical feature passing a threshold, etc.) and tolerate loading. Accordingly, this block can include the one or more computing devices notifying a practitioner that the organ is normal, healthy, recommended to be transplanted, cleared to be transplanted, functioning properly, the organ is suitable for transplantation, etc. In some cases, this notification can be presenting, on a display, a corresponding indication (e g., that the organ is normal, healthy, recommended to be transplanted, cleared to be transplanted, functioningproperly, does not need mechanical assistance after surgery, etc.). In some cases, this can include the one or more computing devices transmitting a message that includes the notification (e.g., to a health chart of a recipient patient).

[0137] In some non-limiting examples, the block 414 can include the one or more computing devices notifying that the organ can be loaded (e.g., unblocked, unclamped, etc.) or that a blood vessel supplying blood to the organ can be occluded (e.g., blocked, clamped, etc.) since the organ is functioning normally and can tolerate the stress of loading. In some cases, this notifying can be in the form of a message transmitted similarly as described above.

[0138] In some non-limiting examples, the block 414 can include the one or more computing devices controlling an operation of a perfusion device. For example, this can include increasing a loading of an organ coupled to the perfusion device, decreasing or maintaining the oxygen level of a perfusate of the perfusion device, decreasing or maintaining the level of a nutrient (e.g., glucose) in the perfusate of the perfusion device, increasing the heat of the perfusate (e.g., via the heater), decreasing a cooling of the perfusate (e.g., via the cooler), increasing the mean aortic pressure, decreasing the coronary blood flow (e.g., the one or more computing devices adjusting a rate of infused medication, such as via a pump that draws medication from a reservoir and to the patient), etc.

[0139] In some cases, after the block 414, the process 400 can proceed back to the block 402 to receive a different plurality of images (e.g., the organ can be reanalyzed). In this way, the organ can be periodically evaluated, such as during transportation on an ex vivo perfusion device. In some cases, therefore, the block 414 is only implemented after multiple iterations of the process 400 that indicates passing, including when the organ is actually in the operating room.

[0140] At the block 416, the one or more computing devices can provide an action. At this block, by failing to pass all or at least some of the thresholds, the organ is predicted to function poorly (e.g., the organ failing to pass a threshold, an anatomical feature failing to pass a threshold, etc.) and is unlikely to tolerate loading well. Accordingly, this block can include the one or more computing devices notifying a practitioner that the organ is abnormal, unhealthy, not recommended to be transplanted, not cleared to be transplanted, not functioning properly, the organ is unsuitable for transplantation, needs mechanical assistance after surgery, etc. In some cases, this notification can be presenting, on a display, a corresponding indication(e g., that the organ is abnormal, unhealthy, not recommended to be transplanted, not cleared to be transplanted, not functioning properly, etc.). In some cases, this can include the one or more computing devices transmitting a message that includes the notification (e.g., to a health chart of a recipient patient).

[0141] In some non-limiting examples, the block 416 can include the one or more computing devices notifying that the organ should be unloaded (e.g., unblocked, unclamped, etc.) or that a blood vessel supplying blood to the organ can be unoccluded (e.g., unblocked, unclamped, etc.) since the organ is not functioning normally and is not tolerating or will not tolerate the stress of loading. In some cases, this notifying can be in the form of a message transmitted similarly as described above.

[0142] In some non-limiting examples, the block 416 can include the one or more computing devices controlling an operation of a perfusion device. For example, this can include decreasing a loading of an organ coupled to the perfusion device, increasing the oxygen level of a perfusate of the perfusion device, increasing the level of a nutrient (e.g., glucose) in the perfusate of the perfusion device, decreasing the heat of the perfusate (e.g., via the heater), increasing a cooling of the perfusate (e.g., via the cooler), decreasing the mean aortic pressure, increasing the coronary blood flow (e.g., the one or more computing devices adjusting a rate of infused medication, such as via a pump that draws medication from a reservoir and to the patient), etc. _[

[0143] In some cases, after the block 416, the process 400 can proceed back to the block 402 to receive a different plurality of images (e.g., the organ can be reanalyzed). In this way, the organ can be periodically evaluated, such as during transportation on an ex vivo perfusion device. Accordingly, in some cases, the block 416 is only implemented after multiple iterations of the process 400 that indicates failing, including when the organ is actually in the operating room. In this way, the organ is not deemed to have failed based only on one set of images.EXAMPLES

[0144] The following examples have been presented in order to further illustrate aspects of the disclosure, and are not meant to limit the scope of the disclosure in any way. The examples below are intended to be examples of the present disclosure and these (and other aspects of the disclosure) are not to be bounded by theory.EXAMPLE 1

[0145] FIG. 9 shows a schematic of a software pipeline in accordance with an non-limiting example of the invention.

