Bedside noninvasive imaging test to assess kidney function

Abstract

The present teachings include methods of predicting kidney transplant outcomes, monitoring kidney transplant outcomes, and identifying a kidney for transplantation using non-contrast ultrasound (US) imaging.

Description

[0001] BEDSIDE NONINVASIVE IMAGING TEST TO ASSESS KIDNEY FUNCTION CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit under 35 U.S.C. § 119(e) of U.S.

[0003] Provisional Patent Application No. 63 / 776,684, filed March 24, 2025, the entire disclosure of which is incorporated herein by reference.

[0004] GOVERNMENT SUPPORT CLAUSE

[0005] This invention was made with government support under DK128851 awarded by the National Institutes of Health. The government has certain rights in the invention.

[0006] FIELD

[0007] The present disclosure generally relates to methods for imaging and assessing kidney function.

[0008] BACKGROUND

[0009] Transplant kidneys are a life-saving but limited resource. Many donated kidneys are unfortunately deemed “poor quality” and are discarded when we do not know if they could be used to improve or save lives. This is because we do not have tools to predict the health of these kidneys. To solve this problem, the development of new imaging tools to predict which kidneys can be used for transplant is necessary. If successful, such methods will provide new tools to better match kidneys to patients in need.

[0010] The number of patients in need of a kidney is three times higher than the number available for transplant. This imbalance directly contributes to the death of nearly 5000 patients waiting for a kidney transplant each year (Collins et al., Kidney International Supplements, 2015; 5: 2-7). Astonishingly, -2500 kidneys are deemed “poor quality” and are discarded despite evidence that even “lower quality” kidneys can confer a survival benefit over continued dialysis (Hart et al., Am J Transplant 2019. 19 Suppl 2, 19-123; Moeckli et al., Transpl Int, 2019, 32, 459-469; and Stewart et al., Clin J Am Soc Nephrol, 2017, 12, 2063-2065). This disclosure addresses the widely recognized need to increase the number of available allografts for patients who require a kidney transplant. The United StatesDepartment of Health and Human Services’ Advancing American Kidney Health Initiative has mandated quality metrics be identified to encourage the use of “less than perfect” deceased donor kidneys, a viable option for patients on the waitlist. Currently, the results of the donor kidney biopsy are the most common reason to discard a kidney. However, specific pathologies in deceased donor kidneys, such as glomerulosclerosis and vascular damage, cannot be fully measured from a biopsy, despite the significant impact these metrics may have on transplant outcomes (Kaths et al., Transplantation, 2016, 100, 1862-1870 and Kaths etal., Am J Transplant, 2017, 17, 2580-2590). The kidney maintains glomerular filtration and osmotic regulation of the blood, despite constant and potentially damaging natural fluctuations in systemic blood pressure. There are several identified mechanisms of autoregulation of local blood pressure and flow, including a myogenic (MYO) response and tubuloglomerular

[0011] feedback (TGF) (Bayliss et al., J Physiol, 1902, 28, 220-231 ; Gilmore et al., Circ Res, 2018, 47, 226-230; Ploth et al., Kidney Int, 1977, 12, 253-267; and Ploth et al., Am J Physiol-Ren Physiol, 1978, 235, F156-F162). The MYO response arises from passive modulation of arterial diameter in response to systemic blood pressure fluctuations, protecting the downstream glomerular capillaries from barotrauma (Bayliss et al., J Physiol, 1902, 28, 220-231). The TGF mechanism arises from signaling by the macula densa to modulate glomerular arteriolar diameter in response to changes in NaCI concentration in the distal convoluted tubule. These two mechanisms become dysfunctional in pathologies associated with diabetic and hypertensive nephropathy, due in part to glomerular and tubular injury from uncontrolled fluctuations in pressure (Mattias et al., Physiol Rev, 2015, 95, 405- 511; Christensen et al., Kidney Int, 1997, 52, 1369-1374; Christensen et al., Kidney Int, 1999, 56, 1517-1523; and Haymann et al., Sci Rep, 2021, 11, 11682). Each mechanism is associated with distinct, but spatially variable bands of low-frequency fluctuations in perfusion rates and arterial diameters (Niels-Henrik et al., Am J Physiol-Ren Physiol, 2011 , 300, F319-F329 and Dmitry et al., Elife, 2022, 11, e75284). The MYO response causes fluctuations from 0.1-0.3 Hz, and the TGF from 0.02-0.05 Hz. Currently, there are no clinically translatable tools to measure and map autoregulation in the kidney. These fluctuations are intrinsic to the kidney, and reflect the functional capacity of the kidney because they are driven by vascular tone and nephron function, independent of sympathetic input.It has recently been demonstrated that autoregulatory mechanisms can be mapped using noninvasive imaging by measuring spontaneous physiological fluctuations in the kidney (Bennett, National Institutes of Health, 2024, https: / / www.niddk.nih.gov / news / meetings-workshops / 2024 / reimagining-kidney-function-assessment-workshop and Baldelomar et al., AJP Renal Physiol, 2024, 327(1), F113-F127). The technique is based on simply observing changes in image features caused by these fluctuations between repeated images (e.g., MRI or ultrasound). The images are acquired such that the bulk motion of the kidney is easily corrected with post-processing. The time- series in each location in the images are then processed to produce a frequency power spectrum. This power spectrum is spatially variable, and specific peaks in the spectra occur at the frequencies associated with distinct mechanisms of autoregulation. The distributions of features of these spectra can then be compared to measure kidney functional parameters, including loss of vascular tone or perfusion, or loss of single nephron autoregulation. This approach requires no contrast or breath hold, and the images are acquired over a period of 10 minutes, making the technique highly translatable. It is therefore hypothesized that ultrasound (US) imaging can be used in the setting of the deceased kidney donor to sensitively measure these low-frequency physiological fluctuations as a sensitive measure of kidney health.

