Quantitative and large‐format histochemistry to characterize peripheral artery compositional gradients

The femoropopliteal artery (FPA) is a long, flexible vessel that travels down the anteromedial compartment of the thigh as the femoral artery and then behind the kneecap as the popliteal artery. This artery undergoes various degrees of flexion, extension, and torsion during normal walking movements. The FPA is also the most susceptible peripheral artery to atherosclerosis and is where peripheral artery disease manifests in 80% of cases. The connection between peripheral artery location, its mechanical flexion, and its physiological or pathological biochemistry has been investigated for decades; however, histochemical methods remain poorly leveraged in their ability to spatially correlate normal or abnormal extracellular matrix and cells with regions of mechanical flexion. This study generates new histological image processing pipelines to quantitate tissue composition across high‐resolution FPA regions‐of‐interest or low‐resolution whole‐section cross‐sections in relation to their anatomical locations and flexions during normal movement. Comparing healthy ovine femoral, popliteal, and cranial‐tibial artery sections as a pilot, substantial arterial contortion was observed in the distal popliteal and cranial tibial regions of the FPA which correlated with increased vascular smooth muscle cells and decreased elastin content. These methods aim to aid in the quantitative characterization of the spatial distribution of extracellular matrix and cells in large heterogeneous tissue sections such as the FPA.

The femoropopliteal artery (FPA) is the predominant blood-supplying vessel to the leg and courses through the anterior compartment of the femoral region (Swift & Bordoni, 2019).The FPA is a highly dynamic artery as it is subjected to biomechanical stressors such as flexion, extension, torsion and compression as a result of everyday movement such as standing, walking and crouching (Adiguzel et al., 2009;Kamenskiy et al., 2017).The ability of the FPA to cyclically reproduce these healthy mechanics across a lifetime depends on its composition of extracellular matrix (ECM) and cells, such as elastin, collagens, and smooth muscle cells, which provide compliance and resistance during elastic deformation.The dysregulation of FPA composition and its impact on FPA mechanics are hypothesized to be critical factors and indicators of peripheral artery disease (PAD).PAD is most commonly characterized through the manifestation of atherosclerosis, causing a narrowing of the arteries or the complete loss of arterial patency (Figure 1) (Hirsch et al., 2006).PAD occurs within the FPA in 80% of cases, causing ischemia of the lower limb (Losordo et al., 2015;O'Donnell et al., 2011) with severe cases requiring revascularization (Dhaliwal & Mukherjee, 2007).Surgical intervention, in particular endovascular intervention, is often utilized for patients with severe PAD (O'Donnell et al., 2011).Despite the continuous prevalence of PAD, better methods to detect and understand FPA composition are critical to future PAD treatments.
Histology remains a gold-standard technique used to characterize the composition of healthy and diseased tissue, although these methods remain largely manual processes which require experienced technicians.This, however, provides an opportunity for the adoption of automated computational analyses to improve workflows and diagnoses.Quantitative histological image analyses could help reduce operator error and bias and enable the characterization of largeformat tissue samples where manual interpretations remain prohibitively time-consuming.This transition toward digital histology has relied on the development of computational image analysis algorithms, such as for the effective and reliable assessment of diseases (Kim et al., 2011).This study describes the use of NIH ImageJ software for quantitative image analysis of arterial wall composition, particularly vascular smooth muscle cells (VSMCs), elastin and collagen across regions of the FPA.This quantification was then compared with quantitative biomechanical analysis radiographically imaged FPA deformations.To compare analytical trade-offs between image resolution and size or sectioning method, this study assessed quantitative differences in transverse versus longitudinal sectioning methods and low-resolution whole-slide imaging versus high-resolution regionof-interest ("ROI") imaging.
This study aims to describe histological preparation and image processing methods to investigate variations within FPA cytochemical content at various locations and degrees of flexion.With supporting literature, it is hypothesized that an increase in VSMC composition and decrease in elastin and collagen composition are associated with regions of an artery undergoing high degrees of biomechanical stress.
This aims to provide a greater understanding of the artery itself and its environment which may aid the design of safer, personalized and patient-specific PAD therapies (Ansari et al., 2013).This study uses an ovine model to investigate arterial morphology, which are commonly used in pilot research for human stenting as they possess comparable arterial hindlimb anatomy (Byrom et al., 2010;Schleimer et al., 2018;Swartz & Andreadis, 2013;Tannast et al., 2018).

