Phosphotungstic Acid Staining to Visualize the Vagus Nerve Perineurium Using Micro‐CT

ABSTRACT

Background and Purpose

Peripheral nerve stimulation is approved by the US Food and Drug Administration for treating various disorders, but it is often limited by side effects, highlighting the need for a clear understanding of fascicular and fiber organization to design selective therapies. Micro‐CT imaging of contrast‐stained nerves enables the visualization of tissue microstructures, such as the fascicular perineurium and vasculature. In this work, we evaluated phosphotungstic acid (PTA) as a contrast agent and assessed its compatibility with downstream histology.

Methods

Human vagus nerve samples were collected from three embalmed cadavers and subjected to three different staining methods, followed by micro‐CT imaging: Lugol's iodine, osmium tetroxide, and PTA. Contrast ratios of adjacent tissue microstructures (perineurium, interfascicular epineurium, and fascicle) were quantified for each stain and compared. We further developed a pipeline to optimize micro‐CT scan acquisition parameters based on objective metrics for sharpness, noise, and pixel saturation. The PTA‐stained samples underwent subsequent histological processing and staining with hematoxylin and eosin, Masson's trichrome, and immunohistochemistry and were assessed for tissue degradation.

Results

PTA enhanced the visualization of perineurium, providing high contrast ratios compared to iodine and osmium tetroxide. Optimized scanning parameters for PTA‐stained nerves (55 kV and 109 µA) effectively balanced noise and sharpness. While we found that PTA is generally nondestructive for downstream histology, higher concentrations and longer exposure could alter the optical density of nuclei and affect stain differentiation in special stains.

Conclusion

PTA serves as a valuable micro‐CT contrast agent for nerve imaging, effective in visualizing the perineurium with minimal impact on histological integrity.

1. Introduction

Electrical peripheral nerve stimulation (PNS) is Food and Drug Administration‐approved for treating various disorders, providing an alternative to drug treatments [1]; for example, vagus nerve stimulation (VNS) is used to treat epilepsy [2], depression [3], and rheumatoid arthritis [4]. The vagus nerve contains both sensory and motor fibers, and it innervates most organs in the trunk; consequently, therapeutic stimulation effects can be limited by side effects [1, 5]. A clear understanding of fascicular and fiber organization in peripheral nerves is crucial for designing novel and selective PNS therapies [6, 7, 8].

Histological sectioning followed by light microscopy is the current standard for studying fascicular organization (i.e., nerve morphology). For example, histology has enabled the quantification of nerve diameter, fascicle diameters, and fascicle counts in rat, pig, and human vagus nerves [9, 10, 11, 12, 13]; these images were then used to implement a population of models for rat, pig, and human VNS [14]. As another example, immunohistochemistry with an antibody against choline acetyltransferase (ChAT) revealed bimodal organization in the pig cervical vagus nerve, with ChAT‐positive efferent fibers and ChAT‐negative afferent fibers on separate halves of the nerve [15]; models of pig VNS could then account for this spatial organization of functional fiber groups [16]. However, as histology only shows one cross section at a time, cost and labor are prohibitive for visualizing morphology along the length of a nerve [17, 18]. Fascicles of the human vagus nerve split or merge every 0.56 mm on average [19], the contacts of the standard implanted vagus nerve stimulator are spaced 8 mm center‐to‐center [20], and computational models of human VNS are typically 50‐mm long [14]. A histological image every 100 µm along a single 50‐mm‐long nerve sample would require 500 cross sections, and even that tremendous data collection effort still would not provide complete three‐dimensional (3D) morphology.

In contrast to 2D histological images, 3D imaging modalities also exist to visualize nerve morphology, including ultrasound, microscopy with ultraviolet surface excitation (MUSE), magnetic resonance imaging (MRI), fast neural electrical impedance tomography (FN‐EIT), and light‐sheet microscopy; however, they are limited by their resolution, field of view along the nerve's length, and/or depth of imaging from the nerve's surface. Ultrasound was used to image human median nerves following traumatic neuroma [21] and to image pig vagus nerves nondestructively [22]; however, its resolution is low even with research‐grade (i.e., higher frequency) transducers, and its field of view is limited [22, 23]. Block‐face imaging techniques like MUSE provide very high resolutions (0.9 µm), but only ∼1.5 cm of nerve length can be embedded in each block, and imaging is time‐consuming, especially when tiling multiple images to capture a large cross section [24, 25]. MRI is highly valuable in clinical settings for visualizing peripheral nerves and diagnosing disease conditions, but it still lacks in terms of acquiring high‐resolution images and is more expensive, especially for higher field strengths (e.g., >3 T) required to achieve sub‐fascicular visualization [26, 27]. Light‐sheet microscopy has also been used as an alternative for visualizing fascicle structures but requires tissue‐clearing treatments that can be challenging with nerve tissues [28, 29]. FN‐EIT is another imaging modality being developed to visualize nerve structures. In EIT, the impedance of tissue at different points is measured, which is followed by a tomographic reconstruction, whereas in FN‐EIT, the evoked activity is imaged with electrode arrays around the nerve but lacks the ability to visualize anatomy at high resolution to visualize nerve fascicles [30]. Another promising imaging modality is polarization‐sensitive optical coherence tomography, which helps visualize biological structure in tissue by detecting changes in the polarization of light, but it is still limited by penetration depth due to higher light scattering in nerve tissue [31].