[0012] Therefore this disclosure aims to develop functional biomarkers from ultrasound (US) images in deceased donors to evaluate kidney health. The proposed work provides a translational tool to assess kidney functional capacity, aimed at accurately identifying kidneys that can be confidently allocated for transplantation.

[0013] Furthermore, the focus of this work is to increase the number of available kidney allografts from deceased donors by developing sensitive markers of kidney quality. To this end, a proposed non-contrast ultrasound (US) imaging technique to improve the assessment and allocation of deceased donor kidneys is disclosed. The proposed approach is highly translational and could be implemented immediately if established. A technique using spectra from resting state US (rsUS) to measure autoregulatory function, a critical component of kidney physiology will be established. The technique is tested by performing rsUS to measure kidney function ventilated, deceased donors before transplantation. The method has two primary embodiments: 1) Establishing the use of rsUS to predict outcomes aftertransplant and 2) Identification of correlates between pathology and rslIS spectral features. The methods and techniques will provide noninvasive methods to accurately determine kidney health before transplantation.

[0014] SUMMARY

[0015] Among the various aspects of the present disclosure is the provision of methods of predicting kidney transplant outcomes by non-contrast ultrasound (US) imaging.

[0016] Among the various other aspects of the present disclosure is the provision of methods for identifying a subject for kidney transplantation by non-contrast ultrasound (US) imaging.

[0017] Among the various aspects of the present disclosure is the provision of methods of monitoring kidney transplant outcomes by non-contrast ultrasound (US) imaging.

[0018] Briefly, therefore, the present disclosure is directed to methods of predicting kidney transplant outcomes, monitoring kidney transplant outcomes, and identifying a kidney for transplantation using non-contrast ultrasound (US) imaging.

[0019] Among other various aspects of the present disclosure is the provision of methods of predicting kidney health by non-contrast ultrasound (US) imaging.

[0020] Among other various other aspects of the present disclosure is the provision of methods for identifying kidney health of a subject by non-contrast ultrasound (US) imaging.

[0021] Among other various aspects of the present disclosure is the provision of methods of monitoring kidney health in a subject by non-contrast ultrasound (US) imaging.

[0022] Among other various aspects of the present disclosure is the provision of methods to determine kidney health of a subject by non-contrast ultrasound (UA) imaging wherein kidney health is determined by measuring kidney functional capacity, kidney autoregulatory function, vascular tone, vascular perfusion, loss of single nephron autoregulation, or any combination thereof.

[0023] Briefly, therefore, the present disclosure is also directed to methods of predicting kidney health, monitoring kidney health, identifying kidney health of asubject, and determining kidney health of a subject using non-contrast ultrasound (US) imaging.

[0024] DESCRIPTION OF THE DRAWINGS

[0025] Those of skill in the art will understand that the drawings, described below, are for illustrative purposes only. The drawings are not intended to limit the scope of the present teachings in any way.

[0026] FIG. 1 A shows time series and spectra of voxels from resting state MRI of the human kidney, with spontaneous physiological fluctuations. MRI of a male volunteer. Time series of MR signal filtered between 0-0.2 Hz from an individual voxel exhibits spontaneous low frequency oscillations.

