Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Aug 1.
Published in final edited form as: Kidney Int. 2020 Mar 3;98(2):355–365. doi: 10.1016/j.kint.2020.02.011

Ultrasound super-resolution imaging provides a noninvasive assessment of renal microvasculature changes during mouse acute kidney injury

Qiyang Chen 1,2,†, Jaesok Yu 1,2,†,§, Brittney M Rush 3, Sean D Stocker 3, Roderick J Tan 3,*, Kang Kim 1,2,4,5,6,*
PMCID: PMC7387159  NIHMSID: NIHMS1570725  PMID: 32600826

Abstract

Acute kidney injury (AKI) is a risk factor for the development of chronic kidney disease (CKD). One mechanism for this phenomenon is renal microvascular rarefaction and subsequent chronic impairment in perfusion. However, diagnostic tools to monitor the renal microvasculature in a noninvasive and quantitative manner are still lacking. Ultrasound super-resolution imaging is an emerging technology that can identify microvessels with unprecedented resolution. Here, we applied this imaging technique to identify microvessels in the unilateral ischemia-reperfusion injury mouse model of AKI-to-CKD progression in vivo. Kidneys from 21 and 42 day post- ischemia-reperfusion injury, the contralateral uninjured kidneys, and kidneys from sham-operated mice were examined by ultrasound super-resolution and histology. Renal microvessels were successfully identified by this imaging modality with a resolution down to 32 μm. Renal fibrosis was observed in all kidneys with ischemia-reperfusion injury and was associated with a significant reduction in kidney size, cortical thickness, relative blood volume, and microvascular density as assessed by this imaging. Tortuosity of the cortical microvasculature was also significantly increased at 42 days compared to sham. These vessel density measurements correlated significantly with CD31 immunohistochemistry (R2=0.77). Thus, ultrasound super-resolution imaging provides unprecedented resolution and is capable of noninvasive quantification of renal vasculature changes associated with AKI-to-CKD progression in mice. Hence, this technique could be a promising diagnostic tool for monitoring progressive kidney disease.

Keywords: acute kidney injury, chronic kidney disease, fibrosis, microvascular rarefaction, diagnostic imaging, ultrasound super-resolution

Graphical Abstract

graphic file with name nihms-1570725-f0001.jpg

Introduction

Acute kidney injury (AKI) is a rapid loss of renal function occurring in up to 20% of hospitalized patients 1. The presence of AKI is associated with both increased immediate hospital mortality and the long-term development of permanent or chronic kidney disease (CKD) and eventually end-stage renal disease. It is well accepted that AKI can lead to CKD and this risk is affected by AKI severity, duration, and frequency as well as age and presence of CKD or other comorbid conditions 2–6. One proposed mechanism for AKI-to-CKD progression is the deterioration of the renal microvasculature after initial injury, a process known as microvascular rarefaction. Loss of microvascular density, particularly in the peritubular capillaries, has been described in a variety of studies of ischemic AKI 7–11, as well as in human kidneys that undergo ischemia in the process of transplantation 12. The mechanisms of microvascular rarefaction are still under investigation, but studies have implicated endothelial dysfunction and/or apoptosis, as well as altered vascular endothelial growth factor (VEGF) secretion 8. The loss of the microvasculature would lead to the impairment of renal perfusion and therefore a predisposition for acute and chronic ischemic injury 7,13–15. However, the diagnostic tools that enable noninvasive and quantitative monitoring of renal microvascular changes during AKI-to-CKD progression are still lacking.

Several diagnostic imaging technologies, including magnetic resonance imaging (MRI) 16,17, micro-computed tomography imaging (μCT) 18,19, and ultrasound (US) imaging 20–22, have been employed in pre-clinical studies to noninvasively and quantitatively evaluate the changes of renal microvasculature and renal perfusion for predicting and monitoring progressive kidney disease. However, MRI and CT scans have important limitations when used in humans. MRI requires long imaging times and the gadolinium-based contrast has been associated with development of nephrogenic systemic fibrosis, a debilitating disease 23. Meanwhile, the risk of contrast-induced nephropathy and exposure to radiation limits the widespread or repeated use of enhanced CT scans in patients with kidney disease 24–27. In addition, these imaging systems are bulky and costly, especially when used for serial imaging.

US imaging has the advantage of safety, noninvasiveness, portability, affordability and ease of use. Several approaches, such as Doppler US imaging 22 and contrast-enhanced ultrasound (CEU) imaging 20,21, have been explored to diagnose AKI-to-CKD progression in animals and humans. However, neither technique provides spatial resolution high enough for assessing microvessels, especially in the cortex. This is mainly because of the insufficient sensitivity and the acoustic diffraction limit of the operating US frequency. Ultrasound super-resolution (USR) imaging 28–37, is an emerging technology that can achieve a high spatial resolution of vasculature beyond the acoustic diffraction limit 28. The unprecedented spatial resolution is accomplished by combining the use of US contrast agents that enhance hyperechoic contrast, ultrafast frame rate imaging 38, advanced clutter filtering technique that extracts signals only from microbubbles circulating in the vessels 39, and novel center localization techniques that pinpoint the center of each microbubble from the extracted signals 29,33,34,40,41. Thus, the individual microvessels can be identified with unmatched spatial resolution up to one third of the wavelength of the US waves 28, significantly outperforming conventional US approaches. In previous studies, USR imaging technology has been successfully tested in vivo for imaging microvessels in the rat brain 29, rat kidney 42, rabbit lymph node 43, mouse ear 32, and the vasa vasorum in a rabbit atherosclerotic plaque model 33, for purpose of demonstrating the technical capability. Unlike iodinated CT contrast, US contrast agents are not nephrotoxic and are generally safe when given systemically 44–47. They are already widely used clinically for left ventricular opacification in cardiac echocardiography 46. In combining high spatial resolution with the benefits of traditional US, USR has the potential to accurately assess kidney vasculature without harming the subject.