Micro‐computed tomography (micro‐CT) is another 3D imaging modality that recently provided pioneering insights into the complexity of the 3D morphology of human vagus nerves [32, 33]. Micro‐CT has also been used to image rat and pig nerves [19, 32, 34]. Micro‐CT studies of peripheral nerves enable the quantification of anatomical metrics [35], provide crucial inputs to anatomically realistic computational models of PNS to analyze mechanisms of action and to design improved therapies [20], and allow the assessment of PNS electrode placement and migration [32, 36, 37]. Imaging nerve samples with micro‐CT presents a significant opportunity to visualize the perineurium, an electrically resistive tissue layer. Accurately measuring the thickness of the perineurium, particularly in three dimensions, is crucial as it significantly influences the electrical activation of nerve fibers. Enhanced precision in these measurements can greatly improve the accuracy of computational models [38].

Prior to micro‐CT imaging, excised nerves must be stained to provide contrast between different neural tissues: typically, osmium tetroxide and Lugol's iodine are used. These stains primarily bind to lipid‐rich regions in the nerve, especially myelin [32, 39]. To visualize the boundaries of the fascicles, there is a need for a micro‐CT stain that highlights collagen‐rich regions, as the fascicles are wrapped in perineurium, which is rich in collagen. Further, osmium tetroxide is incompatible with subsequent histological processing, which is an important complementary approach to micro‐CT to visualize small fascicles, perineurium, and fiber types.

As an alternative to osmium tetroxide and Lugol's iodine, phosphotungstic acid (PTA) has been used to stain soft tissues for micro‐CT imaging, such as porcine arteries [40], human atherosclerotic plaques [40], and embryos [41]. PTA preferentially binds to collagen‐rich regions, rather than lipid‐rich regions [40]. However, the performance of PTA for visualizing neural structures has not been evaluated.

Herein, we used PTA to visualize fascicular structures in peripheral nerves with micro‐CT imaging, and we confirmed its compatibility with subsequent histology. We developed a protocol for staining cadaveric human vagus nerves with PTA, followed by micro‐CT imaging. We selected micro‐CT imaging parameters (excitation voltage, current, and detector integration time) to improve sharpness and reduce noise. We quantified contrast as the intensity ratio between neural tissues (epineurium, perineurium, endoneurium) to compare performance with osmium tetroxide versus Lugol's iodine versus PTA staining. We then evaluated the compatibility of the PTA‐stained nerves with the subsequent histology and immunohistochemistry.

2. Methods

2.1. Sample Acquisition and Preparation

We collected human cervical vagus nerve samples, each 6 cm in length, from three embalmed cadavers. All specimens were harvested from de‐identified donor sources, and no protected personal health information was collected. A letter of IRB exemption (non‐human‐subjects determination) was sought and approved by the Case Western Reserve University Institutional Review Board. The samples were stored in 10% neutral buffered formalin (NBF) (Surgipath, 3800598) until ready for staining and scanning (stored for up to 6 months). Each nerve sample was cut into three 2‐cm subsections; the subsections from each sample were stained with either PTA, Lugol's iodine, or osmium tetroxide.

2.2. Micro‐CT Staining

2.2.1. Lugol's Iodine

Nerve sections were washed with 1× phosphate‐buffered saline (PBS) (Fisher BioReagents, BP399‐4) three times and submerged into 10 mL of Lugol's iodine (Sigma–Aldrich, L6146‐1L). The sample in iodine was placed on a shaker for gentle agitation for 48 h, following the protocols of Thompson et al. [32].

2.2.2. Osmium Tetroxide

Nerve sections were washed in 1× PBS three times and placed in 10 mL of 1% osmium tetroxide (Electron Microscopy Sciences, 19170) [42]. The sample in osmium tetroxide was placed on a shaker for gentle agitation for 72 h. The sample was then removed from the osmium tetroxide and dehydrated with 70% ethanol three times for 30 min with gentle agitation, followed by dehydration with 95% ethanol three times for 30 min with gentle agitation [25].

2.2.3. PTA Staining Protocol for Micro‐CT Imaging

We prepared a solution of 3% PTA by diluting a standard 10% PTA solution (Sigma–Aldrich, HT152‐250ML) with deionized (DI) water. Samples were then submerged in this solution and stained overnight. The staining time was based on previous work that performed PTA staining on embryonic tissue [43].

2.2.4. PTA Staining Protocol to Compare Its Effects on Histology

We prepared two solutions of 3% PTA, one with DI water and one with 100% ethanol. PTA at 3% with DI water was prepared by diluting three parts of standard 10% PTA solution (Sigma–Aldrich, HT152‐250ML) with seven parts of DI water (v/v). PTA at 3% solution in ethanol was prepared by mixing three parts of stock solution of 10% PTA with seven parts of 100% ethanol (v/v). The different nerve sections were submerged into the staining solution fully and placed on the shaker for gentle agitation for 8 and 48 h.