[0027] FIG. 1 B shows time series of voxels from the region of interest (ROI) shown in the MR image in FIG. 1 A with black shade indicating cortical voxels and red shade indicating voxels in medulla. Cortex and medulla were manually identified based on MRI signal contrast.

[0028] FIG. 1C shows (left) average power spectra (without error displayed for comparison) are calculated from cortex (black) and medulla (red) in the ROI. Asterisks (*) indicate selected examples of peaks with significant difference with p < 0.05 between cortex and medulla, (right) the same average power spectra from cortex (black) and medulla (red) in the ROI is shown between 0.02-0.1 Hz and with standard error shown as shaded area. Asterisks (*) indicate peaks in Bands l-ll with a significant difference with p < 0.05 between cortex and medulla.

[0029] FIG. 2A shows time series and spectra of voxels from resting state MRI of a rat kidney, acquired over 10 minutes. Schematic of a kidney and a representative MR image.

[0030] FIG. 2B shows the power spectrum calculated from the time series confirms peaks at those same frequencies in bands: Band I = 0.0125-0.05 Hz; Band II = 0.05-0.1 Hz; Band III = 0.1-0.3 Hz. Fractional total power in each band was: I-28.2%, 11-12.1%, IIIA-15.1%, IIIB-17.0%, IIIC-11.0%, IIID-16.6%.

[0031] FIG. 3A shows correlations of histopathology with total power and peak power calculated from all voxels in cortex in all animals using wavelet analysis of resting- state MRI in the rat kidney. Fixed sections stained with Periodic Acid-Schiff stain. Shown are 20x (top) and 40x (bottom) examples of glomeruli from a healthy Sprague Dawley (SD) rat (left) and a Zucker Diabetic Sprague Dawley (ZDSD) with long exposure (LE) to hyperglycemia (right).

[0032] FIG. 3B shows peak power from Band I frequencies (0.0125-0.05Hz) and Band II frequencies (0.05-0.1 Hz) was calculated in all ZDSD rats, and were plotted against a composite index for total glomerular injury. Calculated R2 was 0.76 and 0.96, respectively. One rat was an outlier in regression (+).

[0033] FIG. 3C shows total power and peak power in Band I and in Band II were each highly correlated with glomerular pathology in ZDSD rats. Scale bar, A-top -50um; A-bottom = 20um.

[0034] FIG. 4 shows resting state ultrasound (rsUS) and Doppler US of the kidney of a ventilated, deceased donor at Mid America. rsUS spectra are acquired from the cortex and medulla using image processing. Fluctuations in the doppler images are compared to the rsUS spectra. Spectra from these data were acquired in under 10 minutes.

[0035] FIG. 5 shows resting state ultrasound of a human kidney. Regions of interest (in color) are evaluated overtime as a time series. The time series from different regions are used to compute a power spectrum, which exhibits distinct peaks that reflect the frequencies of physiological fluctuations.

[0036] FIG. 6 shows power spectra from resting state ultrasound varies between subjects but many peaks in the spectra are similar. Some variability in the spectra may reflect natural differences that are also associated with physiological variables such as resting heart rate.

[0037] FIG. 7 shows peak power from measured resting state ultrasound is correlated with estimated nephron number in different human subjects.

[0038] FIG. 8 shows nephron number in donor kidneys, which can be measured using resting state ultrasound, correlates inversely with serum creatinine in recipients after transplantation.

[0039] FIG. 9 shows donor eGFR and single nephron GFR do not appear to correlate with recipient BUN or creatinine after transplantation.

[0040] FIG. 10 shows peak frequencies of resting state ultrasound power spectraare spatially variable in the kidney and may reflect spatially heterogeneous function.

[0041] DETAILED DESCRIPTION

[0042] In certain aspects, the methods of predicting and / or monitoring kidney transplant outcomes and / or identifying a kidney for transplantation comprises obtaining resting state non-contrast ultrasound (rslIS) imaging over 10 minutes in ventilated deceased donors, on both kidneys, and determining whether spectral features of autoregulation from rslIS predicts the delayed graft function (DGF) and / or serum creatinine (sCr) at three and six months post- transplant. In some aspects, machine learning will be applied to investigate whether features in the spectra, in combination with histopathology and donor data, can improve prediction of outcomes. It is hypothesized that spectral features from US in ventilated deceased donors predict DGF and / or sCr trajectory after kidney transplantation.