| Histological tissue sectioning and staining
Gross anatomical dissection was performed by resecting right and left femoropopliteal arteries which were then cleaned of connective tissue.Following resection, the arterial tissue was fixed in 4% paraformaldehyde for 4 days then washed twice with phosphatebuffered saline and placed in 70% ethanol followed by a 9 hour tissue processing run using ThermoFisher Excelisor™ AS Tissue Processor according to published protocols (Ren et al., 2021).After processing, the femoropopliteal artery was segmented into 1 cm segments with alternating segments dedicated for transverse or longitudinal sectioning as illustrated in Figure S1.Segments dedicated to transverse sectioning were further segmented into approximately 0.33 cm segments to allow for more comprehensive analysis.Segments were embedded in Tissue-Tek ® TEKIII paraffin wax (Sakura Finetek USA Inc., California, United States) using a ThermoFisher Shandon Histocentre 3 (Thermo Fisher Scientific, Massachusetts, United States) and underwent microtomy using a Leica Biosystems RM2245 (Leica Biosystems, Wetzlar, Germany).Regions of the artery were correlated to anatomical landmarks to ensure the histology reflected the arterial region.The start of femoral artery was indicated through the bifurcation of the external iliac artery into femoral and deep femoral artery, the start of popliteal artery was indicated by the continuation of the femoral artery passing the popliteal fossa, and the start of the cranial tibial artery was indicated by the bifurcation into the cranial tibial and caudal tibial arteries.Three consecutive sections of 3 μm each were obtained from each axial segment and were stained using hematoxylin and eosin (H&E), Verhoeff-Van Gieson (VVG) and Masson's Trichrome (MT) stains, respectively.This was replicated three times: thus, nine total sections were obtained per segment.H&E staining was conducted using Leica Autostainer and Autocoverslipper ST5010-CV5030 (Leica Biosystems, Wetzlar, Germany) using the histology lab standard operating procedure (SOP).
VVG staining was completed manually as follows; sections were brought to water and stained with Verhoeff's hematoxylin comprised of 5% hematoxlylin (Acros Organics, Geel, Belgium), 10% ferric chloride (Sigma Aldrich, Missouri, United States), Lugols Iodine (Univar/ Ajax Finechem, Massachusetts, United States), after washing with distilled water, sections were dipped in 2% aqueous ferric chloride (Sigma Aldrich, Missouri, United States) to differentiate black elastic fibers on a pale gray background.After differentiation, sections were rinsed with distilled water and colored blue once dipped in 0.1% ammonia.Ammonia was replaced with 1% aqueous eosin then washed with 90% ethanol to remove potential excess iodine staining.
Sections were then stained with Van Gieson solution (acid fucshin CI24685, saturated aqueous picric acid; Sigma Aldrich), blotted dry, dehydrated with absolute alcohol (Univar/Ajax Finechem, Massachusetts, United States), and placed in xylene in preparation for mounting.
MT staining was also completed manually.Sections were brought to water then underwent post-fixation in 60 C picric acid (Sigma Aldrich) for 1 h then washed in water.Sections were then stained with Weigert's iron hematoxylin (HT107 Part A and HT109 Part B; Sigma Aldrich).Hematoxylin was washed with water and sections were then quickly dipped in 0.5% acid alcohol for differentiation and immediately washed to avoid over-differentiation.Sections were then blued using 0.1% ammonia and briefly rinsed.Sections were stained with Masson's Red (Ponceau Xylidine CI27000, Sigma Aldrich) and quickly replaced with 1% phosphomolybdic acid (Sigma Aldrich) to differentiate red muscle tissue from collagen.Sections were then stained with methyl blue (Alfar Aesar) until its color intensity matched the previous red staining and rinsed with 1% glacial acetic acid (Univar, Ajax Finechem) to remove excess staining.Sections were then dehydrated in absolute ethanol (Univar, Ajax Finechem) and placed in xylene in preparation for staining.
Stained sections were scanned at 40Â lens magnification using the 3DHistech Slide Scanner and viewed on 3D Histech Case Viewer software (3DHISTECH Kft., Budapest, Hungary).