[0146] FIGS. 10A-10F show video frame snapshots of a beating heart with overlaid pixels for tracking motion.

[0147] Ex vivo perfusion has increased the number of organs being considered for transplantation. However, a reliable prognostic biomarker for post-transplant organ function that can be measured during ex vivo perfusion has not yet been established. A total of 6 hearts (2 human and 4 porcine) were procured and initiated on ex vivo perfusion. Hearts were maintained in a Langendorff model for 6 hours, and those that demonstrated continued function underwent ventricular loading for a subsequent 6 hours. Laboratory tests and videos were obtained every 30 minutes. Ten second videos were taken with a consumer cell phone in slow-motion mode (240 frames per second). Software was used to monitor the movement of virtual trackers overlaid on the left and right ventricles. Video kinematic parameters were calculated and compared to laboratory values at each time point, as well as to the organ’s ability to tolerate ex -vivo loading. Analysis of the kinematic derived metrics identified distinct patterns over time that correlated with future heart function (FIGS. 11A-F and FIGS. 12A- F). Two hearts (FIGS. 11 and 12, E and F) did not have significant improvement in their contractility or compliance after being initiated on ex vivo perfusion and had to be stopped after 4 hours in Langendorff. Two hearts (FIGS. 11 and 12 B and C) demonstrated initial improvement in function after re-perfusion but then had deterioration in their kinematic parameters with a corresponding rising lactate and did not tolerate any subsequent loading. The remaining two hearts (FIGS. 11 and 12 A and D) demonstrated improvement in their kinematic metrics that were better sustained in Langendorff and both tolerated 6 hours of loading.

[0148] This study suggests video kinematic parameters correlate with laboratory markers of organ function and can provide a non-invasive method of quantifying cardiac function during ex-vivo perfusion.

[0149] Summary of Software for Analyzing Heart Function from Surface Video

[0150] The software utilized may be designed to analyze heart function by processing surface video recordings of the heart. In one non-limiting example, the software operates through the following steps:

[0151] Step One - Video Collection: The software begins by collecting surface video of the heart, capturing real-time motion and activity. This could be recorded from a phone, tablet, camera, or other mobile device.

[0152] Step Two Automatic Segmentation: The software automatically segments the heart within the video, distinguishing it from the background, including any overlying tubing, cannulas, or other objects. This may be done autonomously or involve user input to correctly identify the heart. The segmentation is performed using a convolutional neural network (CNN).

[0153] Step Three Mesh Network Overlay: After segmentation, the software overlays a mesh network of points onto the heart. Alternatively, users can manually select specific locations they wish to track.

[0154] Step Four Heart Motion Tracking: The software tracks the motion of each point within the mesh over the course of the video, allowing for precise analysis of individual heart movements. The tracking is performed utilizing a recursive neural network.

[0155] Step Five Background Mesh Overlay: The software also overlays a separate mesh network of points onto the background content of the video.

[0156] Step Six Background Motion Tracking: It then tracks the motion of these background points to account for any camera movement during the recording. The tracking is again performed utilizing a recursive neural network.

[0157] Step Seven Motion Subtraction: The software subtracts the background motion from the heart motion to isolate the heart’s movement. This isolation provides a clearer understanding of the heart’s true motion, free from external influences like camera movement.

[0158] Step Eight Data Utilization: The isolated heart movement data can be used for various purposes, including: quantifying the overall function of the heart; analyzing the motion of different parts of the heart individually; tracking changes in heart function over time; and prognosticating future heart function.

[0159] This software serves as a powerful tool for clinicians and researchers by providing detailed and accurate assessments of heart function from surface video recordings.

[0160] The video analysis software could be used to assess ex vivo human hearts while they are being perfused during the transplantation process.

[0161] The database that was created contains longitudinal video of hearts that perform well or start to struggle while undergoing ex vivo normothermic perfusion gives insights into early prognostic markers of hearts that are high risk of having poor performance if transplanted, which has immense clinical value and is not easily reproduced.EXAMPLE 2

[0162] Over the last decade, there has been increasing adoption of a new technique for organ transplantation termed donation after circulatory death (“DCD”). In this method, the donor has suffered injury that is deemed too severe to ever achieve meaningful recovery, and their family decides that they would want to be an organ donor. However, they do not technically meet all criteria necessary for legal brain death, but with this new category of donation they are still able to donate. They are brought to the operating room and life sustaining artificial support is removed (mechanical ventilation and medications). If they then expire, their organs are procured for transplantation according to their wishes. Unlike brain death donation, however, the organs experience the effects of true circulatory death during this process, which poses the risk of causing damage to the organs, especially the heart. Heart transplantation in this DCD methodology has shown acceptable survival rates comparable to brain death donation, but immediately after the transplantation these DCD hearts more frequently require other forms of mechanical circulatory support (“MCS”) including extracorporeal membrane oxygenation, (“ECMO”) short-term. Most recover after a few days, but this process inherently increases risk of morbidity to the patient. Being able to better quantify the performance of the heart during transportation from the donor to the recipient hospital would be immensely valuable to better ascertain which hearts are at an increased risk of developing early poor performance requiring MCS. Moreover, many potential hearts from seemingly marginal donors (e.g., older age, high risk mechanism of death, or unknown cardiac history) that would perform adequately if transplanted into a patient in need, are not evenattempted to be transplanted because there is not sufficient methodology to confirm that they have adequate function.