[0043] In one aspect, the methods of predicting and / or monitoring kidney transplant outcomes and / or identifying a kidney for transplantation comprises comparing rsUS-measured autoregulatory spectral features to interstitial fibrosis, glomerular sclerosis, glomerular features, and vascular pathology from site-directed biopsy, multi-location biopsy, and microstructural and anatomical MRI. In further aspects of the method, specific spectral bands that correlate with the biopsy features will establish the mechanistic basis of the autoregulatory features. It is hypothesized that spectral features from US in ventilated deceased donors correlate with whole kidney histopathologic scores. In some aspects, to facilitate the broader implementation of these methods, rsUS measurements by non-technical users (e.g., clinicians and nurses) will be compared to those obtained by certified technicians.

[0044] In certain aspects, the methods further comprise measuring kidney volume by determining kidney length, kidney left-right diameter, and kidney anterior-posterior diameter with non-contrast ultrasound imaging.

[0045] In other aspects, the methods further comprise determining cortical volume by segmenting kidney and cortex images. In certain aspects, the spectral feature is kidney nephron number, kidney density, or a combination thereof. In some aspects, the nephron number is determined at a spectral peak of about 0.025Hz.In some aspects, the non-contrast ultrasound (US) imaging is segmented into cortex tissue and medulla tissue based on signal intensity differences of the cortex tissue and the medulla tissue.

[0046] In another aspect, the methods further comprise measuring microvascular motion, blood flow, or a combination thereof by non-contrast ultrasound (US) imaging.

[0047] All publications, patents, patent applications, and other references cited in this application are incorporated herein by reference in their entirety for all purposes to the same extent as if each individual publication, patent, patent application or other reference was specifically and individually indicated to be incorporated by reference in its entirety for all purposes. Citation of a reference herein shall not be construed as an admission that such is prior art to the present disclosure.

[0048] Having described the present disclosure, it will be apparent that modifications, variations, and equivalent embodiments are possible without departing from the scope of the present disclosure defined in the appended claims. Furthermore, it should be appreciated that all examples in the present disclosure are provided as non-limiting examples.

[0049] EXAMPLES

[0050] The following non-limiting examples are provided to further illustrate the present disclosure. It should be appreciated by those of skill in the art that the techniques disclosed in the examples that follow represent approaches the inventors have found function well in the practice of the present disclosure, and thus can be considered to constitute examples of modes for its practice. However, those of skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments that are disclosed and still obtain a like or similar result without departing from the spirit and scope of the present disclosure.

[0051] EXAMPLE 1

[0052] The approach builds on the technique of resting-state magnetic resonance imaging (rsMRI), which has had a major impact on human neuroscience and diagnostic imaging to detect brain disease (Biswal et al., Magn Reson Med, 1995,34, 537-541 and Fox etal., Front Syst Neurosci, 2010, 4, 19). In rsMRI, MR images are acquired in rapid sequence over a period of time. The voxels (3D pixels) in the images each exhibit a magnitude measured at each time point. This magnitude is evaluated as a time series. In the brain, natural fluctuations in these time series at different locations can be used to reveal spontaneous neural activity correlated with changes in flow and oxygenation. It has recently been demonstrated that rsMRI can be applied in the kidney to detect spontaneous physiological fluctuations associated with natural processes, including autoregulation. An example of this approach is shown in FIGS. 1A-1C, where the kidney of a human subject was imaged over a period of 10 minutes and spectral features of the time series across the entire organ were extracted (Baldelomar et al., AJP Renal Physiol, 2024, 327(1), F113-F127). Several important features of the spectra resulted such as 1) there are prominent spectral peaks in voxels in cortex and medulla that correspond to frequencies of autoregulatory mechanisms established from micro puncture and doppler studies of isolated kidneys and 2) these peaks occur in bursts, and in response to oscillations in perfusion in the renal artery. This technique allows for the first time to evaluate natural physiological processes that provide sensitive, early measurements of altered physiology that indicate or predict the development of kidney pathology.

[0053] To investigate the sensitivity of autoregulatory spectra to kidney disease, rsMRI in two animal models was applied: a rat model of metabolic syndrome progressing over a period of 30 weeks to diabetes diabetic nephropathy, and a rat model of acute kidney injury (AKI) due to injection of folic acid, which damages the tubules and leads to chronic kidney disease (Baldelomar et al., AJP Renal Physiol, 2024, 327(1), F113-F127). Both of these models closely mimic human CKD progression during metabolic syndrome and diabetes and after AKI.

