Here we first show the pipe-like structures in the DOCT (LIV and Fast-DOCT) projection image of the normal kidney (Figs. 1, 2 and 3) are renal tubules. The nephron is the microscopic structural and functional unit of the kidney and consists of two parts: renal corpuscles and renal tubules (Fig. 7a)1. The renal corpuscles contain cluster of capillaries, called the glomeruli, which are surrounded by Bowman’s capsules. The renal tubules consist of the proximal convoluted tubule (PCT), the distal convoluted tubule (DCT), and the loop of Henle. Several previous studies demonstrated renal tubular imaging in animal kidneys using OCT. For example, Chen et al. demonstrated fine superficial renal tubular structural imaging in Munich–Wistar-strain rat kidney ex vivo using a high-resolution OCT (axial − 3.3 \(\upmu\)m, and lateral − 6 \(\upmu\)m) with a broadband laser light source with a center wavelength of 1.3 \(\upmu\)m)14. Three-dimensional morphology of the renal tubule was also illustrated in Munich-Wistar-strain rat kidney in vivo15 by the same group. In both studies, the renal tubule appeared as a lumen in the OCT cross-section and as a hypo-scattering structure in the en face image, similar to ours (Fig. 3a, b). Thus, we can conclude that the low-backscattering structures in the OCT are renal tubules.
We determined the diameter of these pipe-like structures from the OCT projection image and compared it with previous OCT measurements14,15. We found that the diameter of this structure was around 20–25 \(\upmu\)m, which is close to that described in previous studies (30–40 \(\upmu\)m). In addition, the structures observed in the DOCT projection images have a similar convoluted appearance to those observed in previous OCT studies, and also in the corresponding histology. Therefore, we believe that the pipe-like structures in the DOCT projection images are renal tubules. Note that the diameter found from the LIV image is slightly larger than that from the OCT, around 38–51 \(\upmu\)m. We will discuss the reason for this difference in a later paragraph of this “Discussion” section.
There might be some curiosity regarding why the pipe-like structures are not clearly visible in the standard OCT intensity in this study, while they were clearly visible in previous studies14,15. This can be explained by two reasons. First, the resolution of the OCT system used in this study (18.1 \(\upmu\)m laterally and 14 \(\upmu\)m axially) is lower than that of previous OCT studies of the kidney, making it difficult to visualize renal tubule structures in the OCT intensity image. Second, the current study used mouse kidney tissue, whereas previous studies used Munich-Wistar strain rat, which is known to have more-visible fine superficial renal tubules. In the future, it might be possible to visualize such renal tubular structures in the LIV as well as in the OCT with our high-resolution OCT35 or by imaging the Munich-Wistar strain rat.
We suspect that the high DOCT signal in the pipe-like structures originates from active renal tubular epithelial cells (RTECs) around the renal tubule. A schematic of renal tubule function is shown in Fig. 7. The kidney relies on the renal tubules to reabsorb most of the filtered water and solutes, including glucose and fatty acids. The renal tubule is made up of epithelial cells that are responsible for the active transport process through the renal tubule. RTECs have mitochondria that provide an excessive amount of ATP energy to glucose and fatty acid transporters, and hence RTECs are functionally active. Glucose is transported through the SGLT1/2 and GLUT1/2 transporters36. Fatty acid transport through the renal tubules by the CD36 receptors, fatty acid-binding proteins (FABP), and fatty acid transport proteins (FATP)9,37. We believe that after we sacrifice the mouse, the active transport process stopped in the renal tubule, but the RTECs around the renal tubules were still highly active. The active RTECs can cause intercellular motion resulting in speckle fluctuation, and hence could result in a high DOCT signal around the renal tubule.
We mentioned earlier that the diameter of the pipe-like convoluted structures was larger in LIV than in the OCT image. This can be explained by the previous discussion that what we are observing in the LIV image might be the functional activity of the RTECs around the renal tubule. As the high LIV signals are from the surrounding tissue around the renal tubule, it seems plausible that the diameter of these structures is higher in LIV compared to OCT. Note that, in our previous study on mouse liver29, we found similar surrounding tissue activity. Specifically, we observed high activity just beneath the liver vessel using LIV imaging, which was due to the high metabolism of the periportal and pericentral regions of the liver microvasculature.