2.2.5. Imaging

After staining, each nerve sample was washed with 1× PBS three times. Due to the toxicity of osmium tetroxide, the osmium‐stained nerves were wrapped in parafilm and placed in a 9‐mm‐diameter plastic tube prior to being placed in the micro‐CT scanner (ScanCo, Wangen‐Brüttisellen, Zurich). We imaged the samples using a 0.5‐mm aluminum filter at a 10.2‐mm field of view at native resolution (3000 projections). There were n = 3 vagal nerve sections for each stain type; each sample was scanned at 16 different scan parameters (Table 1), and each scan was a stack of 314 images with a 3.3 µm isotropic voxel resolution.

TABLE 1.

Saturated pixel count for phosphotungstic acid‐stained nerve across varying scan parameters.

Scan voltage (kV)	Scan current (µA)	Integration time (ms)	Saturated pixel count
45	88	800	29,405
45	133	800	22,775
45	200	800	15,206
55	72	800	1992
55	109	800	1256
55	145	800	851
55	200	800	199
70	57	800	55
70	85	800	0
70	114	800	0
70	200	500	10
90	44	500	0
90	66	500	0
90	88	500	0
90	155	387	0
90	200	300	0

Note: The table indicates lower voltage scans cause higher pixel saturation.

2.2.6. Image Analysis

There were two types of analysis done on the acquired images. The first analysis was done to generate a line profile analysis for each of the staining modalities. The second analysis was to compare the contrast ratios of different tissue regions across the three different stains. To perform these two analyses, scan settings chosen for iodine were based on Thompson et al. [19], and for osmium, they were based on Upadhye et al. [32]. We performed a third analysis called parameter analysis (which was performed only for the PTA stain), which aimed at determining the optimal scan settings for PTA‐stained nerve based on three metrics: image sharpness, noise, and pixel saturation.

2.2.7. Line Profile Analysis

To generate the line profile for each stain, we imported the images into MATLAB (R2023, MATLAB, The MathWorks, Inc., US, http://www.mathworks.com) and drew 10 lines per image across 30 images (151 pixels in length), approximately evenly spaced around the perimeter of each fascicle, perpendicular to and centered on the perineurium (Figure 1D). The mean intensity profile was then plotted for each of the three different stains along the lines that were defined.

FIGURE 1.

Example micro‐CT images of cadaveric human vagus nerve tissues stained with phosphotungstic acid (A), Lugol's iodine (B), or osmium tetroxide (C) with scale bars of 500 µm. (D) Example image of a PTA‐stained nerve depicting lines along which intensity profiles were sampled. Each line was defined such that it spanned across the fascicle, perineurium (midpoint), and interfascicular epineurium. (E) Average (mean) line intensity profiles for the three stains.

2.2.8. Contrast Ratio Analysis

We quantified contrast as the ratio of intensity between perineurium (P) and adjacent tissues: fascicle (F) or interfascicular epineurium (IFE) (Figure 2). To obtain the ratios, we created lines of 40 pixels (i.e., 132 µm) on the images, ensuring that these lines traversed across the IFE and fascicle region such that the midpoint of the line was on the perineurium. The code to draw these lines was written such that the user manually selects the midpoint to be on the perineurium and the user also defines the angle at which the line is drawn, ensuring that the edges of the line are either on the fascicle region or on the IFE region while simultaneously making sure that the midpoint is on the perineurium. The averaged intensity of three pixels at both edges of each line was used to define the intensities for fascicle and IFE, whereas the averaged intensity of three pixels at the midpoint of the line (which is on the perineurium) was used to define the voxel intensity for perineurium. This process was repeated on 13 different images (of the 314 images per scan) per sample (for n = 3 nerve sections) for each stain. We then calculated the contrast ratios as P/F and P/IFE.

FIGURE 2.

The contrast ratios for perineurium/fascicle (P/F) and perineurium/interfascicular epineurium (P/IFE) for the three stains, showing the mean and confidence interval (CI).

2.2.9. Parameter Analysis for PTA‐Stained Nerve Tissue

We used three metrics to quantify image quality: sharpness, noise, and number of saturated pixels. Sharpness was calculated as the variance of the image gradient, noise was calculated as the standard deviation of the background (a fixed square ROI was chosen across all samples outside the nerve boundaries), and saturated pixels were counted as the number of pixels with an intensity value of 32,767. The parameter analysis was done on every 10th image (i.e., we analyzed 30 images per scan) across the 16 scans.

2.3. Histology

Both 3% PTA‐ and 10% PTA‐stained nerves underwent histological staining with hematoxylin and eosin (H&E) staining and Masson's trichrome, as well as immunohistochemistry labeling with antibodies against myelin basic protein (MBP) and neurofilament (NF). The unstained sample was acquired from a separate deidentified cadaver.