In this study, we applied USR imaging technology with high spatio-temporal resolution in a mouse ischemia-reperfusion (IRI) kidney injury model to demonstrate its ability to quantitatively evaluate the microvasculature during AKI-to-CKD progression. US scan was performed in vivo on kidneys at 21 and 42 days after IRI, uninjured contralateral kidneys, as well as sham kidneys. We quantitatively assessed the kidneys for changes in size, relative blood volume (rBV), vessel density, and vessel tortuosity, based on the reconstructed USR kidney images. The results from US assessment on vessel density was correlated with histological analysis as a gold standard for validation.

Results

IRI leads to the development of renal fibrosis

Figure 1 presents the experimental design of our study including groups, experimental protocol and analysis. Kidneys subjected to IRI in our mouse model exhibited extensive interstitial fibrosis at both 21 and 42 days after injury. Masson’s Trichrome stain demonstrated large areas of fibrosis only in affected kidneys. Picrosirius red staining confirmed these results, and when these slides were viewed under polarized light, collagen fibers specifically exhibited birefringence, confirming deposition of extracellular matrix (Figure 2A). Although there was a significant increase in fibrosis in the IRI kidneys over contralateral kidneys, we did not find any differences between the 21 day and 42 day IRI kidneys (Figure 2B). This suggests that fibrotic development, as assessed by our methods, reaches a plateau in our injury model, with timing that is in agreement with previously published reports 18. We confirmed these results with quantitative real-time PCR showing that collagen mRNA transcripts are enhanced to similar levels in the 21 and 42 day kidneys (Figure 2C). Consistent with previous reports 48,49, we found that there was a decrease in vascular endothelial growth factor (VEGF) mRNA in all injured kidneys (Figure 2D).

Figure 1. Experimental design for in vivo ultrasound super-resolution imaging on mouse acute kidney injury model.

Figure 1.

Open in a new tab

(A) Timeline of the experiment (B) Ischemia-reperfusion injury (IRI) is performed on the right kidney to induce the acute kidney injury (AKI). The sham, contralateral, and injured kidneys at 21-days and injured kidneys at 42-days post injury (n=5 for each group) were scanned by ultrasound in long axis for ultrasound super-resolution (USR) images. The kidneys were excised immediately after the scan for fibrosis analysis and immunohistochemistry with CD31 staining focusing on renal vasculature around the corticomedullary junction. (C) From the reconstructed USR images of the kidneys, kidney changes including area, vessel density, and microvasculature tortuosity were assessed.

Figure 2. IRI leads to renal fibrosis.

Figure 2.

Open in a new tab

Mice were subjected to unilateral IRI and both the affected and contralateral kidneys were recovered at either 21 or 42 days after injury. (A) Fibrosis was detected with Masson’s Trichrome stain (blue staining) as well as picrosirius red stain (dark red). When the picrosirius slides were viewed under polarized light, birefringence denotes specific staining for collagens. (B) Fibrosis scoring of kidneys. (C-D) mRNA levels of collagen (Col3) are dramatically elevated in all injured kidneys compared to the contralateral kidneys. mRNA levels of VEGF are dramatically decreased in all injured kidneys. Data are expressed as mean ± standard error. * P < 0.0001 compared to contralateral kidneys, no differences were found between the 21 and 42 day injured kidneys.

In vivo USR imaging enables a qualitative assessment of the overall changes of the mouse kidney with IRI

Figure 3 depicts the overlaid B-mode and USR images of sham kidneys (Column A, n = 5), contralateral uninjured kidneys (Column B, n=5), and the injured kidneys at 21 days (Column C, n = 5) and 42 days (Column D, n = 5) after IRI. The 2-D cross-sectional US scan was aimed at the center plane of the kidneys in the long axis. The anatomical contour of the kidneys was shown from the B-mode layer in gray-scale in the background, and the vascular tree was identified from the overlaid USR images in hot color scale with high spatial resolution resolving microvessels down to 32 μm. The major renal vessel branches, aorta, cortex, medulla, and dorsal skin were marked by white arrows in the image of the injured kidney from mouse number 6. Although there was some interanimal variability, an overall decrease in kidney size and cortical thickness after IRI was observed. Of note, obvious rarefaction of the renal vasculature was consistently identified in the injured kidneys, compared to sham and contralateral.

Figure 3. Overlaid B-mode and Ultrasound super-resolution (B-USR) images for sham kidneys, contralateral kidneys, injured kidneys at 21-days post injury, and injured kidneys at 42-days post injury.

Figure 3.

Open in a new tab

Column A shows the USR images of the five sham kidneys. Column B shows the contralateral kidneys. Column C and column D show the USR images of injured kidney scanned at 21 days and 42 days after injury, respectively. To provide an anatomical landmark, the major renal vessel branches, aorta, cortex, medulla, and dorsal skin were marked by white arrows in the image of the injured kidney from the mouse number 6. Overall decrease in size and increase in vasculature rarefaction were observed over time.

In vivo USR imaging enables a quantitative assessment of the changes in overall morphology and renal perfusion of the mouse kidney with IRI