| Computational histology analyses
The computational quantitation of cells and ECM were compared in two methods: a comparison of transverse versus longitudinal artery slices and a comparison of "region-of-interest" (ROI) versus "wholesection quantitation" as illustrated in Figure S2.The "ROI quantification" method is similar to previous publications (Avolio et al., 1998;Clarke et al., 2003;Dettmer et al., 2013;Meshram et al., 2017;Moguillansky et al., 2011) and includes the random selection of three rectangular "ROIs" at 15Â export magnification (182 Â 323 μm, 0.32 μm/pixel) containing all layers of the arterial wall (tunica intima, tunica media and tunica adventitia) and the mean across the three "ROIs" were used to represent the material content for the complete arterial section.The "whole-section quantification" method consisted of a low-magnification image of the whole artery (WA) section at

| Statistical analyses
The Kruskal-Wallis test was performed to identify statistically significant variations in material (elastin, VSMC, collagen) composition across the femoropopliteal and cranial tibial regions.ANOVA parameters (groups) are represented through a gross location such as the femoral artery, popliteal artery, and cranial tibial artery.The Student's T-test was performed to identify significant discrepancies in quantification measurements between the "ROI" and "whole-section quantification" methods and transverse and longitudinal sections.All histological procedures were replicated across the two hindlimbs of all four sheep.Radiographical imaging was only performed on the right hindlimb of sheep 4. All statistical analyses were performed using GraphPad Prism with statistical significance defined by p values less than .05.

| Large-format histology accurately preserves tissue architecture
The H&E stain showed overall cellular morphology of the arterial wall.
VSMCs were most clearly represented through their blue-purple stained nuclei, which were elongated in transverse sectioning or round in longitudinal sectioning (Figure 1a,b).Collagen and elastin were stained a pale-pink color, VSMC bodies were more saturated in pink stain.MT stain depicted VSMC through red staining with collagen stained blue in color (Figure 1c).Elastin is stained pale-red and was not clearly recognized in MT staining in comparison to VVG staining (Figure 1c,d).The VVG stain presented elastin as thin, black, wavy lines in the tunica media of transverse sectioning, layered with VSMCs and collagen (Figure 1d,e).In contrast, adventitial elastin is presented as black dot-like structures between layers of collagen (Figure 1e).In longitudinal sectioning, fine elastin in the tunica media is visualized as black dot-like structures whereas elastin in the tunica adventitia appears as black, layered lines (Figure 1f).This confirms the circumferential orientation of fine elastin in the tunica media and longitudinal orientation of elastin in the tunica adventitia.VSMCs appear as yellow-orange (in VVG stain) within the tunica media of the arterial wall (Mozafari et al., 2019).In transverse sectioning, VSMCs appeared spindle-shaped with elongated nuclei whereas in longitudinal sectioning VSMCs appeared more round with circular nuclei (Figure 1e,f).This suggests a circumferential orientation of VSMCs in the tunica media as it is layered with similarly circumferential elastin and collagen.

| Arterial flexion is predominately found in femoropopliteal and not tibial regions
For one cadaveric sheep (sheep #4), contrast was injected and X-ray imaging was undertaken under different amounts of induced limb flexion to observe the resultant contortions of the femoral, popliteal, and cranial tibial regions in situ (Figure 2a-c).Considerable angular bending was observed within the popliteal region in all full flexion (Figure 2a), neutral (Figure 2b) and full extension (Figure 2c) positions of the cadaveric ovine limb.Arterial compression was observed within the popliteal region in full flexion directly posterior to the tibiofemoral joint visualized though the disruption of contrast (Figure 2a, orange arrow) compared to neutral limb position (Figure 2b) where no disruption is identified.This disruption resulted in decreased contrast within the cranial tibial artery.Longitudinal stretch was also observed within the femoral and popliteal region between neutral and full extension position, characterized by the straightening of the artery (Figure 2b,c).No gross changes were observed within the caudal tibial artery across the different flexion positions.Altogether, the x-ray images indicated that femoral and popliteal arteries undergo a greater degree of flexion than tibial arteries, in agreement with current literature (Poulson et al., 2018).