2.3.1. Tissue Processing

After micro‐CT staining imaging, excess staining PTA was washed out of the tissue samples with 1× PBS. The samples were then placed in 10% NBF for at least 24 h to ensure the tissue was fixed. After initial fixation, further tissue processing was done using a LEICA Pegasus Automated Tissue Processor (Leica Biosystems), with a protocol set to hold the samples in 10% NBF for 1 min, 70% ethanol for 5 min, and a series of increasing ethanol concentrations (15, 15, 15, and 45 min, respectively) until 100% is reached, all at ambient temperature. The processor can automatically calculate the ethanol concentrations based on the number of cassettes placed in the retort. Samples will be incubated in 100% ethanol for 30 min. To begin the process of paraffinization, three wash steps of xylenes (Avantik, RS4050) (5, 15, and 30 min, respectively) were used to clear the alcohol; the final xylene step was heated to 45°C. Lastly, samples were immersed in Surgipath paraffin (Leica, 39601006) for (5, 15, and 30 min, respectively) under vacuum at 65°C. After tissue processing, samples were embedded in paraffin (Leica, 3801360) and were sectioned on an Autocut microtome (Leica Biosystems) at a thickness of 4 µm.

2.3.2. H&E Staining

Tissue sections were placed in the oven at 70°C for 20 min to help with tissue adhesion to slides and to remove excess water before deparaffinization. Automated H&E staining was performed using an automated stainer (HistoCore SPECTRA Stainer; Leica Biosystems). The tissue sections were first deparaffinized with two xylene dewaxing steps, each lasting 2 min, followed by a 4‐min incubation in xylene. The slides were then treated with 100% alcohol for 2 min, followed by 95% alcohol for 2 min, and rinsed with tap water. Next, the slides were incubated in hematoxylin (SL401; Statlab) for 4 min, rinsed with tap water, incubated in an acid rinse (SL404) for 1 min, rinsed with tap water again, and incubated in a magnesium sulfate bluing rinse (SL403; Statlab) for 1 min. After rinsing, the slides were placed in 95% alcohol for 1 min and in eosin (SL406, Statlab) for 1 min. The slides were then incubated in 95% alcohol for 1 min, in 100% alcohol for 2 min, and subsequently in xylene for 4 min. Finally, automated coverslipping was performed (HistoCore SPECTRA CV; Leica Biosystems), and slides were left to dry before imaging.

2.3.3. Masson's Trichrome

Slides were deparaffinized as described above, and trichrome staining was conducted using an automated stainer (Artisan Link Pro; Agilent). Slides were incubated in Bouin's solution for 12 min and rinsed with 1× wash solution (AR10211; Agilent). Next, slides were placed in Weigert's hematoxylin A and Weigert's hematoxylin B solutions for 10 min, respectively. After rinsing with wash solution, Biebrich scarlet acid fuchsin was applied for 2 min. Subsequently, 2.5% PTA and 2.5% phosphomolybdic acid were applied for 12 min each. After a final quick wash, 2.5% aniline blue was applied for a few seconds. The slides were then rinsed five times in the wash solution. Slides were allowed to dry for 5 min and dipped in xylene before coverslipping.

2.3.4. Immunohistochemistry

All samples underwent 3,3'‐diaminobenzidine (DAB) staining using an automated stainer (Autostainer Link 48; Agilent) for both MBP and NF. Samples underwent deparaffinization, followed by automated antigen retrieval (AR) (DAKO PT Link; Agilent). AR was done in low‐pH (6) citrate target retrieval solution (K8004; Agilent). Slides were placed in a prewarmed 65°C retrieval solution, heated to 97°C, and held at that temperature for 20 min. The solution is then cooled to 65°C. After cooling, slides were placed in 1× envision FLEX wash buffer (K8007; Agilent) for 5 min. After rinsing, slides were then protein‐blocked for 10 min (Background Sniper; BioCare Medical). After rinsing, slides were incubated with either anti‐MBP (1:4000, Abnova MAB20219) or anti‐NF (1:1000, Neuromics MO22103) for 30 min. The slides were then rinsed with DI water and buffer, followed by a 20‐min incubation with EnVision FLEX horseradish peroxidase (K8002; Agilent). After incubation, slides were rinsed with DI water, and EnVision FLEX DAB+ Sub‐Chromo (DM847; Agilent) was applied for 10 min. A counter stain of hematoxylin (K8008; Agilent) was then applied for 10 min. After a quick DI water wash, slides were rinsed with 70% ethanol and were quickly rinsed in two changes of 95% ethanol, two subsequent changes of 100% ethanol, and lastly two subsequent changes of xylene. After xylene rinses, the slides were automatically coverslipped.

2.3.5. Immunofluorescence

Immunofluorescence (IF) was performed using an automated stainer (Autostainer Link 48; Agilent). To avoid tissue wash‐off, slides were placed in the oven at 70°C for 20 min and were deparaffinized as above. AR was performed (DAKO PT Link; Agilent) using 1× citrate buffer (pH 6) at 85°C for 25 min. Slides were rinsed for 5 min in buffer and then protein‐blocked for 10 min. After rinsing with buffer, sections were incubated in anti‐MBP (1:2000, Abcam 7349) for 60 min, rinsed, and then incubated in anti‐NF (1:1000, Neuromics MO22103) for 60 min for dual staining. Secondary fluorescence antibodies Alexa Fluor 488 goat anti‐mouse IgG (1:200, Invitrogen, A‐11029) and Cy5 goat anti‐rat IgG (1:200, Invitrogen, A‐10525) were used for anti‐NF and anti‐MBP, respectively, at an incubation time of 60 min. Sections were rinsed with DI water and manually coverslipped using DAPI Fluoromount‐G (Southern Biotech).