Figure 4A plots the average of the cross-sectional area in the long axis of the kidney measured from US B-mode image, which demonstrated a significant decrease in size in injured kidneys of 21 days (37.54 ± 1.55 mm2) and 42 days (35.67 ± 2.41 mm2) compared to sham (49.20 ± 1.58 mm2). This agreed with the measured kidney weights (Figure 4B), which decreased from 160.50 ± 4.83 g (sham) to 96.94 ± 6.22 g (21 days) and 80.70 ± 6.14 g (42 days). As expected, the weight of the contralateral kidney (197.70 ± 8.08 g) was increased compared to sham, consistent with compensatory hypertrophy. However, this change was too subtle to observe in cross-sectional ultrasound imaging. The areas of the cortex and medulla were manually identified according to the structural B-mode and overlaid vasculature images. For sham, contralateral, 21 days and 42 days post-injury kidneys, the average thickness of the cortex was estimated to be 1.76 ± 0.03 mm, 1.88 ± 0.08 mm, 1.23 ± 0.04 mm and 1.10 ± 0.08 mm respectively (Figure 4C), supporting the trend of continuous reduction in the kidney size with the progression of kidney injury. There was significant decrease of the cortical thickness by comparing 21 days (P < 0.001) and 42 days (P < 0.001) post injury kidneys to either sham or contralateral. No significant difference was found between 21 days and 42 days post injury. The vessel densities in different regions of interest, including the entire kidney, cortex and corticomedullary junction of the kidneys were quantified from the USR images. The rBV, which is defined as the percentage of the blood volume to the total organ volume 18,50, was estimated at 34.66 ± 1.99% in sham, 35.85 ± 1.38% in contralateral, 22.35 ± 1.38% in 21 days post injury, and 26.30 ± 1.92% in 42 days post injury (Figure 4D). By statistical analysis, a significant decrease between sham and 21 days (P < 0.001), and between sham and 42 days (P < 0.05) was demonstrated. The vessel density in the cortex was measured to be 64.44 ± 1.80% in sham, 66.96 ± 2.69% in contralateral, 39.77 ± 2.69% in 21 days post injury, and 46.47 ± 2.47% in 42 days post injury (Figure 4E). A significant decrease between sham and 21 days (P<0.001), and between sham and 42 days (P<0.001) were found. Compared to 21 days, a slight nonsignificant increase of rBV and vessel density in the cortex at 42 days post injury was noted. The vessel density in the corticomedullary junction of the sham, contralateral, and in the 21 and 42 days injured kidneys were estimated to be 47.14 ± 2.41%, 49.59 ± 2.42%, 26.17 ± 1.28% and 27.60 ± 1.37%, respectively (Figure 4F). These densities were obtained by calculating the area fraction of the segmented vessels in the region of interest in USR images of the kidneys. The vessel density in the corticomedullary junction of 21 and 42 days post-injury kidneys were significantly decreased compared to sham (P<0.001) and contralateral (P<0.001), but a significant difference was not found between the two injury groups. USR demonstrated a small nonsignificant increase in vessel density between the sham and contralateral groups. These quantitative assessments support prior findings of perfusion impairment chronically after a single episode of AKI 18,20.

Figure 4. Quantitative assessment of the changes in overall morphology and renal blood volume of sham, contralateral, and IRI kidneys.

Figure 4.

Open in a new tab

(A) Kidney cross-sectional area measured from the long-axis US B-mode images. Cross-sectional areas of the sham, contralateral, 21 days post injury and 42 days post injury kidneys were 49.20 ± 1.58 mm2, 51.61 ± 3.57 mm2, 37.54 ± 1.55 mm2, 35.67 ± 2.414 mm2. A significant decrease in kidney area after IRI was observed by ultrasound measurement. (B) Weight of the sham (160.50 ± 4.83 g), contralateral (197.70 ± 8.08 g), and injured kidneys (21 days: 96.94 ± 6.22 g, 42days: 80.70 ± 6.14 g). Significant reduction in weight was found, which supports the size decrease measured by US. (C) Cortex thickness of the kidneys measured from the US images. Significant decrease of the cortex thickness of the injured kidneys (21days: 1.23 ±0.04 mm, 42 days: 1.10 ±0.08 mm) compared to sham (1.76 ±0.03 mm) and contralateral (1.88 ±0.08 mm) was found. (D) US estimation of relative blood volume (rBV). The average rBV of the sham kidneys, contralateral kidneys, and IRI kidneys at 21 days and 42 days were 34.66% ± 1.99%, 35.85% ± 1.88%, 22.35% ± 1.38%, and 26.30% ± 1.92%, respectively. Significant decrease of rBV on IRI kidney was observed. (E) Vessel density in the cortex measured by US. A significant reduction in 21 days (39.77 ± 2.69%) and 42 days (46.47 ± 2.47%) post-injury kidneys compared to sham (64.44 ± 1.80%) and contralateral (66.96 ± 2.66%) kidneys was found. (F) Vessel density in the corticomedullary junction measured by US. A significant decrease in 21 days (26.17 ± 1.28%) and 42 days (27.60 ± 1.37%) compared to sham (47.14 ± 2.41%) and contralateral (49.59 ± 2.42%) was found. (n=5, ANOVA with post-hoc Tukey HSD test, *P<0.05; **P<0.01; ***P<0.001.)

In vivo USR imaging correlates with histology in vessel density estimation

Figure 5A shows CD31 immunostaining of the corticomedullary junction area of the sham, contralateral, and IRI kidneys at 21 and 42 days, a region that is particularly susceptible to IRI 7. The mean vessel density in the corticomedullary junction from each group, determined as the fractional area of staining from a midsagittal slice of the kidney, was 12.14 ± 0.75% for sham kidneys, 15.02 ± 0.47% for contralateral, 7.56 ± 0.22% for IRI kidneys 21 days post injury, and 8.82 ± 0.73% for IRI kidneys 42 days post injury (Figure 5B). Significant reduction of vessel density through histologic assessment was demonstrated between sham and 21 days (P < 0.001), and between sham and 42 days (P < 0.01). The significant reduction was also found by comparing contralateral to 21 days (P < 0.001) and 42 days (P < 0.001). The statistical significance by comparing injury groups to sham or contralateral agrees with the US assessment in the same area shown in Figure 4F. Moreover, the vessel density of contralateral was shown to be significantly higher than the sham group (P<0.05) by histology, which suggests the compensatory increase in blood flow or remodeling induced by the injury on the other side of the kidney. Figure 5D compares the histology with US assessment of the vessel density in the corticomedullary junction on a total of 20 kidneys over four different groups. There was a significant correlation between vessel density values determined by histology and USR in the corticomedullary junction (P < 0.001, R2 = 0.77), supporting the accuracy of the microvascular density assessment by USR imaging.

Figure 5. Vessel density in the corticomedullary junction by histology and correlation with US imaging.

Figure 5.