[0163] Having an accurate and reproducible method of quantifying the function of DCD hearts when they are undergoing ex vivo reperfusion before implantation would both improve outcomes for current DCD heart recipients and increase the number of transplantations that could be performed, saving many more lives.

[0164] To this end, techniques in this disclosure have been developed to quantify the function of a heart while it undergoes normothermic ex vivo perfusion. The medical team takes a 10 second slow-motion video of the beating heart, which is provided to the system. Virtual markers are placed on the heart and the software tracks these pixels frame-by-frame. Then, the tracking is analyzed, identifying different parts of the cardiac cycle to quantify beat-to-beat heart performance. More than 200 videos of ex vivo pig hearts were analyzed longitudinally as they undergo normothermic perfusion, providing insights into performance characteristics.

[0165] There is not a reliable biomarker for identifying which hearts are at highest risk for primary graft dysfunction. It is hypothesized that ex-vivo video kinematics (“EVVK”) of unloaded beating hearts can help predict post-implant heart performance. Porcine hearts procured following a DCD protocol were re-animated in unloaded ex situ heart perfusion (“ESHP”) before transitioning to a loaded configuration. Five second videos were recorded using a consumer cell phone. Machine learning EVVK software was developed to autonomously quantify heart function. Of 22 porcine hearts, 6 (27%) did not tolerate loading. EVVK demonstrated prognostic capability to differentiate the future loaded performance in terms of both compliance (p=0.004) and contractility (p=0.003). Proof-of-concept has been demonstrated of using EVVK on human hearts (n=4) to provide a quantitative functional assessment of the allograft during transportation from the donor to recipient center. Overall, non-invasive EVVK may be able to aid clinicians in quantitative heart evaluation.

[0166] Ex-situ heart perfusion (“ESHP”) is a promising tool for increasing the number of heart transplants performed through transplantation of hearts after circulatory death (“DCD”) and from extended criteria donors (“ECD”). However, DCD hearts undergoing ESHP have higher rates of primary graft dysfunction (“PGD”) and currently there is not a reliable biomarker to identify which allografts are most at risk. ESHP provides the unique opportunity to directly visualize the beating allograft, but clinically it is currently limited to only aqualitative and subjective assessment. Ex -vivo video kinematics (“EVVK”) is a promising non-invasive method for providing a quantitative and reproducible assessment of heart function. This example aims to assess the utility of EVVK during normothermic reperfusion to predict post implant allograft function in a large animal model and demonstrate feasibility to quantify human heart function using this technology during ESHP in a clinical setting.

[0167] This study was approved by IACUC (#2024N000002) and IRB (#2017P001969). Porcine heart procurement was performed via a DCD process with a lower MAP definition for the start of warm ischemia time to make the hearts higher risk. The procured hearts underwent 1 hour Langendorff followed by 3 hours of loading on ESHP. Videos were captured using a consumer cell phone for 5 seconds at the end of the Ih of Langendorff perfusion immediately before switching to the loaded configuration . For the human study, videos were taken of hearts on an organ care system (“OCS”) (Transmedics, Andover MA) at the donor center immediately before covering the heart within the OCS to leave the facility, and again at the recipient center immediately before cooling the heart to arrest for implantation .

[0168] FIG. 13 shows software developed for automated, non-invasive quantification of ex-vivo beating heart function using video kinematics.

[0169] Video analysis software was developed to automate the quantification of the ex vivo video kinematics as demonstrated in FIG. 13. The software segments the heart using a finetuned convolutional neural network, places a mesh grid of virtual markers over the myocardium, and utilizes a transformer-based machine learning model to track the movement of each virtual tracker . The movement of each virtual tracker is analyzed individually and then averaged to quantify ventricular compliance (total displacement) and peak contractility (maximum ). These values were normalized by heart rate and heart size.

[0170] The EVVK metrics were compared to outcomes. The porcine hearts were dichotomized into those that tolerated loading (lasted at least 2h of loaded “working mode” configuration) and those that did not tolerate loading (failed within 2h of loading). Chart review was performed to determine outcomes of the OCS recipients.