[0054] FIGS. 2A-2C show representative rsMRI spectra in a rat kidney in vivo, demonstrating spatially variable, natural physiological fluctuations throughout the organ. The same spectral peaks corresponding to known frequencies associated with autoregulatory mechanisms, at -50 mHz (TGF) and -150 mHz (MYO) were observed. Similar to humans, the features of the spectra vary by tissue compartments, with most prominent TGF peaks in the cortex. These oscillations are characteristic of delayed feedback systems, specifically here the feedback between either sodium and chloride concentrations in the distal tubule and afferentarteriolar perfusion (TGF), and between arterial and arteriolar diameters and arteriolar pressure.

[0055] In the model of diabetic nephropathy (Zucker diabetic Sprague Dawley), the power in the peaks tracked exposure to hyperglycemia after 30 weeks (Baldelomar et al., AJP Renal Physiol, 2024, 327(1), F113-F127). Power in Band I (10-50 mHz, TGF), was significantly lower after short exposure, but was significantly higher and exhibited multiple peaks after longer exposure. This likely demonstrates that power in these fluctuations reflects loss of autoregulation after short exposure, and hyperfiltration after longer exposure. Critically, the power in Band I was strongly correlated with severity of glomerular histopathology scores, shown in FIGS. 3A-3C. In the rat model of AKI, induced injury using folic acid was followed by longitudinal imaging, blood chemistry analysis, and direct measurements of glomerular filtration rates over 6 weeks. The first imaging session, 2 weeks after injury, demonstrated that the Band I resting state spectral power was significantly elevated in rats with AKI compared to controls. At this point in progression, during the phase of acute kidney disease (AKD) before any change in glomerular filtration rate (GFR), spectra correlated with severity of injury based on BUN / creatinine taken just after injury.

[0056] These studies have led to the breakthrough concept of applying resting state imaging to evaluate high-risk kidneys. While MRI is not currently feasible in an OPO, ultrasound (US) instruments are widely available and inexpensive. And importantly, US is extremely sensitive to blood flow and motion, both of which are modulated by the autoregulatory mechanisms seen in preliminary studies. This system and approach was tested on a deceased, ventilated donor and has established the feasibility and practicality of the experiments, data collection, and analysis. The data was easily collected within 10 minutes, with almost no donor motion. Typical images from rsUS and doppler are shown in FIG. 4, where low frequency fluctuations occurring over minutes in the images is observed, similar to those observed in MRI. These observations establish the feasibility of the studies.

[0057] These preliminary studies demonstrate that imaging can be used to detect low frequency physiological fluctuations in the kidney that reveal autoregulatory function. Autoregulation is tightly linked to both vascular health and single nephron filtration, so this approach provides a sensitive marker of loss of kidney function, likely at the earliest stages of disease. In the setting of the deceased donor kidney,sensitively detecting kidney function and functional capacity at this resolution could allow clinicians to determine which donor organs are viable for transplantation. Specifically in high- risk kidneys, this approach could be used to identify and rescue kidneys that could provide a survival benefit for certain recipients.

[0058] Thus, this disclosure proposes to apply resting state ultrasound (rslIS) to measure autoregulation in the ventilated, deceased donor, to determine whether it can be used to 1) predict outcomes after transplantation and 2) noninvasively detect microstructural pathology. Finally, the practical implementation of this approach for a non- technical user will be established. This work will provide a sensitive, noninvasive measure of kidney function in deceased donors, aimed at reducing the number of discards and increasing the number of kidneys available for transplantation.

[0059] Determine the relationship between autoregulatory function in the kidney and transplant outcomes.

[0060] Design: Whether rslIS spectral features of the kidney cortex predict delayed graft function in 100 deceased donor kidneys will be investigated. rsUS over a period of 10-20 minutes on ventilated, deceased donors will be performed. The scans will be performed between 24-48 hours of admission. Average time for donors in the ICU is 40 hours. Blood will be sampled within the first 24 hours and at the time of US, blood pressure at the time of imaging. Donor initial, peak, and terminal creatinine will be collected. Vasopressor administration history including type and dose will be recorded. Vasopressors are administered to 80% of donors upon arrival, and 35-40% are off of them within four hours. The rest are usually off within 12 hours. Mean arterial pressure is maintained at > 65 mmHG. Donors are prone, which allows access for kidney US imaging. Scanning will be performed by a clinical ultrasound technician. Exclusion criteria for this study are: Research consent body mass index < 40, ages from 10-90, and eGFR <15 or renal replacement therapy, and no history of diabetes.