Open in a new tab

(A) Representative CD31 staining of the sham, contralateral, and IRI kidneys. (B) Vessel density measured as positively stained area fraction of the vessels in corticomedullary junction. (Sham: 12.14 ± 0.75%, contralateral: 15.02 ± 0.47%, 21 days: 7.56 ± 0.22%, 42 days: 8.82 ± 0.73%) Injured kidneys exhibited a decrease of vessel density in the ROI compared either sham or contralateral kidney. (n=5, ANOVA with post-hoc Tukey HSD test, *P<0.05; **P<0.01; ***P<0.001.) (C) Significant correlation between the histology and USR measurement of the vessel density in the corticomedullary junction was found (P value < 0.001, correlation coefficient: 0.77). (n=20, Pearson’s correlation analysis)

In vivo USR imaging enables a quantitative assessment of the tortuosity changes in the cortical vasculature

The tortuosity of the cortical microvasculature, which is defined as a ratio of the vessel length between two nearby branching points and the linear distance between those, was also assessed for each kidney from the four groups. Figure 6A shows representative USR images of the cortical microvasculature from the sham kidney and the kidney at 42 days post-injury. It is observed that the more terminal vessels are overall more tortuous and aggregated at 42 days post-injury, while most vessels have straight branches in the sham group. Figure 6B plots the average vessel tortuosity of the sham kidneys (1.129 ± 0.016), contralateral kidneys (1.143 ± 0.008), 21 days post-injury kidneys (1.146 ± 0.017) and 42 days post-injury kidneys (1.194 ± 0.012). A significant increase of the tortuosity in 42 days post-injury kidneys was found compared to the sham group (P < 0.05).

Figure 6. Tortuosity of the sham, contralateral and IRI kidneys.

Figure 6.

Open in a new tab

(A) Representative USR images of the cortical vessels from the sham and 42-days post-injury kidneys are shown. White arrows indicate the curved and aggregated vessels from the 42-days post-injury kidney. (B) Significant increase of cortical microvasculature tortuosity is shown in the kidneys at 42-days post injury compared to sham. (n=5, ANOVA with post-hoc Tukey HSD test, *P<0.05.)

Discussion

The unilateral IRI mouse model adopted for this study is a well-established model of AKI-to-CKD progression 51. In this model, the kidney exhibits a reduction in size over time, along with perfusion impairment and microvascular rarefaction. As expected, we demonstrated that this is associated with the development of extensive renal fibrosis, as well as a reduction in VEGF which would contribute to vessel rarefaction 52. During the US scan, in order to have consistent comparisons, all mouse kidneys were imaged at the center plane in the long axis with same settings of the US system, and vessel segmentation threshold was applied at −25dB for all the images. The results show that USR imaging is able to visualize renal microvessels with very high spatial resolution. The changes in kidney size, cortex thickness, rBV, microvascular rarefaction, and tortuosity were successfully quantified from high quality USR images and demonstrate progressive pathologic changes over time. The close correlation (R2 = 0.77) between the USR and histology validates the feasibility and accuracy of USR for renal vasculature assessment.

The absolute values of the rBV and vessel density from USR were generally higher than those from the histology readings. The discrepancy could be mainly attributed to the finite transducer elevational beamwidth (or imaging slice thickness) that is around 1.6mm. The US echo signals that originate slightly off-center from the imaging plane but within the elevational beamwidth of the transducer will still be recognized by the imaging system and contribute to the signal intensity in the final reconstructed images 53–56. Thus, the rBV and vessel density can be overestimated by dividing the volumetric vascular signals within the imaging slice thickness by the 2-D area. In addition, a relatively low signal to noise ratio (SNR) in some areas of the kidney may have affected the deconvolution process during image reconstruction 57–59, resulting in the overestimation of the vessel population. The relatively low SNR may also contribute to the lower sensitivity of USR compared to histology, especially in the area with high vessel density that has already overcrowded microbubbles during imaging. Therefore, while kidney weight and vessel density measured with histology were significantly higher in the contralateral compared to sham kidneys (consistent with reactive changes during unilateral IRI), USR could not detect this difference with significance. The microvessels inside the mouse renal cortex are reported to be as small as several microns. Although the spatial resolution of USR is greatly improved compared to conventional US imaging, small arterioles and capillaries as well as venules below the spatial resolution of USR imaging cannot be identified as individual vessels. Thus, the single vessel shown in the USR images could actually represent one or several adjacent smaller vessels under 30 microns. This also might have contributed to the slight overestimation in rBV and vessel density measurement. Although these minor discrepancies exist, it does not affect our major findings in the correlation between USR and CD31 stain, and the statistical difference between the injured and sham groups. Our overall immunohistochemical staining and measurements of vascularity show a similar trend to prior published data 18. While our sham kidneys have a different percentage area than the prior published report, this is likely due to differences in thresholding used in our study. However, the same threshold was utilized for every image for consistency in analysis.

According to our tortuosity analysis, the microvascular tortuosity was significantly increased at the late time point of 42 days after IRI, which is in concordance with ex-vivo μCT results in an earlier study 18. This alteration could be a useful index for predicting disease progression. It should also be noted that microvessels with high tortuosity will aggregate together in a finite area of the kidney. These individual vessels will be more difficult to separate at the later stages of injury, which may result in the overestimation of vessel density. This could be the reason for the slight increase in cortical vessel density and rBV at 42 days compared to 21 days. It is also possible that increases in rBV and vessel density, as a ratio of percentage of blood volume to area, are due to the overall loss of kidney and cortical size at the late time point of our study. This trend of slight increase of vessel density at 42 days was also found by the histological assessment.

There are some limitations of this study. Since we aimed in this study to first validate the feasibility of USR technology in this study, the kidneys were harvested for histology after each US scan. In the future, a longitudinal monitoring of the kidney by USR would be helpful to minimize interanimal variations, and investigate the effects of drug interventions during the disease. In addition, the accumulation of the dense microbubbles in cortical blood vessels would block ultrasound energy from deeper transmission into the medulla, leading to lower intensity of medullary blood vessels and discontinuity in the USR images. This is especially prevalent in sham and contralateral kidneys that have higher cortical vessel density. A carefully selected microbubble concentration would be needed to compromise the image quality between the cortex and medulla. Moreover, there is still room for further improving the performance of USR in the future, such as 3-D US scan to achieve volumetric information, using US probe with higher frequency to enhance the spatial resolution, and advanced signal processing techniques to enhance the SNR and sensitivity of USR. Finally, the reduction in kidney size over time may have affected our quantification of vessel density and rBV, but this would have actually led to an overestimation since these measurements were adjusted to total tissue area.