[0171] A total of 22 pigs (30-35 kg) underwent heart procurement following the DCD simulated protocol. After 1 hour of Langendorff, 16 (73%) tolerated subsequent loaded configuration for at least 2 hours. Compared to hearts that tolerated loading, hearts that did not tolerate loading had significantly lower compliance (median [IQR], 68 [66, 79] vs 142 [111,166] pixel / heartbeat, p = 0.004) and contractility (281 [246, 330] vs 612 [461, 973] pixel s / second, p=0.003) at the end of Langendorff (FIG. 14).

[0172] FIG. 14 shows two graphs, a left hand graph of different hearts’ average distance traveled at the donor center compared to at the recipient center, and a right hand graph of different hearts’ median peak velocity at the donor center compared to at the recipient center. FIG. 14 shows a significant difference in the quantifiable function of unloaded hearts that did and did not tolerate subsequent loading in terms of their video kinematic quantified metrics of compliance and contractility.

[0173] Videos were obtained for a total of 4 human hearts on the commercially available OCS system. Total time on ESHP was 271±79 minutes. There was a significant decrease in both compliance (median [IQR], 775 [671, 872] vs. 376 [348, 416] pixel / heartbeat, p=0.026) and contractility (2842 [2607, 3212] vs. 1379 [1065, 1771] pixel / sec, p=0.026) for all hearts during this time of transportation (FIG. 15).

[0174] FIG. 15 shows two graphs, a left hand graph of the average distance traveled of hearts that did not load versus hearts that tolerated loading, and a right hand graph of the mean peak velocity of hearts that did not load versus hearts that tolerated loading. FIG. 15 shows a significant change in compliance and contractility of human hearts on ESHP while traveling form the donor to recipient centers.

[0175] None of the four patients experienced early severe PGD. Additional studies demonstrated robustness of the processes to inconsistency in the angle and distance of video acquisition and the ability to quantify regional functional assessments such as left and right heart function individually .

[0176] In the present example an automated method of quantifying ex-vivo heart function non-invasively was developed. Early data from a porcine model suggests a prognostic potential in identifying hearts that are at high risk for PGD. Moreover, a proof-of-concept was shown implementing this technology into the current standard of care workflow for DCD heart transplantation. Currently, clinicians assess allograft function during ESHP using visual inspection, which is highly subjective and hard to standardize. This non-invasive, automated method could reduce this variability. It relies on only a 5 second video taken from a consumer grade mobile phone camera, making it easily reproducible without expensive equipment andcan be integrated seamlessly into current clinical workflow. Moreover, the non-invasive nature eliminates any concern for causing damage to the allograft.

[0177] Early data from a porcine DCD model demonstrates the non-invasive technique’s potential in predicting future loaded heart function from videos of the unloaded hearts. Interestingly, data from human hearts demonstrated a significant decrease in the video kinematic derived contractility and compliance metrics while the hearts were being transported on normothermic reperfusion. Reassuringly none of the patients in the present study experienced PGD, but these data suggest some degree of decrease in heart function during ESHP may be occurring clinically.

[0178] In conclusion, an automated approach is provided to obtain video kinematics on unloaded beating hearts which may be able to predict post-implant heart function. Further study is warranted to substantiate these early findings and determine what functional metrics or change in function while on ESHP may indicate a high risk for PGD. Moreover, combining the real-time quantitative assessment of heart function with donor factors, recipient factors, and other biomarkers that can be obtained on ESHP into a single multi-modal prognostic model has significant potential for helping optimize hearts on ESHP and aid clinicians in identifying hearts suitable for transplantation.

[0179] FIG. 16 shows an image of the setup including the camera, a tripod to support the camera, and the heart. FIG. 16 also shows the heart along with an output of one of the trackers.

[0180] FIG. 17A shows a graph of the path traveled for a marker and FIG. 17B shows a second graph of the y position traveled for the marker over time.

[0181] FIG. 18A shows a graph of the contractility of five virtual markers for two hearts, and FIG. 18B shows the compliance of five virtual markers for the two hearts.

[0182] FIG. 19 shows a graph of the true positive rate versus the false positive rate for average distance traveled to predict the likelihood of tolerating loading.

[0183] FIG. 20 shows a graph of the true positive rate versus the false positive rate for mean peak velocity to predict the likelihood of tolerating loading.

[0184] FIG. 21A-C shows video kinematics quantified from five different videos of the same heart taken at different angles to produce consistent metrics. Specifically FIG. 21A shows images of the five different angles of that video was taken at, FIG. 2 IB shows a graphof the average distance traveled versus the video angle, and FIG. 21 C shows a graph oof the mean peak velocity versus the video angle.