| Elastin content is directly and VSMC content is inversely associated with peripheral artery contortion and distance from heart
An increased degree of arterial contortion was observed in inducedlimb flexion, particularly within the popliteal region (Figure 2a-c).
Comparing to the more distal cranial tibial region that did not undergo the same degree of arterial deformation/contortion, the popliteal region possessed a proportionally larger elastin content (Figure 2d-f).
A proportionally greater amount of longitudinally orientated elastin was observed in the tunica adventitia and lesser VSMC content in the tunica media of the popliteal region compared to the cranial tibial region (Figure 2e,f).Qualitative image analysis of the stained sections shows a proportional decrease of elastin between the proximal femoral and distal cranial tibial regions (Figure 3a).A decrease was particularly observed within the tunica adventitia with fewer elastin fibers present within distal regions of the femoropopliteal artery and a decrease of fine elastin within the tunica media was also observed (Figure 2d-i).In contrast, an overall significant increase in VSMCs was observed particularly from the femoral to popliteal region and from the popliteal region to the cranial tibial region (Figures 2d-f, 3b).No significant differences were observed in collagen content across the femoral, popliteal, and cranial tibial regions (Figure 3c).A significant increase in VSMCs was observed within the tunica media and led to   b) Statistically significant variation in VSMC content was observed between femoral and popliteal region (p = .0138)and popliteal and cranial tibial region ( p = .0118).VSMC content between ROI and WA quantification in cranial tibial region also significantly varied ( p = .0037),as shown with an orange box.(c) Statistically significant variations were observed between ROI and WA quantification of femoral ( p = .0201)and popliteal ( p = .0004)regions, as shown with orange boxes.A statistically significant difference was observed for collagen content between transverse and longitudinal sections in the cranial tibial region using ROI quantification ( p = .0424),shown with a green asterisk also in Figure 3c,f.Statistically significant variations of VSMC content was observed between ROI and WA quantification in the popliteal region ( p = .0074),shown by an orange box.
the thicker appearance of the tunica media as it occupies most of the arterial wall in distal regions (Figure 2d,f).Results from longitudinal segments present similar trends in elastin and VSMC content, however, a nonsignificant increase in collagen was observed between the femoral and popliteal region (Figure 2g-i).In longitudinal segments, no statistically significant differences were found in elastin, VSMC and collagen composition between the femoral, popliteal, and cranial tibial regions (Figure 3d-f).

| Elastin and VSMC analyses are more sensitive to image magnification than image area
Two methods of material composition quantification were utilized within this study.The "ROI quantification" method averaged three randomly chosen images of the arterial wall in high magnification (15Â).In contrast, "whole-section quantification" utilized low magnification (4.4Â and 1.5Â) images of the WA sections to represent a trade-off between image resolution and size.Differences between results from ROI and whole-section quantification methods can be qualitatively observed (Figure 3) particularly in the amount and trend of collagen composition within both transverse (Figure 3a-c) and longitudinal segments (Figure 3d-f).In whole-section sections, transverse segments' collagen content showed a decrease from the femoral region to the cranial tibial region, while longitudinal segments showed an increase (Figure 3d,f).However, transverse whole-section measurements yield higher standard deviations than "ROI" measurements.Statistically significant differences were observed for collagen composition for the femoral and popliteal regions of transverse segments between ROI and whole-section quantitation (Figure 3c).Significant differences were also present in VSMC measurements in the cranial tibial region of transverse segments and popliteal region of longitudinal segments (Figure 3b,e), however elastin measurements were nearly identical in value and deviation (Figure 3a,d).The high variability in WA section characterization is illustrated in Figure 4 and Figures S2 and S3.It is more difficult to identify individual elastin fibers or cells from WA section magnifications in comparison to ROI magnifications.

| Collagen quantitation depends significantly on sectioning and image analysis method
Collagen alignment is perpendicular to elastin within the tunica adventitia (where majority of elastin is found) indicating that traditional transverse sectioning of vessels appeared to under-estimate the content of collagen (Figure 3c,f).In transverse sectioning, collagen appears as long segments within the tunica media however, in longitudinal sectioning, they appear as round clusters.This is due to the circumferential alignment of collagen within the arterial wall (Figure 4).Variation in collagen composition was found in ROI quantification between transverse sectioning and longitudinal sectioning (Figure 3c,f).Although these discrepancies were observed, statistically significant differences of quantified collagen between transverse and longitudinal sections were only found within the cranial tibial region (p = .0424),where collagen content was highest.Collagen content also varied depending on the image analysis method used.Paired t-tests for collagen between ROI and WA quantification methods were performed for both transverse and longitudinal sectioning.