In previous studies by other groups, μCT 18,19 and CEU 20 were proposed as diagnostic methods for noninvasively evaluating progressive kidney disease. In the CEU study 20, renal perfusion impairment was detected and quantified noninvasively. The kidney from IRI mouse model was scanned in the long axis and the renal cortical perfusion was measured according to video intensity, similar to our study design. However, the precise microvascular structure within the kidney, especially in the area of the medulla and cortex, cannot be visualized clearly by CEU due to its limited spatial resolution and inability to identify ROI with accuracy and consistency. For the same reasons, CEU cannot provide detailed information on vessel tortuosity. In the μCT studies 18,19, spatial resolution down to 20 μm voxel size were achieved in vivo allowing the quantification of blood volume. However, the in vivoμCT approach still suffers from radiation and contrast toxicity issues, which hinders serial monitoring in practical applications. USR for imaging renal microvascular changes has been studied before in a rat IRI model 42. However, that study was performed on limited animal samples and without histological and statistical support. In this study, we used increased number of animals and groups in order to determine statistical significance. We have also validated our technology with histologic analysis.Compared to other previous imaging methods, USR imaging provides competitive performance for noninvasive visualization and quantification of vessel characteristics. Since it is also safe and affordable, USR is a promising technology to aid in the monitoring and potential prognostication of kidney disease. For instance, in patients who sustain AKI, an assessment of the microvasculature could predict the future development of CKD. Microvascular assessment could also aid in the understanding of the vascular changes occurring over time in humans with AKI or CKD. It should be noted that US microbubbles are widely used in clinics, and have a good safety profile 44–47. Translating this technology for human use requires adaptation of the technique for the curved array ultrasound probe commonly used in clinical abdominal imaging which provides a deeper and wider field of view. While the algorithms used in this study are being continually optimized and adapted to the curved probe, it is notable that the technique described exactly as in this manuscript can be (and has been) applied with success in humans. In fact, kidney imaging in humans is in many ways easier since the size of the kidneys and their blood vessels are much larger than in mice. Studies on more human subjects including healthy volunteers and CKD patients are ongoing for translating this USR technology to human clinical use.

In conclusion, we show that USR imaging is able to identify renal microvessels with unprecedented high spatial resolution up to 32 μm in vivo and allow for quantification of the changes in kidney morphology and vasculature, including size, rBV, vessel density and tortuosity. This was demonstrated in a clinically relevant rodent model of AKI-to-CKD progression. A relatively high correlation between USR and traditional immunohistochemistry of vessel density was found with R2 = 0.76 which validates the accuracy of the quantitative assessment by USR. Therefore, USR has a great potential to provide key clinical data to aid in the care of patients with progressive kidney disease.

Method

The animal study was approved by the Institutional Animal Care and Use Committee at the University of Pittsburgh (Protocol #: 16098999).

Mouse kidney injury model

C57BL/6 mice (Jackson Laboratory cat# 000664, Bar Harbor, ME) were subjected to unilateral ischemia-reperfusion injury as previously described 51. Briefly, mice were anesthetized and placed on a heating pad to maintain a 37°C body temperature which was confirmed with continuous rectal measurements. Using a ventral incision, the right kidney and its pedicle was isolated and clamped with atraumatic surgical clips (#RS-5459, Roboz, Gaithersburg, MD) for 30 minutes. The clip was removed and incision closed. The contralateral (left) kidney was not injured and served as an internal control. CD31 and all US analysis were performed on contralateral, 21 days post-injury and 42 days post-injury kidneys, as well as the sham kidneys from mice that did not undergo surgery (n = 5 for each group).

Experiment protocol

USR imaging was performed on the sham kidneys, contralateral kidneys, injured kidneys at 21 days post-injury and 42 days post-injury (n = 5 each group) (Figure 1A). Mice were anesthetized with isoflurane (2% in 100% O2) for the placement of a microrenathane catheter (MR-025, Braintree Scientific) in the jugular vein for contrast agent injection. Longer term anesthesia was achieved with ketamine (100 mg/kg) / xylazine (10 mg/kg) for the ultrasound procedure. Paraspinal fur was shaved for US access (Figure 1B). The USR imaging sequence was operated by a programmable US system (Vantage 128, Verasonics, Kirkland, WI) equipped with a 15.6MHz linear array US probe (L22–14v, Verasonics, Kirkland, WI). Regular B-mode US imaging was used to locate the position and orientation of the US probe, so that the entire mouse kidney in the maximum longitudinal axis was viewed in the imaging plane. Then, the position and orientation of the US probe was fixed by a probe holder to ensure the stability of the probe throughout the data acquisition process. Commercial lipid-shelled microbubbles (Definity, diameter 1– 4 μm, Lantheus Medical Imaging, N. Billerica, MA), served as US contrast agents, were diluted with phosphate-buffered saline (PBS) to concentration of 10% and intravenously injected with a bolus of 0.1 mL through the jugular catheter. About ten seconds post injection, imaging data of 1000 effective frames were acquired using multi-angle US plane wave imaging with five angles (−3°, −1.5°, 0°, 1.5°, 3°) for spatial compounding at an effective frame rate of 250Hz for subsequent signal processing. The kidneys were imaged without the need for organ exteriorization or skin incision. After the US scan, the mouse was euthanized and kidneys harvested for weight measurement and histology staining.

Histology and Immunohistochemistry

Formalin-fixed, paraffin-embedded kidney tissue was sectioned at 3 μm thickness in the long axis of the kidney at the midsagittal plane, corresponding to the plane of USR imaging. Staining with Masson’s Trichrome and picrosirius red were performed according to manufacturer’s instructions (Thermo Fisher, Pittsburgh, PA). For fibrosis quantification, at least 10 high powered (40X) images were taken in the corticomedullary junction for each slide after Masson’s trichrome staining. Large vessels were avoided. A 10×13 grid was then overlaid upon each image and the grid intersections containing fibrosis were counted compared to total number of intersections. The counter was blinded to identifier and treatment group of the sample being viewed. For immunohistochemistry, sections were deparaffinized and hydrated before inactivation of endogenous peroxidases with 3% hydrogen peroxide in methanol. After blocking, slides were incubated in primary antibody against CD31 (#77699S, Cell Signaling Technology, Danvers, MA) overnight. Fluorescent anti-rabbit secondary antibody (Jackson Immunoresearch, West Grove PA) was then added for 2 hours prior to imaging on an immunofluorescence microscope. For CD31 immunostaining, images were photographed in a blinded fashion for each mouse in the study. At least 5 low-power (10X) images were obtained which cumulatively covered the entirety of the corticomedullary junction on a mid-sagittal section of the kidney. Quantitation was performed using the percent area function of Image J (NIH, Bethesda MD) using a common threshold for all images.