[0185] FIG 22A shows an illustration of the experimental process investigating the use of video kinematics during a porcine study with ESHP and FIG. 22B shows a human proof-of- concept study during current clinical workflow of DCD heart transplantation.FIG. 23A-C show the ability of the techniques herein to separate virtual markers by different regions and calculate regional specific metrics such as peak velocity and distance traveled for each marker within the region. Specifically, FIG.23A shows an image of a heart with left markers and right markers. FIG. 23B shows a graph of the peak velocity for the left and right sides. FIG. 23C shows a graph of the distance traveled for the left and right sides.EXAMPLE 3

[0186] Current methods of assessing heart function during ex-vivo perfusion are limited to lactate and visual inspection (i.e., a qualitative, subjective standard). Hearts, six hears (4 swine, 2 human). Various procurement types, cold ischemia times and perfusate. Target protocol for each was 6h Langendorff and 6h loaded. Video Kinematics. Hourly videos obtained from iPhone 12 Pro Max (240fps, 1080p). Virtual trackers retrospectively placed on lateral wall of the LV and RV respectively. 10 seconds of video analyzed at each time point Primary Outcome: Ability to tolerate ex vivo loading.

[0187] Video Kinematics Correlate with Performance

[0188] Heart A and B lasted 6h Langendorf and 6h loaded. Hearts C and D lasted 6 hours Langendorff, didn’t tolerate loading, Hearts E and F didn’t last 6h Langendorf. Next steps, automated models to track many virtual markers, DCD human heart videos on OCS to correlate with post-implant function. This example shows a promising proof of concept demonstrating video kinematics obtained from an iPhone may be able to provide a clinically useful, objective assessment of cardiac function during ex-vivo perfusion.

[0189] The present disclosure has described one or more preferred non-limiting examples, and it should be appreciated that many equivalents, alternatives, variations, and modifications, aside from those expressly stated, are possible and within the scope of the invention.

[0190] It is to be understood that the disclosure is not limited in its application to the details of construction and the arrangement of components set forth in the accompanying descriptionor illustrated in the accompanying drawings. The disclosure is capable of other non-limiting examples and of being practiced or of being carried out in various ways. Also, it is to be understood that the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting. The use of “including,” “comprising,” or “having” and variations thereof herein is meant to encompass the items listed thereafter and equivalents thereof as well as additional items. Unless specified or limited otherwise, the terms “mounted,” “connected,” “supported,” and “coupled” and variations thereof are used broadly and encompass both direct and indirect mountings, connections, supports, and couplings. Further, “connected” and “coupled” are not restricted to physical or mechanical connections or couplings.

[0191] As used herein, unless otherwise limited or defined, discussion of particular directions is provided by example only, with regard to particular non-limiting examples or relevant illustrations. For example, discussion of “top,” “front,” or “back” features is generally intended as a description only of the orientation of such features relative to a reference frame of a particular example or illustration. Correspondingly, for example, a “top” feature may sometimes be disposed below a “bottom” feature (and so on), in some arrangements or nonlimiting examples. Further, references to particular rotational or other movements (e.g., counterclockwise rotation) is generally intended as a description only of movement relative a reference frame of a particular example of illustration.

[0192] In some non-limiting examples, aspects of the disclosure, including computerized implementations of methods according to the disclosure, can be implemented as a system, method, apparatus, or article of manufacture using standard programming or engineering techniques to produce software, firmware, hardware, or any combination thereof to control a processor device (e.g., a serial or parallel general purpose or specialized processor chip, a single- or multi-core chip, a microprocessor, a field programmable gate array, any variety of combinations of a control unit, arithmetic logic unit, and processor register, and so on), a computer (e.g., a processor device operatively coupled to a memory), or another electronically operated controller to implement aspects detailed herein. Accordingly, for example, nonlimiting examples of the disclosure can be implemented as a set of instructions, tangibly embodied on a non-transitory computer-readable media, such that a processor device can implement the instructions based upon reading the instructions from the computer-readablemedia. Some non-limiting examples of the disclosure can include (or utilize) a control device such as an automation device, a special purpose or general purpose computer including various computer hardware, software, firmware, and so on, consistent with the discussion below. As specific examples, a control device can include a processor, a microcontroller, a field- programmable gate array, a programmable logic controller, logic gates etc., and other typical components that are known in the art for implementation of appropriate functionality (e.g., memory, communication systems, power sources, user interfaces and other inputs, etc.).

[0193] The term “article of manufacture” as used herein is intended to encompass a computer program accessible from any computer-readable device, carrier (e.g., non-transitory signals), or media (e.g., non-transitory media). For example, computer-readable media can include but are not limited to magnetic storage devices (e.g., hard disk, floppy disk, magnetic strips, and so on), optical disks (e.g., compact disk (CD), digital versatile disk (DVD), and so on), smart cards, and flash memory devices (e.g., card, stick, and so on). Additionally it should be appreciated that a carrier wave can be employed to carry computer-readable electronic data such as those used in transmitting and receiving electronic mail or in accessing a network such as the Internet or a local area network (LAN). Those skilled in the art will recognize that many modifications may be made to these configurations without departing from the scope or spirit of the claimed subject matter.