[0061] Imaging will be done with the Sonosite Exporte with linear and phased array for 1-5 MHz. The kidney will be identified and the probe will be positioned so that the motion of the kidney is in plane, as in the preliminary studies. The probe will be positioned using a 3D printed holder. Imageswill be acquired in image mode and the videos stored through the DVR port and transfer via USB stick to a computerfor analysis, using each frame in the video as individual images for the spectral measurements. Both kidneyswill be scanned (-10 minutes per kidney). Both rslIS and continuous doppler US will be performed for comparison.

[0062] A prospective study will investigate the association between ultrasound spectral features in the kidneys of ventilated deceased donors and images in the kidney and transplant outcomes. Continuous doppler measurements of flow in the same regions over the same time will be evaluated. It is anticipated that approximately 90 kidney scans will be obtained over the entire project, though will increase this if more become available will be performed. Regions of interest in cortex or medulla will be measured at evenly distributed, but nonoverlapping locations (-10 cm3). The primary outcomeswill be 1) delayed graft function (DGF), defined by the need for dialysis within the first seven days following transplant, and 2) sCr and sCr T rajectory from 1 month and 6 months after transplant. Information on several important confounding variables including immunosuppressive regimen including induction medications, transplant institution, rejection episodes, cause of death, and comorbidities (diabetes, heart disease, hypertension, and underlying kidney disease) and any factors that may influence rsUS will be collected. Ideally, this study will establish ultrasound as a potential clinical tool to predict allograft outcomes, helping to inform allograft matching.

[0063] Analysis and expected outcomes

[0064] MATLAB (Ver. R2018b, The Mathworks, USA) and AFNI (Ver.

[0065] AFNI_23.0.07, National Institute of Health, USA) will be used for image coregistration and processing detailed below. Images will be co-registered to a base image selected by user, using a rigid body, least squares affine co-registration algorithm in AFNI (AFNI functions: 3dAllineate and 3dvolreg). An initial coregistration will be used to correct for large motion. A second co-registration corrected for smaller motion. The quality of co-registration will be quantified using the root mean square error (RMSE) of all images with respect to the base image (RMSE), the image ata minimum in the trajectory of motion. For this, the difference of all images to base image will first be calculated and values in all voxels then squared. The average of the squared-difference image will be calculated and the square root of the average then calculated as the RMSE. Time series will then be band-pass filtered between 0-0.3 Hz using the AFNI ‘3dTproject’ function. Kidney images will be segmented into cortex, and medulla based ondistance from the surface of the kidney. Cortex will be identified using distance to the kidney surface; 0-3.5mm. Remaining inner kidney will be considered medulla. Voxels <= 0.5mm from kidney surface will also be omitted from masks and analysis to avoid edge-associated artifacts.

[0066] Power spectral analysis will be used to investigate frequency- varying features from time series of the rsUS signal in each voxel. A developed protocol for this analysis from the preliminary studies will be used. Analysis of spectra in frequency bands consistent with expected autoregulatory mechanisms will be performed: Band I, 0.0125-0.05 Hz; Band II, 0.05-0.1 Hz; and Band III, 0.1-0.3Hz. AFNI will be used to calculate power spectra. Average power and peak power will be measured in each band, in each voxel. A “superlet”, will be used to visualize the temporal variation of spectral features. Open-source code for MATLAB will be used for superlet calculation with frequency resolution of 0.003 Hz, three wavelet cycles, and a 1x20 interval of super-resolution orders.

[0067] Machine learning models will be applied to identify spectral and doppler features that correlate with both of the outcomes. For this, several models will be labeled and trained, such as U-net, transformer models, on the spectra from a subset of -30 of the segmented image sets from major compartments (cortex and medulla). The models will then be tested on the remaining kidneys to determine whether they can predict allograft outcomes. Because deep-learning models can identify subtle features which may be difficult to see by human eye from images, it is expected that the successful models will be more accurate in predicting outcomes. From this effort the machine learning approach that provides the highest prediction of our outcomes will be identified.

[0068] Spectral features (Power in Bands 1-11, burst frequency and timing) will be compared, to both primary outcomes (DGF and Serum creatinine trajectories), alone and in aggregate using K-means clustering from PCA of the features. It is expected that spectral features that correlate, alone or in combination, with primary outcomes will be identified. Logistic regression will be used to assess the association of delayed graft function and power spectra in band I in cortex. It is expected that these will correlate more significantly with outcomes than doppler, because doppler is mainly sensitive to larger vessels. Both male and female outcomes will be examined separately. The multivariate analysis will consider other donor and recipient parameters, taken during the stay or from UNOS / SRTR.Determine the relationship between ultrasound temporal spectral features to pathology in deceased donor kidneys

[0069] Discards are common in kidney transplantation, and it is expected that -15% of the kidneys imaged will be discarded. This presents an opportunity to investigate the relationship between the measured spectral features and histopathology at the same locations. An understanding of this relationship is critical to establishing the spectral features as a noninvasive biomarker of kidney function.