2.3.6. Microscopy

All stained slides were imaged using either the 20× or 40× objective on a Zeiss Axioscanner Z7. For IF‐stained sections, samples were imaged with fluorescent channels: Cy5 for MBP, 488 for NF, and a standard DAPI reference channel.

2.3.7. Analyses of H&E Images

Automated image analysis was performed using QuPath (v5.0, QuPath docs authors, https://qupath.readthedocs.io/en/stable/index.html) and color deconvolution algorithms within the software [44]. The optical density (OD) was used to measure the staining intensity of H&E images. The OD is the measure of absorbance of light through a sample and is proportional to the stain concentration where greater stain concentration correlates to a higher value of greater optical intensity [45]. QuPath's algorithms define the mean OD for H&E pixel layers separately based on the original colorimetric data or by assigning color vectors to each layer based on what is visually observed [44]. After the separation of staining layers through automatic color deconvolution, the regions encapsulating the tissue section were chosen, and the individual hematoxylin‐positive nuclei were segmented automatically using QuPath's cell detection. A pixel size of 0.25 µm was used along with a background radius of 2 µm to best segment the nuclei. We corrected the mean OD based on the number of nuclei segmented on the entire section. The OD is relative to a batch performed at the same conditions; thus, for this study, all samples were stained by batch to ensure consistency. Replicates of the 3% PTA stained in water and the 10% stock solution were performed with a different section of tissue. A one‐way ANOVA test was performed to evaluate statistical significance using Minitab (v21.2, Minitab, LLC, 2024, University Park, PA, minitab.com).

We manually segmented the fascicle boundaries and then automatically segmented the Schwann cell nuclei therein, which were stained dark purple by the hematoxylin (Figure 7). The OD of the hematoxylin‐stained nuclei (dark purple) and the eosin‐stained collagenous tissue (pink) was quantified using QuPath's internal tools, which digitally separate stain components into the appropriate RGB channels via color deconvolution and calculate the OD per pixel following image thresholding.

FIGURE 7.

Masson's trichrome‐stained cross sections of the vagus nerve treated with 3% phosphotungstic acid (PTA) for 8 and 48 h (top row) and 10% PTA for 8 and 48 h (bottom row), using water as the solvent. (A) PTA at 3% in water (8 h). (B) PTA stock solution at 10% (8 h). (C) PTA at 3% in water (48 h). (D) PTA stock solution at 10% (48 h). (E) Unstained sample (control). Enlarged views of each section are provided on the right to emphasize the trichrome staining.

3. Results

3.1. Perineurium Intensity Across Stains

With PTA staining, the perineurium had higher voxel intensity than the fascicular or IFE voxels (Figure 1E), thus exhibiting clear bright boundaries of the nerve morphology (Figure 1A). Conversely, Lugol's iodine staining had relatively constant intensity across the tissues (Figure 1B,E). Osmium tetroxide staining had a marked decrease in intensity at the perineurium, persisting throughout the fascicle; higher voxel intensities occurred in the epineurium due to the osmium‐binding fatty deposits surrounding the fascicles (Figure 1C,E).

From the graphs (Figure 2), it can be observed that PTA had the highest P/F and P/IEE ratios across all nerve samples (across 39 lines per staining modality), indicating superior visualization of perineurium in PTA‐stained samples. Conversely, as seen in Figure 1, iodine resulted in similar intensity across all three neural tissues—and thus weak perineurial contrast—as reflected by contrast ratios near 1. Osmium tetroxide also had a contrast ratio near 1 for P/IFE, but a lower P/F ratio, indicating that the fascicles were brighter than the perineurium.

3.2. Optimization of Scan Parameters for PTA‐Stained Nerve Tissue

To optimize the micro‐CT images for PTA‐stained nerve tissues, we systematically evaluated the effects of voltage, current, and detector integration time on image sharpness, noise, and pixel saturation. We imaged three nerve samples with all evaluated scan parameters, for a total of 16 scans per sample (scan parameters mentioned in Table 1). As shown in Figure 3, increasing the voltage decreased the pixel saturation (across the rows), and increasing the current decreased image sharpness (across the columns), consistent with established CT imaging principles, where higher currents are known to influence the focal spot size, ultimately contributing to image blurring [46]. The integration times were used based on the machine's limitations at the particular scan settings. We quantified image sharpness as the variance of the image gradient [47], noise levels as the standard deviation of pixel intensities, and pixel saturation as the number of pixels with a saturated intensity value of 2 [16]. We observed a trade‐off between image sharpness and noise: increasing current at a fixed voltage or increasing voltage at a fixed current decreased image noise but also decreased image sharpness (Figure 4). Avoiding pixel saturation is also critical for image quality; saturation was negatively correlated with voltage and reached negligible levels at 55 kV. Therefore, to compromise between sharpness and noise—given that the noise saturated between 55 and 70 kV—we identified 55 kV and 109 µA as suitable scanning parameters for our experimental requirements.

FIGURE 3.