Next, we discuss our findings in the obstructed kidney models. The anatomical pipe-like structures that were observed in the LIV of normal kidneys were almost not found in the 1-week model and were completely absent from the 2-week model. Literature suggests that obstruction of the ureter for a certain period results in renal hemodynamic and metabolic changes, followed by tubular injury, and finally apoptosis or necrosis38. During the apoptosis or necrosis process, the function of the renal tubule can gradually alter and is finally lost due to cell death. The histology of the 2-week obstructed model also suggests renal structural damage and dilation, which further highlight the process of renal injury due to obstruction. The absence of the pipe-like structures in the LIV and Fast-DOCT images in the obstructed models may therefore indicate the loss of renal tubule function due to the effect of obstruction.
Instead of any anatomical structures, we found a high-LIV superficial layer around the peri-renal cortex region in both obstructed kidney models (Figs. 4, 5). Similar results to those in Figs. 4 and 5 were obtained for all obstructed kidney model samples. The histology images of these samples did not reveal any changes around the peri-renal cortex region and appeared similar to the those of normal kidney. To the best of our knowledge, such an appearance was not reported previously by fluorescence microscopy imaging.
Literature suggests that the obstruction of the ureter for a certain period induces renal inflammation39,40. Bai et al. reported that several pro-inflammatory cytokine IL-1\(\beta\) and transforming growth factor-\(\beta\)1 (TGF-\(\beta\)1) are recruited in the renal environment during the obstruction41. Due to regulatory mechanisms involved during renal inflammation, these inflammatory cells (IL-1\(\beta\) and TGF-\(\beta\)1) can be highly metabolically active. We suspect that during obstruction, renal inflammation occurs near the peri-renal cortex region, and since the inflammatory cells are active, they can produce speckle fluctuation and result in a high LIV signal. Although we discuss a possible cause, it is not conclusive, and such a peri-renal layer appearance on the LIV image is worthy of further investigation.
In this study, we employed two different DOCT algorithms to access the slow and fast functional activity of renal tubules; one based on logarithmic intensity variance and one on complex correlation analysis of time-sequence OCT intensity. In our pilot study, we detected several vertical artifacts in the LIV cross-section of normal kidneys, similar to the OCT angiography (OCTA) projection artifacts of retinal imaging42. This artifact suggests that there might be some structure in the renal tissue that may exhibit relatively fast motion. The LIV contrast is too sensitive to visualize this fast motion in the tissue. On the other hand, our previous study also found that when the time window is small (milliseconds), LIV cannot visualize the fine high dynamics structures18. To overcome the above limitation of LIV, we employed complex-correlation-based Fast-DOCT contrast in addition to LIV to visualize structures with high dynamics, and imaging with this contrast did not exhibit any tailed artifacts.
We also note that LIV and Fast-DOCT were reconstructed from two volumes using two different scanning protocols (details are in the “Methods” section). In the future, a new scanning protocol may overcome this issue allowing both contrasts to be reconstructed from a single volume43.
Note that, out of the two DOCT contrasts employed in our study, LIV consistently reveals pipe-like structures in all normal renal tissues and remains reproducible (see the supplementary file S1). In contrast, the Fast-DOCT projection reveals pipe-like structures in some renal tissues (Fig. 3e, f) but not in others (Fig. 2e, f). This variation in Fast-DOCT led us to initially hypothesize that sequential measurements might contribute to these discrepancies. To validate this hypothesis, we measured additional four normal kidney samples at a consistent time point, 30 minutes post-sacrifice. In all normal kidney tissues, LIV continues to exhibit similar structures, consistent with Figs. 2c, d and 3c, d. In contrast, the Fast-DOCT signal revealed pipe-like structures in two samples while in the other two samples, it did not reveal these structures despite measuring at the same time point.