Quantitative, real-time reverse transcriptase PCR (qRT-PCR)

After RNA isolation with TRIzol reagent (Thermo Fisher, Pittsburgh PA) and reverse transcriptase reaction, the cDNA was run in a qRT-PCR reaction with SYBR green in a CFX Connect (Biorad, Hercules CA). Forward and reverse primer sequences were CCACGTCAGAGAGCAACATCA and TCATCTCTCCTATGTGCTGGCTTT for VEGF, AGGCAACAGTGGTTCTCCTG and GACCTCGTGCTCCAGTTAGC for collagen III, and CAGCTGAGAGGGAAATCGTG and CGTTGCCAATAGTGATGACC for β-actin.

USR signal processing procedure

The signal processing procedure after US data acquisition was implemented off-line in MATLAB (Mathworks, Natick, MA). The acquired raw radio-frequency (RF) channel data were processed through beamforming, motion compensation, clutter filter 39, Richardson-Lucy deconvolution 57–59, and frame summation to reconstruct the final USR images (described in detail in Supplementary Material).

Ultrasound Image assessment

The reconstructed 2-D cross-sectional overlaid USR and B-mode images of the long-axis mouse kidney was used for quantifying changes in kidney size, cortex thickness, rBV, vessel density at the corticomedullary junction, and tortuosity of the cortex microvasculature (Figure 1C). The contour of the kidney was manually selected according to the B-mode layer, by which the long axis area of the kidney was determined. In the USR layer, the vessels were identified by threshold-based segmentation. The cortex thickness was determined by the average distance between the outer boundary of the kidney and the corticomedullary junction according to the vasculature structure depicted in the USR layer. rBV was calculated as the ratio of vessel area inside the kidney to the area of the entire kidney. The vessel density in the area of corticomedullary junction was calculated as the ratio of vessel area inside the region of interest (ROI), which was selected manually according to the anatomical structure in the USR image, to the total area of the ROI. The analysis above was performed in MATLAB (Mathworks, Natick, MA). The tortuosity of the cortical microvasculature was assessed using ImageJ (NIH, Bethesda, MD), with plugin Skeletonize3D 60 and AnalyzeSkeleton 61,62.

Statistical analysis

All the quantitative results were presented as mean ± standard error. To evaluate the statistical significance, all the data among the four groups (sham, contralateral, 21 days post-injury, 42 days post-injury, n = 5 each) were compared using one-way analysis of variance (ANOVA) with post-hoc Tukey’s Honestly Significant Difference (HSD) Test. The correlation between US and histology quantification of the vessel density in ROI was performed using Pearson’s correlation analysis. The value of P < 0.05 was considered statistically significant. All the statistical analysis and plots are generated using GraphPad Prism 5.0 (San Diego, CA).

Supplementary Material

1

Translational Statement.

Currently, there are no noninvasive and safe approaches to monitor renal microvascular changes over time in humans with kidney disease. This research demonstrates that ultrasound super-resolution (USR) imaging achieves unprecedented resolution when assessing renal microvasculature in live mice after ischemia-reperfusion AKI. USR can assess changes in size, relative blood volume, vessel density, and vessel tortuosity. This USR methodology, with adaptation to the curved array ultrasound probe used in clinical abdominal imaging, can be readily applied to humans to assess kidney disease severity and aid in prognosis.

Acknowledgments

This work was supported by a Pilot grant from P30 DK079307 (Pittsburgh Center for Kidney Research). RJT is supported by an American Society of Nephrology Carl W. Gottschalk Research Scholar Grant, a National Kidney Foundation Edith H. Blattner Young Investigator Grant, and an American Heart Association Fellow-to-Faculty Award (13FTF 16990086).Mouse techniques were learned at the Vanderbilt Mouse Kidney Injury Workshop. This research was supported in part by the University of Pittsburgh Center for Research Computing through the resources provided.

Footnotes

Disclosure

The authors have declared that no conflict of interest exists.

Supplementary Material

Supplementary Methods. Block diagram of image reconstruction procedure and details of signal processing of ultrasound super-resolution.