| DISCUSSION
Arterial wall morphology is comprised of three distinct layers; tunica intima, tunica media and tunica adventitia, each with differing tissue compositions dependent on their function.The tunica intima is a thin layer containing the endothelium, smooth muscle cells, and the internal lamina comprised of elastin (Mercadante & Raja, 2020).The tunica media is responsible for arterial tone, consisting of a thick layer of concentric smooth muscle and fine elastin (Mercadante & Raja, 2020).
The tunica adventitia is mainly comprised of connective tissue as it serves to anchor the position of the artery with surrounding tissues.
The FPA is categorized as a muscular artery as the tunica media is heavily concentrated with VSMCs and fewer elastin fibers compared to their elastic artery counterparts (Tucker & Mahajan, 2018).Elastin located within the external elastic lamina is oriented longitudinally to facilitate the FPA's dynamic contortions (Kamenskiy et al., 2017).
Peripheral arteries are well-known to increase in elastin composition and decrease in VSMC composition as they get closer to the heart, to accommodate for an increase in blood pressure (Tucker & Mahajan, 2018).This corroborates with prior studies that also show arteries with increasing and constant biomechanical stress present a decrease in elastin composition, particularly stretch and torsion mechanics from limb flexion (Han et al., 2016;Wang et al., 2018).This decrease in elastin can be due to VSMCs' response to biomechanical stressors.VSMCs similarly enable an artery's compliance to deformation and response, and VSMCs likewise increase in content with biomechanical stress from tibial to femoropopliteal regions.VSMCs proliferate and migrate to the site of biomechanical stress, synthesize collagen and produce metalloproteinases which can break down elastin (Owens et al., 2004;Peeters et al., 2015;Willis et al., 2004).A decrease in elastin composition has shown to influence FPA stiffening (Kamenskiy et al., 2015) and decreased resilience which may lead to hypertension and increased stress resulting in a further cascade of arterial remodeling (Cocciolone et al., 2018;Desyatova et al., 2017).
These structural changes to the artery wall influenced through the biomechanical environment likely affect disease development such as atherosclerosis (Van Varik et al., 2012) and the artery's response to therapeutic interventions (Kamenskiy et al., 2017).A better understanding of peripheral arteries, particularly a quantitative understanding of their distributed biochemical, cellular, and mechanical environments, will be crucial to the improved understanding, detection, and development of therapies for PAD (Ansari et al., 2013;Fortier et al., 2014).
The study of ovine femoropopliteal and tibial arteries represents a model system to examine correlations between arterial contortion and histology, but several limitations remain.Although the hindlimb arterial anatomy of sheep is similar to humans, results may not be directly translated to humans and may garner different results.Sheep stifle joints also do not have the same range of motion as the human knee, for example, full extension of a sheep hindlimb does not reach 180 as it does in human lower limbs (Faria et al., 2014).Furthermore, induced-limb flexion on a cadaveric sample lacks arterial sympathetic tone and further deformations possibly caused through skeletal muscle compression on the artery (Armentano et al., 1991;Cocciolone et al., 2018;Desyatova et al., 2017;Jaminon et al., 2019;Smulyan et al., 2016;Solan et al., 2009).Although injected with contrast, the arteries were not pressurized to mimic arterial blood flow.This may cause unnecessary buckling of the artery and inaccuracy in translating results to living models.An in-situ study may be conducted in the future utilizing living human patients to model a more accurate representation of arterial deformation in limb flexion and translatable results.Importantly, the ovine model studied here is of similar age to a young human adult and may not effectively show the prolonged effects of arterial remodeling in response to biomechanical stress (Shu & Santulli, 2018).Future work will apply quantitative largeformat analyses to models of arterial pathology or intervention.
Although this study utilized a visual representation of arterial deformation and contortion to indicate of biomechanical stress, future studies may also consider utilizing quantitative mechanical testing to accurately measure biomechanical stress (Camasão & Mantovani, 2021).Compression testing is an example that may be utilized as it yields stress-strain results in order to accurately determine the structural integrity and strength of an artery and distinguish between areas that may bare greater or lesser load (Teng et al., 2009;Zhao et al., 2018).Similarly, Jadidi et al. also compared physiologic and mechanical characteristics of the femoropopliteal artery though planar biaxial extension and constitutive modeling to calculate physiologic stretch-stress and circumferential stiffness (Jadidi et al., 2021).
Other studies have also used multi-planar biaxial extension (Kamenskiy et al., 2015), computational 3D modeling (Desyatova et al., 2017) and cyclic quasi-static extension-inflation tests (Sommer et al., 2010).These results can be compared to histology and inducedlimb flexion to further correlate variations in material composition and arterial strength.