[0194] Certain operations of methods according to the disclosure, or of systems executing those methods, may be represented schematically in the FIGS, or otherwise discussed herein. Unless otherwise specified or limited, representation in the FIGS, of particular operations in particular spatial order may not necessarily require those operations to be executed in a particular sequence corresponding to the particular spatial order. Correspondingly, certain operations represented in the FIGS., or otherwise disclosed herein, can be executed in different orders than are expressly illustrated or described, as appropriate for particular non-limiting examples of the disclosure. Further, in some non-limiting examples, certain operations can be executed in parallel, including by dedicated parallel processing devices, or separate computing devices configured to interoperate as part of a large system.

[0195] As used herein in the context of computer implementation, unless otherwise specified or limited, the terms “component,” “system,” “module,” and the like are intended to encompass part or all of computer-related systems that include hardware, software, acombination of hardware and software, or software in execution. For example, a component may be, but is not limited to being, a processor device, a process being executed (or executable) by a processor device, an object, an executable, a thread of execution, a computer program, or a computer. By way of illustration, both an application running on a computer and the computer can be a component. One or more components (or system, module, and so on) may reside within a process or thread of execution, may be localized on one computer, may be distributed between two or more computers or other processor devices, or may be included within another component (or system, module, and so on).

[0196] In some implementations, devices or systems disclosed herein can be utilized or installed using methods embodying aspects of the disclosure. Correspondingly, description herein of particular features, capabilities, or intended purposes of a device or system is generally intended to inherently include disclosure of a method of using such features for the intended purposes, a method of implementing such capabilities, and a method of installing disclosed (or otherwise known) components to support these purposes or capabilities. Similarly, unless otherwise indicated or limited, discussion herein of any method of manufacturing or using a particular device or system, including installing the device or system, is intended to inherently include disclosure, as non-limiting examples of the disclosure, of the utilized features and implemented capabilities of such device or system.

[0197] As used herein, unless otherwise defined or limited, ordinal numbers are used herein for convenience of reference based generally on the order in which particular components are presented for the relevant part of the disclosure. In this regard, for example, designations such as “first,” “second,” etc., generally indicate only the order in which the relevant component is introduced for discussion and generally do not indicate or require a particular spatial arrangement, functional or structural primacy or order.

[0198] As used herein, unless otherwise defined or limited, directional terms are used for convenience of reference for discussion of particular figures or examples. For example, references to downward (or other) directions or top (or other) positions may be used to discuss aspects of a particular example or figure, but do not necessarily require similar orientation or geometry in all installations or configurations.

[0199] This discussion is presented to enable a person skilled in the art to make and use non-limiting examples of the disclosure. Various modifications to the illustrated examples willbe readily apparent to those skilled in the art, and the generic principles herein can be applied to other examples and applications without departing from the principles disclosed herein. Thus, non-limiting examples of the disclosure are not intended to be limited to non-limiting examples shown, but are to be accorded the widest scope consistent with the principles and features disclosed herein and the claims below. The accompanying detailed description is to be read with reference to the figures, in which like elements in different figures have like reference numerals. The figures, which are not necessarily to scale, depict selected examples and are not intended to limit the scope of the disclosure. Skilled artisans will recognize the examples provided herein have many useful alternatives and fall within the scope of the disclosure.

[0200] Also as used herein, unless otherwise limited or defined, “or” indicates a nonexclusive list of components or operations that can be present in any variety of combinations, rather than an exclusive list of components that can be present only as alternatives to each other. For example, a list of “A, B, or C” indicates options of: A; B; C; A and B; A and C; B and C; and A, B, and C. Correspondingly, the term “or” as used herein is intended to indicate exclusive alternatives only when preceded by terms of exclusivity, such as “either,” “one of,” “only one of,” or “exactly one of.” Further, a list preceded by “one or more” (and variations thereon) and including “or” to separate listed elements indicates options of one or more of any or all of the listed elements. For example, the phrases “one or more of A, B, or C” and “at least one of A, B, or C” indicate options of: one or more A; one or more B; one or more C; one or more A and one or more B; one or more B and one or more C; one or more A and one or more C; and one or more of each of A, B, and C. Similarly, a list preceded by “a plurality of’ (and variations thereon) and including “or” to separate listed elements indicates options of multiple instances of any or all of the listed elements. For example, the phrases “a plurality of A, B, or C” and “two or more of A, B, or C” indicate options of: A and B; B and C; A and C; and A, B, and C. In general, the term “or” as used herein only indicates exclusive alternatives (e.g. “one or the other but not both”) when preceded by terms of exclusivity, such as “either,” “one of,” “only one of,” or “exactly one of.”