Micro‐CT images for one cross section of one nerve sample acquired using varying current, voltage, and integration time, showing differences in image sharpness, noise, and intensity. Scale bars are 500 µm, except for the zoomed‐in images in the third row with 200 µm scale bars. Intensity was window‐leveled similarly across images to enable comparisons.

FIGURE 4.

Image sharpness (A) and noise (B) across current (x‐axis) and voltage (colors) scan parameters for the same image location.

3.3. H&E Staining of PTA‐Stained Vagal Nerve Varying Solvent, Concentration, and Time

The H&E stain provides a comprehensive visualization of nerve tissue microanatomy by staining nuclear components as well as collagen fibers. We evaluated the effects of PTA staining on H&E histology across different PTA concentrations (0%, 3%, or 10%), solvents (DI water or ethanol), and stain times (8 or 48 h) (Figures 5 and 6). The perineurium stayed consistently intact across PTA conditions, but shrinkage and cracking of the endoneurium and epineurium occurred more prominently with higher concentrations of PTA. This cracking is apparent in other histological studies of non‐PTA‐stained human cadaver vagal nerve but seems to be exacerbated with higher concentrations of PTA [48].

FIGURE 5.

Histology with hematoxylin and eosin of human cadaveric vagus nerves across different phosphotungstic acid (PTA) concentrations (0%, 3%, or 10%), stain times (8 or 48 h), and solvents (deionized water or ethanol). (A) PTA at 3% in water (8 h). (B) PTA stock solution at 10% (8 h). (C) PTA at 3% in water (48 h). (D) PTA stock solution at 10% (48 h). (E) Unstained sample (control).

FIGURE 6.

Hematoxylin and eosin‐stained cross sections of the vagus nerve treated with different durations and solvents. The top row shows tissue stained with 3% phosphotungstic acid (PTA) in water for 8 and 48 h, while the bottom row displays tissue stained with 3% PTA in ethanol for 8 and 48 h. (A) PTA at 3% in water (8 h). (B) PTA at 3% in ethanol (8 h). (C) PTA at 3% in water (48 h). (D) PTA at 3% in ethanol (48 h). (E) Unstained sample (control). Enlarged views of each section are shown on the right to highlight the hematoxylin and eosin staining.

Higher OD values correspond to greater staining density. In this analysis, we primarily focused on using two conditions—PTA in water and PTA stock solution at various time points—as water was the solvent used for PTA staining in the micro‐CT experiments described earlier. Notably, we did not observe any significant differences in OD between the PTA staining conditions with a p‐value of 0.943 (p >> 0.05) (Table 2).

TABLE 2.

Summary of the average optical density.

Phosphotungstic acid staining condition (number of nerve samples = 3)	Mean optical density
Unstained	0.46 ± 0.060
3% PTA, water, 8 h	0.48 ± 0.087
10% PTA, 8 h	0.45 ± 0.100
3% PTA, water, 48 h	0.48 ± 0.103
10% PTA, 48 h	0.50 ± 0.045

Note: The table presents a summary of the mean optical density ± standard deviation across varying phosphotungstic acid (PTA) concentrations and stain durations.

3.4. Masson's Trichrome Staining of PTA‐Stained Vagal Nerve Varying Solvent, Concentration, and Time

Masson's trichrome is a widely used histological stain that offers more detailed morphological information compared to the standard H&E stain [49]. In neural tissue, Masson's trichrome highlights different components: myelin sheaths stain red, collagen appears blue, and the nuclei are dark blue or black.

PTA has been used as a histological agent in stains such as Mallory's stain and has been shown to bind preferentially to collagen [40]. In the case of Masson's trichrome, which contains acid fuchsin, hematoxylin, and aniline blue, it has been observed that PTA selectively blocks the staining of most tissue components, except for aniline blue, which stains collagen‐rich connective tissue [50]. In terms of vagal nerve tissue, this allows for easier visualization of the perineurium and epineurium within the tissue; although this benefits collagen‐rich tissues, the PTA can decrease the differentiation of Biebrich scarlet‐acid fuchsin staining, which appears red and stains the myelin sheath around individual axons. Many other histological and immunological stains can be used to characterize myelin around peripheral nerve tissue, such as toluidine blue, Luxol fast blue, and OsO₄ [51, 52]. Regardless of the concentration or time in PTA, the differentiation between the red staining of scarlet‐acid fuchsin (myelin sheaths) and the aniline blue staining (collagen‐rich connective tissue) remains intact, allowing for the visualization of both components (Figures 7 and 8).

FIGURE 8.

Masson's trichrome‐stained cross sections of the vagus nerve treated with different durations and solvents. The top row shows tissue stained with 3% phosphotungstic acid (PTA) in water for 8 and 48 h, while the bottom row displays tissue stained with 3% PTA in ethanol for 8 and 48 h. (A) PTA at 3% in water (8 h). (B) PTA at 3% in ethanol (8 h). (C) PTA at 3% in water (48 h). (D) PTA at 3% in ethanol (48 h). (E) Unstained sample (control). Enlarged views of each section are shown on the right to highlight the trichrome staining.