As laboratory workflows are adapting to a digital age, quantitative image analyses are becoming an important technology in diagnostic pathology (Ozerdem et al., 2013) as they enable an automated and higher sensitivity evaluation of large amounts of normal and diseased tissue without human bias (Webster & Dunstan, 2014).In this study, we developed a large-format histology pipeline to quantitatively analyze variations in arterial flexion and elastin, VSMC, and collagen area percent composition across the femoral, popliteal and cranial tibial artery regions of ovine hindlimbs.The results suggest a significant decrease in elastin and significant increase in VSMC from the femoral to popliteal regions and from popliteal to cranial tibial (proximal to distal), consistent with current literature surrounding normal human femoropopliteal and tibial arteries without onset of arterial remodeling and pathology (Basu et al., 2010).Our results did not show significant changes in collagen content, in contrast with previous literature suggesting a decrease in collagen distally in humans (Harkness et al., 1957).However, collagen content significantly varied depending on the method of sectioning and image analysis.Some significant differences were found between transverse and longitudinal sectioning, and ROI and WA quantification methods.These variations may be explained by the potential bias to collagen content in longitudinal sectioning due to the difficulty of obtaining a true and consistent section and lack of image clarity of WA quantification.
This study employed a simple image analysis to detect elastin, VSMCs, and collagen by calculating the number of pixels within a range of RGB values.H&E stain was initially used to view the general arterial histology, VVG and MT stains were then performed to segregate proteins with different colors for analysis.We found Verhoff's Van Geisen (VVG)-stained sections were best suited to quantify all arterial components.While collagen quantification was first attempted using Mason's Trichrome (MT) staining, the yellow-orange pixels of VSMC and red-pink pixels of collagen in the VVG stain were more easily differentiated in the NIH ImageJ RGB (red, green, blue) color palette compared to the red VSMC and blue collagen in MT staining.
In the future, the use of a modified MT stain or combination VVG and MT stain may allow for clearer color segregation between elastin, VSMCs, and collagen (Garvey, 1984;Garvey et al., 1987;O'connor & Valle, 1982).For future studies, algorithms for determining protein alignment may be incorporated to further understand fiber alignment and quantity (Hernández-Morera et al., 2016;Kelly et al., 2016).For example, diffusion tensor imaging has been useful in providing the microstructural insight into changes in collagen and elastin (Tornifoglio et al., 2020).The distribution of collagen fibers have shown to be crucial in the mechanical behavior of arterial walls with the direction of these fibers being particularly important (Qi et al., 2015).Collagen is primarily responsible for dictating the mechanical properties within arterial tissue (Dahl et al., 2007), thus interpreting the alignment of collagen in future studies would benefit the understanding of the arterial morphology in association with biomechanical stressors.
An increasing number of quantitative image analysis algorithms are being developed which automatically, rapidly, and accurately char- verse and longitudinal sectioning showed some significant discrepancies particularly when quantifying collagen.This may be due to the irregularity of longitudinal sections as some were not sectioned at a true medial cross-section due to the artery being collapsed, becoming difficult to lie flat when embedding.Thus, more of the tunica media may be present resulting in possible over-quantification of proteins within the tunica media.The most consistent results were derived from transverse "ROI" sectioning, which provided the most accurate and reproducible representation of the arterial histomorphology.This may be adjusted in the future by injecting the lumen of the artery with paraffin after tissue processing and prior to embedding, this will allow for easier embedding and microtomy thus, more accurate visualization of the artery during microscopy and protein quantification.
ROI quantification is a commonly cited method within research (Avolio et al., 1998;Clarke et al., 2003;Dettmer et al., 2013;Meshram et al., 2017;Moguillansky et al., 2011), and yielded a smaller range of deviations with the use of higher-resolution and images at high magnification showing greater reliability compared to whole-section quantification.As ROI quantification methods incorporated the mean value across three "ROIs" to predict the material content across the wholesection section, this mitigated the presence of artifacts created from experimenter errors such as tissue scoring or folding from microtomy thus, provides more representative quantification measurements.A comparison between the two methods shows some statistically significant differences, particularly in VSMC and collagen quantification.
Low-resolution images, although able to show the WA section, possess fewer pixels at higher magnification.Due to this, the segregation between the red-pink color of collagen and yellow-orange VSMC is less distinct, resulting in the higher probability of overÀ/underquantification of either material.Furthermore, whole-section sections required manual removal of remaining surrounding connective tissue to avoid over-quantification of tissues not of interest (Figure S2).
Although this may be a plausible option to quantify "whole-section" images, the ROI method does not require this additional manual step as it already highlights only the tissues of interest.Furthermore, with whole-section images being in low-resolution, segregation between the tunica adventitia and surrounding connective tissue may not be distinct thus the manual deletion of only surrounding tissue may not be accurate resulting in over-or under-quantification.With further development, whole-section quantification may be utilized as a rapid automated quantification of histology in research and clinical use.