We further performed an analysis of the visibility of these pipe-like structures in relation to the time of sacrifice post-dissection by including all the measured normal kidney samples (Fig. 8). For this analysis, we evaluated a total of 16 kidney samples from 8 mice (6 mice of the main study and 2 mice of the additional study) and enlisted a grader to assess visibility based on the appearance of LIV and Fast-DOCT images. Here, visibility is denoted by values: 2 for clearly visible, 1 for moderately visible, and 0 for not visible. The LIV signal consistently visualizes these pipe-like structures across different time points. With regard to the Fast-DOCT signal, the pipe-like structures are clear in 3 renal tissues, moderately visible in 7 renal tissues, and not visible in 7 renal tissues. Upon plotting the visibility of the pipe-like structures (Fig. 8), it became evident that the differences in measurement time may not be the contributor to the variability in the Fast-DOCT contrast. The conclusion drawn from these findings does not support our initial hypothesis. Therefore, we reconsider our initial hypothesis to propose that this variation in the Fast-DOCT contrast in detecting pipe-like structures could be explained by high metabolic activity variation across various kidneys rather than the time difference. This variation in the Fast-DOCT contrast requires further investigation and will be explored in future communications.
The Fast-DOCT signal of normal kidney tissue (Fig. 1e, f) reveals several isolated white (hyper-Fast-DOCT) spots in both cross-sectional and en face views. Due to two reasons, we suspect that these white spots represent cross-sections of renal tubules. First, in both cross-sectional and en face views, these hyper-Fast-DOCT white spots correlated well with the locations of high-LIV spots, and such high-LIV spots formed tubular morphology as previously discussed. Second, this correlation between high-LIV and hyper-Fast-DOCT white spots is also observed in another sample (Sample 2, Fig. 3) of normal kidney tissue. In this case (sample 2), the hyper-Fast-DOCT spots form pipe-like structures. Such pipe-like structures were clearly visible in the projection image because they appeared as vertical tails in the cross-sectional Fast-DOCT signal similar to LIV. These two observations lead us to believe that these white spots might be cross-sectional segments of the renal tubule. However, to support this assumption, more direct evidence is needed, such as correlation with the histological findings.
In comparison to LIV, Fast-DOCT specifically contrasts only the regions with fast activities. Hence, it can visualize only some of the renal tubules. Our previous studies of ex vivo liver tissues29,30 and in vitro spheroid18,25 and organoid44 tissues revealed that the activity in these tissues is relatively slow (order of seconds). Because of this strong selective contrast, the interpretation of Fast-DOCT by comparing it with histology is not easy. It might be a future investigation to interpret these high-dynamic spots in Fast-DOCT and correlate them with the histological findings. One additional point is that the utility of the Fast-DOCT contrast could potentially be more helpful when directed toward the examination of in vivo tissue dynamics. In such scenarios, this contrast might enable the capture of certain fast-dynamic signals that could not be detected by the LIV signal.
Signal-to-noise ratio (SNR) plays a crucial role in the reliability and interpretability of the dynamic OCT contrast. For an accurate interpretation of the dynamic OCT signal, we should consider the regions that show high SNR, where the OCT signal is predominantly high and less affected by noise. For example, the deep tissue regions where the signal is mostly affected by noise appeared as high decorrelated signals in the Fast-DOCT contrast and green in LIV signals. These high-decorrelated signals and high-LIV signals at these regions should be considered artifacts that arise mainly due to low SNR. Note that, out of two DOCT contrasts, the Fast-DOCT signal was computed with noise-offset correction, whereas the LIV contrast was computed without any noise correction method. In the future, a proper noise model will be implemented to obtain a more accurate LIV signal less affected by the noise sources.
In addition to SNR, maintaining phase stability is also important for calculating phase-sensitive dynamic OCT contrasts like the Fast-DOCT signal. In this study, phase offset correction method31 is applied to correct phase instability due to bulk motion before generating the Fast-DOCT contrast. In addition to bulk motion, the hardware of the swept-source OCT system can also introduce phase instability. The phase error can arise due to the asynchronization between A-trigger, which triggers a single spectral acquisition, and the k-clock, which is used for clocking each acquisition point in the spectrum. The mutual shift between the A-trigger and the k-clock induces phase error and hence can corrupt the phase consistency among the volumetric complex OCT datasets. Our previous studies found that the phase remains stable and consistent among the volumetric OCT datasets in our swept-source OCT system32. In conclusion, it is important to consider phase stabilization methods to accurately compute phase-sensitive dynamic OCT contrasts. It is also noteworthy that the LIV uses only OCT intensity, and hence not sensitive to the phase noise.