The effect of solvent on PTA staining has been explored in ethanol and water due to different protocols [53]. Observations between the two solvents have shown less tissue shrinkage compared to that of iodine‐based methods. Histologically, there is no apparent difference between the solvents with overall general morphology conserved.

3.5. Immunohistochemistry of PTA‐Stained Vagal Nerve Varying Solvent, Concentration, and Time

We evaluated the effects of PTA concentration, solvent, and staining time on immunohistochemistry with antibodies against NF and MBP using brightfield (Figures 9, 10, 11, 12) and fluorescence (Figure 13). In all cases, clear NF and MBP labeling is evident, but there is a noticeable difference between PTA‐stained and unstained samples: PTA‐stained samples have more epineural cracking—especially at higher PTA concentrations, as observed with H&E histology—and darker background staining. Without PTA staining, the slices had less folding and better slide adhesion. MBP, NF, and DAPI IF labeling processes were successful with and without PTA (Figure 13).

FIGURE 9.

Neurofilament‐stained cross sections of the vagus nerve treated with 3% phosphotungstic acid (PTA) for 8 and 48 h (top row) and 10% PTA for 8 and 48 h (bottom row), using water as the solvent. (A) PTA at 3% in water (8 h). (B) PTA stock solution at 10% (8 h). (C) PTA at 3% in water (48 h). (D) PTA stock solution at 10% (48 h). (E) Unstained sample (control). Enlarged views of each section are provided on the right to emphasize immunohistochemistry staining.

FIGURE 10.

Neurofilament‐stained cross sections of the vagus nerve treated with different durations and solvents. The top row shows tissue stained with 3% phosphotungstic acid (PTA) in water for 8 and 48 h, while the bottom row displays tissue stained with 3% PTA in ethanol for 8 and 48 h. (A) PTA at 3% in water (8 h). (B) PTA at 3% in ethanol (8 h). (C) PTA at 3% in water (48 h). (D) PTA at 3% in ethanol (48 h). (E) Unstained sample (control). Enlarged views of each section are shown on the right to highlight the immunohistochemistry staining.

FIGURE 11.

Myelin basic protein‐stained cross sections of the vagus nerve treated with 3% phosphotungstic acid (PTA) for 8 and 48 h (top row) and 10% PTA for 8 and 48 h (bottom row), using water as the solvent. (A) PTA at 3% in water (8 h). (B) PTA stock solution at 10% (8 h). (C) PTA at 3% in water (48 h). (D) PTA stock solution at 10% (48 h). (E) Unstained sample (control). Enlarged views of each section are provided on the right to emphasize immunohistochemistry staining.

FIGURE 12.

Myelin basic protein‐stained cross sections of the vagus nerve treated with different durations and solvents. The top row shows tissue stained with 3% PTA in water for 8 and 48 h, while the bottom row displays tissue stained with 3% PTA in ethanol for 8 and 48 h. (A) PTA at 3% in water (8 h). (B) PTA at 3% in ethanol (8 h). (C) PTA at 3% in water (48 h). (D) PTA at 3% in ethanol (48 h). (E) Unstained sample (control). Enlarged views of each section are shown on the right to highlight the immunohistochemistry staining.

FIGURE 13.

Myelin basic protein (MBP)–neurofilament (NF) dual immunofluorescence staining of the vagus nerve. (A) Nerve samples stained with 3% PTA in water, showing separate antibodies in three MBP, NF, and DAPI channels, respectively (top row). (B) Vagus nerve stained with 10% PTA (stock solution) with separated MBP, NF, and DAPI channels (middle row). (C) An unstained vagus sample is used as a control, displaying the separated MBP, NF, and DAPI channels (bottom row).

4. Discussion

The main goal of this study was to develop a staining method that provides superior visualization of fascicle and fascicle boundaries to help develop accurate computational models for simulation and the design of PNS therapy, by enabling robust automated segmentation, accurate quantification of nerve morphology, and compatibility of the stained tissue with downstream histology. Osmium tetroxide and Lugol's iodine have been previously used to stain peripheral nerves for micro‐CT imaging, but osmium tetroxide preferentially stains fatty deposits in the epineurium, while Lugol's iodine offers less pronounced contrast between the perineurium and the inner fascicle edge. To address this challenge, a stain specifically targeting perineurial connective tissue (rich in collagen) is desirable [54]. PTA emerges as a promising candidate due to its established role as a contrast agent for highlighting collagenous structures in micro‐CT imaging. It has been used to visualize tendons, plaques, and other collagenous structures [40, 43, 55]. In this study, we evaluated the staining methods for micro‐CT imaging of nerve fascicles. We compared the intensity profiles of the perineurium relative to the fascicle and IFE across three stains: Lugol's iodine, osmium tetroxide, and PTA. Quantitative analysis of the P/F and P/IFE contrast ratios revealed that PTA staining yielded higher intensities for the perineurium and better visualization of the perineurium. While both Lugol's iodine and osmium tetroxide have established applications in nerve imaging, our findings suggest that PTA offers a distinct advantage in delineating fascicle boundaries. This enhanced visualization may facilitate efficient segmentation of fascicles across large image stacks, a critical step for subsequent analysis. Additionally, we developed a micro‐CT image analysis framework to semi‐objectively optimize scanning parameters for nerve imaging. This framework involved the quantification of image sharpness and noise across various scan settings, enabling the selection of optimal parameters tailored to our specific needs.