| CONCLUSION
This study pilots a quantitative image analysis method by investigating variations in tissue composition within flexible arteries, specifically between femoropopliteal and tibial regions which experience varying degrees of biomechanical stress during normal leg movement.The content of elastin decreased and the content of smooth muscle increased for more distal inferior ovine leg arteries while significant differences in collagen were not found.We compared current methods of transversely-sectioned "region-of-interest" (ROI) histology image analysis with higher-content longitudinally sectioned or "WA section".The developed method within this study may help future studies investigating the same parameters for human diseases or treatments.This study establishes a proposed method and reference to future studies when qualitatively investigating histomorphology and arterial biomechanics, providing a more thorough understanding of the artery itself and provide future applications for detection strategies and therapy design for PAD.
Ethical approvals for the use of animal material related to this project were obtained by the Queensland University of Technology (QUT) University Animal Ethics Committee (UAEC) under tissue use notifications #2000000621 and #2000000716.Femoropopliteal arterial tissue of four 18-month-old ewes (Sheep ID: 1827 (1), 1840 (2), 1832 (3), 1948 (4)) was obtained post euthanasia.Right and left femoropopliteal arteries were resected from hindlimbs of four sheep providing a total of eight samples (n = 8).All methods were carried out in accordance with relevant guidelines and regulations and in compliance with the ARRIVE guidelines (du Sert et al., 2020).
Prior to imaging, Ultravist ® contrast was injected into the right and left femoral vessels of one ovine to visualize the femoropopliteal artery in situ within the radiograph.A 5 cm cut down was initiated at the femoral triangle region to expose a portion of the proximal femoral artery, a 14G BD angiocatheter and needle (Bectin-Dickson, New Jersey, United States) were inserted into the lumen of the femoral artery.Femoral vessels were flushed with a heparin-saline solution (10,000 IU in 1 L saline) solution to displace remaining blood and dissolve potential post-euthanasia thrombi.A 25 mL aliquot of undiluted Ultravist ® (623 mg/mL of iopromide) with the addition of heparin (5000 IU in 5 mL) was injected into the femoral artery using a programmable Aladdin syringe pump (AL-1000, World Precision Instruments, Florida, United States).The proximal femoral artery and upper femoral vein were clamped and sutured closed to contain the contrast for radiograph imaging.This was performed on the right and left sheep hindlimbs.Induced limb flexion of sheep hindlimb stifle joints was performed by manually positioning the limb in full extension (100 ), full flexion (80 ), and neutral positions (85 ).Stifle joints were imaged using a Phillips Veradius C-arm X-ray (Philips Healthcare, Netherlands) to visualize in situ location and deformations or contortions of the femoropopliteal artery.Angles of flexion were determined on the radiograph using the surgical neck of femur, the patellar apex and the surgical neck of the tibia as landmarks.Induced limb flexion was performed on the right and left sheep hindlimbs and radiographically imaged at a lateral projection to best view the femoropopliteal and cranial tibial arteries.The femoropopliteal and cranial tibial arteries were then resected from the sheep for histological analyses.