Supplementary information is available on Kidney International’s website.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  • 1.Basile DP, Bonventre JV, Mehta R et al. Progression after AKI: Understanding Maladaptive Repair Processes to Predict and Identify Therapeutic Treatments. J. Am. Soc. Nephrol. 2016; 27: 687–697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Heung M, Chawla LS. Acute Kidney Injury: Gateway to Chronic Kidney Disease. Nephron Clin. Pract. 2014; 127: 30–34. [DOI] [PubMed] [Google Scholar]
  • 3.Leung KCW, Tonelli M, James MT. Chronic kidney disease following acute kidney injury—risk and outcomes. Nat. Rev. Nephrol. 2013; 9: 77–85. [DOI] [PubMed] [Google Scholar]
  • 4.Chawla LS, Kimmel PL. Acute kidney injury and chronic kidney disease: an integrated clinical syndrome. Kidney Int. 2012; 82: 516–524. [DOI] [PubMed] [Google Scholar]
  • 5.Chawla LS, Amdur RL, Amodeo S et al. The severity of acute kidney injury predicts progression to chronic kidney disease. Kidney Int. 2011; 79: 1361–1369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Chawla LS, Eggers PW, Star RA et al. Acute Kidney Injury and Chronic Kidney Disease as Interconnected Syndromes. N. Engl. J. Med. 2014; 371: 58–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Basile DP, Donohoe D, Roethe K et al. Renal ischemic injury results in permanent damage to peritubular capillaries and influences long-term function. Am. J. Physiol. Physiol. 2001; 281: F887–F899. [DOI] [PubMed] [Google Scholar]
  • 8.Basile DP, Yoder MC. Renal endothelial dysfunction in acute kidney ischemia reperfusion injury. Cardiovasc. Hematol. Disord. Drug Targets 2014; 14: 3–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Hörbelt M, Lee S-Y, Mang HE et al. Acute and chronic microvascular alterations in a mouse model of ischemic acute kidney injury. Am. J. Physiol. Physiol. 2007; 293: F688–F695. [DOI] [PubMed] [Google Scholar]
  • 10.Molitoris BA. Therapeutic translation in acute kidney injury: the epithelial/endothelial axis. J. Clin. Invest. 2014; 124: 2355–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kramann R, Tanaka M, Humphreys BD. Fluorescence Microangiography for Quantitative Assessment of Peritubular Capillary Changes after AKI in Mice. J. Am. Soc. Nephrol. 2014; 25: 1924–1931. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Steegh FMEG Gelens MACJ, Nieman FHM et al. Early Loss of Peritubular Capillaries after Kidney Transplantation. J. Am. Soc. Nephrol. 2011; 22: 1024–1029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Bohle A, Kressel G, Müller CA et al. The Pathogenesis of Chronic Renal Failure. Pathol. - Res. Pract. 1989; 185: 421–440. [DOI] [PubMed] [Google Scholar]
  • 14.Eardley KS, Kubal C, Zehnder D et al. The role of capillary density, macrophage infiltration and interstitial scarring in the pathogenesis of human chronic kidney disease. Kidney Int. 2008; 74: 495–504. [DOI] [PubMed] [Google Scholar]
  • 15.Matsumoto M, Tanaka T, Yamamoto T et al. Hypoperfusion of peritubular capillaries induces chronic hypoxia before progression of tubulointerstitial injury in a progressive model of rat glomerulonephritis. J. Am. Soc. Nephrol. 2004; 15: 1574–81. [DOI] [PubMed] [Google Scholar]
  • 16.Prowle JR, Molan MP, Hornsey E et al. Measurement of renal blood flow by phase-contrast magnetic resonance imaging during septic acute kidney injury. Crit. Care Med. 2012; 40: 1768–1776. [DOI] [PubMed] [Google Scholar]
  • 17.Inoue T, Kozawa E, Okada H et al. Noninvasive evaluation of kidney hypoxia and fibrosis using magnetic resonance imaging. J. Am. Soc. Nephrol. 2011; 22: 1429–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Ehling J, Babi kova J, Gremse F et al. Quantitative Micro-Computed Tomography Imaging of Vascular Dysfunction in Progressive Kidney Diseases. J. Am. Soc. Nephrol. 2016; 27: 520–532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Das NM, Hatsell S, Nannuru K et al. In Vivo Quantitative Microcomputed Tomographic Analysis of Vasculature and Organs in a Normal and Diseased Mouse Model. McCarty JH, ed. PLoS One 2016; 11: e0150085. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Cao W, Cui S, Yang L et al. Contrast-Enhanced Ultrasound for Assessing Renal Perfusion Impairment and Predicting Acute Kidney Injury to Chronic Kidney Disease Progression. Antioxid. Redox Signal. 2017; 27: 1397–1411. [DOI] [PubMed] [Google Scholar]
  • 21.Hull TD, Agarwal A, Hoyt K. New Ultrasound Techniques Promise Further Advances in AKI and CKD. J. Am. Soc. Nephrol. 2017; 28: 3452–3460. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Faubel S, Patel NU, Lockhart ME et al. Renal relevant radiology: use of ultrasonography in patients with AKI. Clin. J. Am. Soc. Nephrol. 2014; 9: 382–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Schieda N, Blaichman JI, Costa AF et al. Gadolinium-Based Contrast Agents in Kidney Disease: A Comprehensive Review and Clinical Practice Guideline Issued by the Canadian Association of Radiologists. Can. J. Kidney Heal. Dis. 2018; 5: 1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Andreucci M, Faga T, Pisani A et al. Acute kidney injury by radiographic contrast media: Pathogenesis and prevention. Biomed Res. Int. 2014: 1–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Susantitaphong P, Eiam-Ong S. Nonpharmacological Strategies to Prevent Contrast-Induced Acute Kidney Injury. Biomed Res. Int. 2014: 1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Braverman IM, Cowper S. Nephrogenic systemic fibrosis. F1000 Med. Rep. 2010; 2: 84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Weisbord SD, du Cheryon D. Contrast-associated acute kidney injury is a myth: No. Intensive Care Med. 2018; 44: 107–109. [DOI] [PubMed] [Google Scholar]
  • 28.Couture O, Hingot V, Heiles B et al. Ultrasound Localization Microscopy and Super-Resolution: A State of the Art. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2018; 65: 1304–1320. [DOI] [PubMed] [Google Scholar]
  • 29.Errico C, Pierre J, Pezet S et al. Ultrafast ultrasound localization microscopy for deep super-resolution vascular imaging. Nature 2015; 527: 499–502. [DOI] [PubMed] [Google Scholar]
  • 30.Desailly Y, Couture O, Fink M et al. Sono-activated ultrasound localization microscopy. Appl. Phys. Lett. 2013; 103: 174107. [Google Scholar]