[0070] Once the kidney is declined for transplantation, site-directed biopsy using the stereotactic system will be performed. The donor’s BP is maintained with a MAP >65 mm Hg in the ICU and the OR up until the time of cross-clamp. At that time the kidney is perfused in situ with cold perfusion solution UW / SPS or HTK (per surgeons’ performance). The kidney is either placed on ice or it is put on pulsatile perfusion pump. It will not be excluded on this basis but will include pumping in the analysis. Organs are only obtained for research when they cannot be used for transplant. Kidneyswill be flushed and cold-stored, as reported in previous studies (Morozov et al., J Am Soc Nephrol, 2022, 33(1), 39-48 and Charlton et al., Am J Physiol Renal Physiol, 2021, 321(3), F293-F304).

[0071] To compare US spectra with regions of pathology, site-directed biopsy using the Virtual Histopathology system will be performed. Virtual Histopathology interfaces with coordinates mapped by a fiducial grid, landmarked to the borders of the kidney identified in US. Standardized and pathology-targeted biopsies will be taken at identified stereotactic coordinates using a Kopf stereotactic system holding a Bard® Monopty® Disposable Core Biopsy Instrument (18 G needle, 22mm penetration depth) to collect: 1) two biopsies from cortex associated with each pyramidal (estimating n=8-12 locations) and 2) four biopsies in each region with reduced glomerular density, glomerular hypertrophy. Each of the derived histopathologic scores will be used in the analysis. This study will identify mechanisms of spectral features and their relationship to pathology, establishing the correlation between functional and histopathologic biomarkers in the kidney.

[0072] Kidney pathology will be evaluated using both histopathology. In each biopsy, the number of glomeruli with glomerulosclerosis will be counted using Period acid Schiff and measure the area of interstitial fibrosis (%) using Masson’s trichrome. The proximal tubules (PT) will be highlighted with a staining procedurefor Lotus tetragonolobus lectin in order to obtain PT volume fraction, and luminal size. Each kidney section will be scanned using a slide scanner (Grundium, Finland) and images will be imported into Amira (Thermo Fisher Scientific, Waltham, MA, USA) for analysis. CF labeling and location will be determined using standard immunohistochemistry methods. Each slide will be examined by a renal pathologist blinded to the rsUS data. A software-based system of virtual histopathology (VHP) has been developed to create an atlas of the human kidney in health and disease. The VHP system comprises a robust, custom workflow for image rendering, and analysis that allows standardized quantitation of the kidney micro structure based on imaging. Coregistration is accurate to ~1.5x image resolution.

[0073] Spectral features that correlate with histopathology scores, either locally from site-the directed biopsy or globally (averaged over the whole kidney) will be identified. Histopathology will include the pathologic features (glomerulosclerosis, interstitial fibrosis tubular atrophy, and arteriosclerosis) from the biopsies across the whole kidney.

[0074] Each of the spectral markers, along with clinical features, will be used to develop a predictive model of pathology. Clinical data will include KDPI, the individual components of the KDPI (age, height, weight, ethnicity, HTN, diabetes, cause of death, serum creatinine, hepatitis C and donation after circulatory death status), sex, and cold ischemia time, and kidney volume from US. This will be accomplished with unsupervised machine learning with principal component analysis (PCA), to investigate how well demographic, rsUS, and histologic features can distinguish heterogenous kidney pathology. K-means clustering analysis of the PCs will be used to reveal the statistical difference between groups based on combinations of metrics. These predictive classification models will be implemented to systematically evaluate rsUS data, patient demographics, and histology to predict kidney quality.

[0075] Machine learning will be used to identify spectral features that correlate with the individual or combined pathologic features. Several models will be labeled and trained on the spectra from a subset of ~30 of the segmented image sets from major compartments (cortex and medulla), and then test the models on the remaining kidneys to determine whether they can predict allograft outcomes. This will allow for selecting the model that provides the most predictive power. It isexpected that the rsUS features will more strongly correlate with histopathologic scores than doppler, because doppler mainly reflects larger vascular perfusion. Composite or individual markers will be compared using Pearson correlation coefficient for continuous parameters, rank correlation for ordinal parameters and ANOVA for categorical parameters.