[0201] Also as used herein, unless otherwise specified or limited, the terms “about” and “approximately,” as used herein with respect to a reference value, refer to variations from the reference value of ± 15% or less (e.g., ± 10%, ± 5%, etc.), inclusive of the endpoints of therange. Similarly, the term “substantially equal” (and the like) as used herein with respect to a reference value refers to variations from the reference value of less than ± 30% (e.g., ± 20%, ± 10%, ± 5%) inclusive. Where specified, “substantially” can indicate in particular a variation in one numerical direction relative to a reference value. For example, “substantially less” than a reference value (and the like) indicates a value that is reduced from the reference value by 30% or more, and “substantially more” than a reference value (and the like) indicates a value that is increased from the reference value by 30% or more.

[0202] Various features and advantages of the disclosure are set forth in the following claims.

Claims

CLAIMSWhat is claimed is:

1. A system comprising: an imaging device; a computing device in communication with the imaging device, the computing device being configured to: receive, from the imaging device, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ; isolate the organ in each image of the plurality of images to generate a plurality of organ images; generate a plurality of markers overlaid on each organ image of the plurality of organ images; track a position of each marker of the plurality of markers across the plurality of organ images; determine an organ function of the organ, based on the tracking of the plurality of markers; and provide an action for the organ, based on the organ function.

2. The system of claim 1, wherein the organ is a heart, the plurality of organ images is a plurality of heart images, and the organ function is of the heart.

3. The system of claim 1, wherein the isolating the organ for each image of the plurality of images to generate a plurality of organ images includes the computing device being further configured to for each image of the plurality of images, isolate the organ from the background.

4. The system of claim 3, wherein the computing device being further configured to: provide each image to a machine learning model that outputs a corresponding organ image, the machine learning model being trained to identify the organ from an inputted image.

5. The system of claim 1, wherein the plurality of markers include a grid of markers.

6. The system of claim 1, wherein the computing device is further configured to: generate tracking information, based on the tracking of each marker of the plurality of markers across the plurality of organ images.

7. The system of claim 6, wherein the computing device is further configured to: generate the tracking information by providing each organ image of the plurality of organ images with the plurality of markers overlaid to a machine learning model that outputs the tracking information, the machine learning model being trained to track the same marker of a plurality of markers across images.

8. The system of claim 7, wherein the tracking information includes sub tracking information that defines the movement of a given marker of the plurality of markers across the organ images.

9. The system of claim 1, wherein the organ function includes at least one of: a contractility of the organ; or a compliance of the organ.

10. The system of claim 1, wherein determining the organ function of the organ, based on the tracking of the plurality of markers includes the computing device being further configured to at least one of: account for a relative size of the organ in the plurality of images; or account for a contraction rate of the organ in the plurality of images.11 . The system of claim 10, wherein accounting for the relative size of the organ or accounting for the contraction rate of the organ includes the computing device being further configured to: standardize tracking information from the tracking the position of each marker across the plurality of images.

12. The system of claim 1, wherein the computing device is further configured to compensate the movement of the plurality of markers, based on the background movement.

13. The system of claim 12, wherein the computing device is further configured to for each image of the plurality of images, isolate the background to generate a plurality of background images.

14. The system of claim 12, wherein the plurality of markers is a first plurality of markers, and wherein the computing device is further configured to: generate a second plurality of markers overlaid on each background image of the plurality of organ images; track the position of each marker of the second plurality of markers across the plurality of background images; and determine the organ function of the organ, based on the tracking of the second plurality of markers.

15. The system of claim 14, wherein the computing device is further configured to: adjust the movement of the first plurality of markers across the plurality of organ images, based on the movement of the second plurality of markers across the plurality of background images.

16. The system of claim 15, wherein adjusting the movement of the first plurality of markers includes the computing device being further configured to: average the movement of the second plurality of markers across the plurality of background images; and subtract the movement of each marker of the first plurality of markers from the average.

17. The system of claim 1, wherein the computing device is further configured to: notify a user of the imaging device, during acquisition of the plurality of images by the imaging device, to at least one of: maintain the position of the imaging device relative to the organ being imaged; avoid changing the imaging angle or field of view (FOV) of the imaging device; or steading the imaging device.

18. The system of claim 1, wherein providing an action for the organ includes the computing device being further configured to: adjust a loading on the organ of a heart-lung machine coupled to the organ; adjust an operation of a perfusion device that is coupled to the organ; notify to a practitioner, that the organ is healthy; notify to a practitioner, that the organ is abnormal; notify to a practitioner, that the organ is cleared to be transplanted; or notify to a practitioner, that the organ is not recommended to be transplanted.

19. The system of claim 1, wherein the computing device is further configured to: compare the organ function of the organ to a threshold value; and based on the organ function being above the threshold, provide the action for the organ.