  • 31.Viessmann OM, Eckersley RJ, Christensen-Jeffries K et al. Acoustic super-resolution with ultrasound and microbubbles. Phys. Med. Biol. 2013; 58: 6447–6458. [DOI] [PubMed] [Google Scholar]
  • 32.Christensen-Jeffries K, Browning RJ, Tang M-X et al. In vivo acoustic super-resolution and super-resolved velocity mapping using microbubbles. IEEE Trans. Med. Imaging 2015; 34: 433–440. [DOI] [PubMed] [Google Scholar]
  • 33.Yu J, Lavery L, Kim K. Super-resolution ultrasound imaging method for microvasculature in vivo with a high temporal accuracy. Sci. Rep. 2018; 8: 13918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Song P, Trzasko JD, Manduca A et al. Improved Super-Resolution Ultrasound Microvessel Imaging With Spatiotemporal Nonlocal Means Filtering and Bipartite Graph-Based Microbubble Tracking. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2018; 65: 149–167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Bar-zion A, Tremblay-darveau C, Solomon O et al. Fast VascularUltrasound Imaging With Enhanced Spatial Resolution and Background Rejection. IEEE Trans. Med. Imaging 2017; 36: 169–180. [DOI] [PubMed] [Google Scholar]
  • 36.Opacic T, Dencks S, Theek B et al. Motion model ultrasound localization microscopy for preclinical and clinical multiparametric tumor characterization. Nat. Commun. 2018; 9: 1527. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Lin F, Shelton SE, Espíndola D et al. 3-D Ultrasound Localization Microscopy for Identifying Microvascular Morphology Features of Tumor Angiogenesis at a Resolution Beyond the Diffraction Limit of Conventional Ultrasound. Theranostics 2017; 7: 196–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Tanter M, Fink M. Ultrafast imaging in biomedical ultrasound. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2014; 61: 102–119. [DOI] [PubMed] [Google Scholar]
  • 39.Demené C, Deffieux T, Pernot M et al. Spatiotemporal Clutter Filtering of Ultrafast Ultrasound Data Highly Increases Doppler and fUltrasound Sensitivity. IEEE Trans. Med. Imaging 2015; 34: 2271–2285. [DOI] [PubMed] [Google Scholar]
  • 40.Rust MJ, Bates M, Zhuang X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat. Methods 2006; 3: 793–796. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Betzig E, Patterson GH, Sougrat R et al. Imaging Intracellular Fluorescent Proteins at Nanometer Resolution. Science (80-. ). 2006; 313: 1642–1645. [DOI] [PubMed] [Google Scholar]
  • 42.Yang Y, He Q, Zhang H et al. Assessment of Diabetic Kidney Disease Using Ultrasound Localization Microscopy : An In Vivo Feasibility Study in Rats. 2018 IEEE Int. Ultrason. Symp 2018: 1–4. [Google Scholar]
  • 43.Zhu J, Rowland EM, Harput S et al. 3D Super-Resolution US Imaging of Rabbit Lymph Node Vasculature in Vivo by Using Microbubbles. Radiology 2019; 291: 642–650. [DOI] [PubMed] [Google Scholar]
  • 44.Jakobsen JA, Oyen R, Thomsen HS et al. Safety of ultrasound contrast agents. Eur. Radiol 2005; 15: 941–945. [DOI] [PubMed] [Google Scholar]
  • 45.Safety Haar G. and bio-effects of ultrasound contrast agents. Med. Biol. Eng. Comput. 2009; 47: 893–900. [DOI] [PubMed] [Google Scholar]
  • 46.Muskula PR, Main ML. Safety With Echocardiographic Contrast Agents. Circ. Cardiovasc. Imaging 2017; 10: e005459. [DOI] [PubMed] [Google Scholar]
  • 47.Wei K, Mulvagh SL, Carson L et al. The Safety of Definity and Optison for Ultrasound Image Enhancement: A Retrospective Analysis of 78,383 Administered Contrast Doses. J. Am. Soc. Echocardiogr 2008; 21: 1202–1206. [DOI] [PubMed] [Google Scholar]
  • 48.Basile DP, Fredrich K, Chelladurai B et al. Renal ischemia reperfusion inhibits VEGF expression and induces ADAMTS-1, a novel VEGF inhibitor. Am. J. Physiol. Physiol 2008; 294: F928–F936. [DOI] [PubMed] [Google Scholar]
  • 49.Leonard EC, Friedrich JL, Basile DP. VEGF-121 preserves renal microvessel structure and ameliorates secondary renal disease following acute kidney injury. Am. J. Physiol. Physiol 2008; 295: F1648–F1657. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Ehling J, Lammers T, Kiessling F. Non-invasive imaging for studying anti-angiogenic therapy effects. Thromb. Haemost. 2013; 109: 375–390. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Skrypnyk NI, Harris RC, de Caestecker MP. Ischemia-reperfusion model of acute kidney injury and post injury fibrosis in mice. J. Vis. Exp. 2013; 78: 50495. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Tanaka S, Tanaka T, Nangaku M. Hypoxia and Dysregulated Angiogenesis in Kidney Disease. Kidney Dis. (Basel, Switzerland) 2015; 1: 80–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Goldstein A, L B. Slice-Thickness Artif acts. J. Clin. Ultrasound 1981: 365–375. [DOI] [PubMed] [Google Scholar]
  • 54.Skolnick ML. Estimation of ultrasound beam width in the elevation (section thickness) plane. Radiology 1991; 180: 286–288. [DOI] [PubMed] [Google Scholar]
  • 55.Richard B Test Object for Measurement of Section Thickness at US. Radiology 1999; 211: 279–282. [DOI] [PubMed] [Google Scholar]
  • 56.Peikari M, Chen TK, Lasso A et al. Effects of Ultrasound Section-Thickness on Brachytherapy Needle Tip Localization Error. In: Springer, Berlin, Heidelberg, 2011, pp.299–306. [DOI] [PubMed] [Google Scholar]
  • 57.Richardson WH. Bayesian-Based Iterative Method of Image Restoration. J. Opt. Soc. Am. 1972; 62: 55–59. [Google Scholar]
  • 58.Lucy LB BL An iterative technique for the rectification of observed distributions. Astron. J. 1974; 79: 745–749. [Google Scholar]
  • 59.Biggs DSC, Andrews M. Acceleration of iterative image restoration algorithms. Appl. Opt. 1997; 36: 1766–1775. [DOI] [PubMed] [Google Scholar]
  • 60.Lee TC, Kashyap RL, Chu CN. Building Skeleton Models via 3-D Medial Surface Axis Thinning Algorithms. CVGIP Graph. Model. Image Process. 1994; 56: 462–478. [Google Scholar]
  • 61.Arganda-Carreras I, Fernández-González R, Muñoz-Barrutia A et al. 3D reconstruction of histological sections: Application to mammary gland tissue. Microsc. Res. Tech. 2010; 73: 1019–1029. [DOI] [PubMed] [Google Scholar]
  • 62.Polder G, Hovens HLE, Zweers AJ. Measuring shoot length of submerged aquatic plants using graph analysis. In: Proceedings of the ImageJ User and Developer Conference 2010,Mondorf-les-Bains Luxembourg, 2010, pp.27–29. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1570725-supplement-1.docx (497.2KB, docx)

RESOURCES