[0076] Establish the guidelines for practical use of ultrasound- detected autoregulation For wider dissemination of this approach, it is critical to establish the use of this technique in the general OPO setting, without the need for trained US technicians. A second scan in a subset of 30 donors using the same protocol, will be performed. The spectral features measured during that scan to those acquired by the US technicians will be directly compared. The % accuracy of measurements by the different users using ANOVA for each of the variables, with p<0.05 considered significant will be determined. Success will be determined by 90% accuracy.

Claims

WHAT IS CLAIMED IS:

1. A method of predicting kidney transplantation outcome in a subject, the method comprising:a. obtaining resting state non-contrast ultrasound (US) imaging over 10 minutes on a kidney or kidneys of the subject; andb. determining the presence of a spectral feature of autoregulation from the resting state non-contrast ultrasound imaging, wherein the spectral features predict delayed graft function (DGF), serum creatinine (sCr), or a combination thereof.

2. The method of claim 1 , further comprising measuring kidney volume by determining kidney length, kidney left-right diameter, and kidney anterior-posterior diameter with non-contrast ultrasound imaging.

3. The method of claim 1 , further comprising determining cortical volume by segmenting kidney and cortex images.

4. The method of claim 1 , wherein the spectral feature is kidney nephron number, kidney density, or a combination thereof.

5. The method of claim 4, wherein the nephron number is determined at a spectral peak of about 0.025Hz.

6. The method of claim 1 , wherein the non-contrast ultrasound (US) imaging is segmented into cortex tissue and medulla tissue based on signal intensity differences of the cortex tissue and the medulla tissue.

7. The method of claim 1 , further comprising measuring microvascular motion, blood flow, or a combination thereof by non-contrast ultrasound (US) imaging.

8. A method for identifying a kidney for transplantation, the method comprising:a. obtaining resting state non-contrast ultrasound (US) imaging over 10 minutes on a kidney or kidneys of the subject; andb. determining the presence of a spectral feature of autoregulation from the resting state non-contrast ultrasound imaging, wherein the spectral features identify delayed graft function (DGF), serum creatinine (sCr), ora combination thereof.

9. The method of claim 8, further comprising measuring kidney volume by determining kidney length, kidney left-right diameter, and kidney anterior-posterior diameter with non-contrast ultrasound imaging.

10. The method of claim 8, further comprising determining cortical volume by segmenting kidney and cortex images.

11. The method of claim 8, wherein the spectral feature is kidney nephron number, kidney density, or a combination thereof.

12. The method of claim 11 , wherein the nephron number is determined at a spectral peak of about 0.025Hz.

13. The method of claim 8, wherein the non-contrast ultrasound (US) imaging is segmented into cortex tissue and medulla tissue based on signal intensity differences of the cortex tissue and the medulla tissue.

14. The method of claim 8, further comprising measuring microvascular motion, blood flow, or a combination thereof by non-contrast ultrasound (US) imaging.

15. A method of monitoring kidney transplantation outcome in a subject, the method comprising:a. obtaining resting state non-contrast ultrasound (US) imaging over 10 minutes on a kidney or kidneys of the subject; andb. determining the presence of a spectral feature of autoregulation from the resting state non-contrast ultrasound imaging, wherein the spectral featuresidentify delayed graft function (DGF), serum creatinine (sCr), ora combination thereof.

16. The method of claim 15, further comprising measuring kidney volume by determining kidney length, kidney left-right diameter, and kidney anterior-posterior diameter with non-contrast ultrasound imaging.

17. The method of claim 15, further comprising determining cortical volume by segmenting kidney and cortex images.

18. The method of claim 15, wherein the spectral feature is kidney nephron number, kidney density, or a combination thereof.

19. The method of claim 18, wherein the nephron number is determined at a spectral peak of about 0.025Hz.

20. The method of claim 15, wherein the non-contrast ultrasound (US) imaging is segmented into cortex tissue and medulla tissue based on signal intensity differences of the cortex tissue and the medulla tissue.

21. The method of claim 15, further comprising measuring microvascular motion, blood flow, or a combination thereof by non-contrast ultrasound (US) imaging.

22. A method of determining kidney health of a subject by non-contrast ultrasound (UA) imaging, the method comprising:a. obtaining resting state non-contrast ultrasound (US) imaging over 10 minutes on a kidney or kidneys of the subject; andb. determining the presence of a spectral feature from the resting state non-contrast ultrasound imaging, wherein the spectral feature identifies kidney functional capacity, kidney autoregulatory function, vascular tone, vascular perfusion, loss of single nephron autoregulation, ora combination thereof.

Images