Histology provides complementary anatomical information to micro‐CT, including the identification of fascicles too small for micro‐CT to resolve, validation of perineurium thickness, and fiber‐level information (such as diameters, locations, and types). Previous studies demonstrated that nerve samples that were stained with Lugol's iodine for micro‐CT imaging can be de‐stained prior to histology [7]. We demonstrated herein that PTA de‐staining is not required for compatibility with H&E histology, Masson's trichrome histology, and immunohistochemistry with antibodies against NF and MBP. Endoneurial shrinkage and, consequently, cracking were observed at higher PTA concentrations; the perineurium remained intact, which would enable the quantification of fascicle sizes and perineurium thickness, but the shrinkage would affect the quantification of fiber diameters and locations. Supported by Masson's trichrome staining and prior literature [56], we observed that PTA binds to positively charged proteins, such as collagen, resulting in a more intense blue from the aniline blue staining. Samples stained with higher concentrations of PTA or exposed for longer durations, however, had differentiation between the red‐stained myelin and blue‐stained collagen, which can enable visualization of the myelin within fascicles and the surrounding collagenous epineurium and perineurium (Figure 8) [40]. Importantly, we demonstrated that PTA staining was compatible with immunohistochemistry and IF, effectively labeling the axons and myelin of individual fibers. This can aid in the analysis and quantification of myelinated and unmyelinated regions within the nerve, as well as in measuring the differences in fiber sizes. This morphological information is essential for accurate computational modeling of fascicles.

Micro‐CT of peripheral nerves enables the visualization, quantification, and computational modeling of 3D fascicular morphology. However, given micro‐CT's high throughput and large field of view, automated segmentation of fascicle boundaries is necessary for tractable and robust segmentation; manual segmentation is too time‐consuming and challenging to conduct accurately in three dimensions. Thus, achieving a clear contrast of the fascicle boundaries is critical. Our novel protocol for PTA staining of peripheral nerves drastically improves perineurial contrast, providing rings with bright intensity compared to the surrounding epineurium and enclosed endoneurium. Such segmented micro‐CT data can then serve as inputs to anatomically realistic computational models of PNS to analyze mechanisms of action and to design improved therapies [16, 20]. With this robust PTA‐based perineurial contrast, unlike histology, micro‐CT captures changes in fascicle size and count along the nerve as the fascicles split and merge in 3D. As the perineurium has a higher impedance, it has important effects on PNS therapies; specifically, thicker perineurium and larger fascicle diameters result in higher activation thresholds [57].

We assessed micro‐CT image quality using three metrics: sharpness, noise, and pixel saturation. Our micro‐CT scanner enabled repeated imaging of a given nerve sample at the same location, enabling a robust comparison across imaging parameters. Our findings were consistent with established CT principles [46], demonstrating that lower current and voltage improved image sharpness but yielded higher noise levels. Thus, this systematic evaluation of data collection parameters and resulting image quality metrics provides a quantitative approach for selecting scan parameters for a given set of objectives.

One of the limitations of our study is that the quality of PTA staining and the depth of penetration depend on the size and composition of the tissue. Therefore, the resulting micro‐CT image quality and PTA's downstream effects on histological quality may vary across samples if different staining protocols are used. To reduce this variability, standardization may be achieved by taking intermittent images throughout the staining period to titrate the time in PTA appropriately. Tissue quality and different CT scanners will also affect image quality. Tissue fixation (type and duration), handling, and processing will likely influence staining quality [58], and appropriate scan parameters must be catered to different tissues and CT scanners. Nonetheless, our study provides a framework to develop staining and imaging protocols for a specific tissue preparation and instrument. Another limitation of our study is that we did not perform a comprehensive and systematic study on the effects of stain duration and concentration on contrast in nerve tissue, and thus, it would need to be repeated for different tissue types and preparations.

Here, we demonstrated the first use of PTA for micro‐CT imaging of peripheral nerves, demonstrating robust visualization of the nerve morphology in cadaveric human vagus nerves, while preserving compatibility with subsequent histological and immunohistochemistry. Specifically, we demonstrated detailed visualization of the perineurium, which allows differentiation of fascicle boundaries. PTA provided much clearer perineurial contrast compared to other established stains (Lugol's iodine and osmium tetroxide). Furthermore, we systematically evaluated micro‐CT imaging parameters and determined that 55 kV and 109 µA balanced noise reduction and image sharpness. It is recommended that these parameters be adjusted according to the specific scanner used; our study provides a framework for such adjustments, using measurements of noise, sharpness, and saturated pixels. Although PTA is predominantly nondestructive and compatible with downstream histology, it is important to note that higher concentrations and extended stain times can potentially degrade the quality of the tissue. H&E histology showed satisfactory results, albeit with minor qualitative aberrations. Special stains such as Masson's trichrome and immunohistochemical staining with DAB were effective, though PTA may slightly affect stain differentiation and increase background staining, respectively. IF staining exhibited no discernible differences, affirming PTA's utility as a robust and minimally invasive contrast agent for nerve imaging in anatomical studies.