20. The system of claim 1, wherein at least one of: the organ function is a contractility and the threshold value includes a threshold contractility; or the organ function is a compliance and the threshold value includes a threshold compliance.

21. The system of claim 20, wherein at least one of: the compliance of the organ corresponds to the distance traveled of the plurality of markers across the plurality of images; or the contractility of the organ corresponds to a peak velocity of the plurality of markers across the plurality of images.

22. The system of claim 1, wherein the plurality of images are acquired while the organ is in an unloaded state.

23. The system of claim 1, wherein the plurality of images are acquired while the organ is in a partially loaded state or a fully loaded state.

24. The system of claim 1, wherein the plurality of images are synchronized.

25. The system of claim 1, wherein each image of the plurality of images is an ex vivo image of the organ.

26. The system of claim 1, wherein each image of the plurality of images is acquired during an open surgery of an organ.

27. The system of claim 1, wherein the plurality of images is a first plurality of images and wherein the computing device being further configured to receive a second plurality of images of the organ of the subject; and wherein determining the organ function is based on the second plurality of images.

28. A system comprising: an imaging device; a computing device in communication with the imaging device, the computing device being configured to: receive, from the imaging device, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ; generate a plurality of markers overlaid on the organ of each image of the plurality of images; track the position of each marker of the plurality of markers across the plurality of images; determine a function of a portion of the portion of the organ, based on the tracking of the plurality of markers; and provide an action for the portion of the organ, based on the function.

29. The system of claim 28, wherein the computing device is further configured to isolate the portion of the organ for each image of the plurality of images to generate a plurality of organ portion images; and wherein the plurality of markers are overlaid on the portion of the organ; wherein determining the function of the portion of the organ is based on the tracking of the plurality of organs of the portion of the organ.

30. The system of claim 28, wherein generating the plurality of markers overlaid on the organ includes the computing device being further configured to: generate the plurality of markers on the portion of the organ in each image of the plurality of images.

31. The system of claim 28, wherein the organ is a heart, and wherein the portion of the organ is the left ventricle or the right ventricle.

32. The system of claim 28, wherein the function is contractility and wherein the computing device is further configured to: determine a contractility value for each marker of the plurality of markers; and based on at least one contractility value for a marker of the plurality of markers exceeding a contractility threshold, determine the function of the portion of the organ.

33. The system of claim 28, wherein the function is contractility and wherein the computing device is further configured to: determine a contractility value for each marker of the plurality of markers; and identify each marker of the plurality of markers having a contractility value being below a contractility threshold; and provide the action based on the identified markers having a contractility value below the contractility threshold.

34. The system of claim 28, wherein provide the action includes the computing device being further configured to: present, on a display, an image of the organ with the identified markers.

35. The system of claim 34, wherein the identified markers are distinguished from the plurality of markers that were not identified.

36. The system of claim 33, wherein the function associated with the identified marker is at least one of: indicative of a previously damaged region of the organ that includes the identified marker; indicative of an infarction of the region of the organ that includes the identified marker; or indicative of ischemia of the region of the organ that surrounds the identified marker.

37. A system comprising: an imaging device; a computing device in communication with the imaging device, the computing device being configured to: receive, from the imaging device, a plurality of optical images of an organ of a subject; generate one or more markers overlaid on the organ of each image of the plurality of images; track the position of each marker of the one or more markers across the plurality of images; determine a function of the organ, based on the tracking of the one or more markers; and provide an action for the organ, based on the function.

38. The system of claim 37, wherein the function of the portion of the organ includes a health score of the organ.

39. The system of claim 38, wherein the computing device is further configured to: determine the action, based on the health score.

40. The system of claim 39, wherein the computing device is further configured to: determine the action, based on the health score exceeding a threshold health score; or determine the action, based on the health score being below a threshold health score.41 . A computer implemented method for evaluating organ health, the method comprising: receiving, using one or more computing devices, a plurality of images of an organ of a subject, each image of the plurality of images being acquired by light reflected off the organ; generating, using the one or more computing devices, one or more markers overlaid on the organ of each image of the plurality of images; tracking, using the one or more computing devices, the position of each marker of the one or more markers across the plurality of images; determining, using the one or more computing devices, a function of at least a portion of the organ, based on the tracking of the one or more markers; and providing, using the one or more computing devices, an action for at least the portion of the organ, based on the function.

42. The method of claim 41, further comprising: determining, using the one or more computing devices, a contractility value of the at least a portion of the organ, based on the tracking of the one or more markers; and determining, using the one or more computing devices, the function based on the contractility value exceeding or being below a contractility threshold.

43. The method of claim 41, further comprising: determining, using the one or more computing devices, a compliance value of the at least a portion of the organ, based on the tracking of the one or more markers; and determining, using the one or more computing devices, the function based on the compliance value exceeding or being below a compliance threshold.

Images