F
I G U R E 1 Legend on next page.
4.4Â export magnification for transverse sections (2093 Â 1102 μm, 1.09 μm/pixel) 1.5Â export magnification for longitudinal sections (6144 Â 3235 μm, 3.20 μm/pixel), where quantification was performed on the complete section.F I G U R E 1 Microscopy image of femoropopliteal artery histomorphology using hematoxylin and eosin (H&E), Masson's Trichrome (MT), and Verhoeff-Van Gieson (VVG).Transverse segments of the femoral region of Sheep 2. (a) H&E stained segment showing a full transverse crosssection.(b) H&E staining showing collagen and elastin stained pale-pink and vascular smooth muscle cells (VSMCs) stained a more saturated pink with elongated dark-purple nuclei.(c) VVG staining showing VSMCs stained red, collagen stained blue and elastin stained pale red.(d) MT staining showing elastin stained black, VSMCs stained yellow-orange and collagen stained pink.(e) VVG stained transverse segment of Sheep 8's left femoral region.(f) VVG stained longitudinal segment of Sheep 2's femoral region captured at Â15 magnification with a scale of 1:1000 μm.Cellular components (smooth muscle, elastin, and collagen) were quantified though ImageJ (NIH) software for VVG stained sections whose pipeline is illustrated in Figure S3.Selection of the cellular components were analyzed through a color threshold using three scales: hue, saturation and brightness.The 'full arterial section' in the image was selected using thresholds hue 0-225, saturation 8-225 and brightness 0-245 for the full arterial section within the image, then quantified.To quantify elastin, thresholds of: hue 0-225, saturation 5-225 and brightness 0-115 were selected and quantified in pixels and subtracted from the full-section quantified value obtain the percentage composition of elastin within the arterial section.To quantify VSMCs, thresholds of hue 215-250, saturation 30-235 and brightness 125-225 were selected and analyzed and subtracted from the full-section quantified value to gain VSMC composition percentage.Quantification of collagen was analyzed through the combined values of hue 0-21 and 248-255 with saturation 32-235 and brightness 126-255, this value was subtracted from the full-section quantified value to provide the percentage of collagen composition.
Observed comparison between extent of limb flexion of variation and arterial wall morphology between femoropopliteal artery (FPA) regions.(a-c) Radiographic image in lateral projection depicting contortions of the femoral, popliteal, and cranial tibial regions during full flexion (a), neutral (b), and full extension (c) positions in right hindlimb of Sheep 4. Blue lines represent the femoral region, red lines represent the popliteal region.Green arrows represent the cranial tibial artery.Orange arrow depicts the disruption in contrast in popliteal region.(d-f) Transverse segment images of femoral (d), popliteal (e), and cranial tibial (f) region of Sheep 4's left artery demonstrating the three arterial wall layers: tunica intima (TI), tunica media (TM) and tunica adventitia (TA).(d) Circumferentially orientated elastin and VSMCs are observed in the TM and longitudinally orientated elastin in the TA.(e) A decrease in fine elastin and increased VSMC in TM is observed in addition to decreased elastin in the TA.(f) Increased thickness of TM with increase in VSMCs and further decrease in fine elastin and elastin in TM and TA, respectively.Images captured at Â15 magnification with a scale of 1:1000 μm.

F
I G U R E 3 Variation in elastin, vascular smooth muscle cell (VSMC) and collagen composition between whole-section sections and regionsof-interest.Comparison of elastin (a), VSMC (b) and collagen (c) composition in transverse sections between "region-of-interest (ROI)" (red line) and "whole artery (WA) quantification" (blue line) (n = 8 arteries).Comparison between elastin (d), VSMC (e) and collagen (f) composition in longitudinal sections between ROI (red line) and WA quantification (blue line) (n = 7 arteries).(a) Statistically significant variation in elastin content was observed between the femoral and cranial tibial region ( p = .0376).(

F
I G U R E 4 Comparison in clarity between low-and high-resolution microscopy for whole-section sections and regions-of-interest.Images are of transverse sections of the right femoral region of sheep 1, scale: 1:1000 μm.(a) Low-resolution microscopy image of transverse section used for "whole artery (WA) quantification" at Â4.4 magnification showing decreased clarity and distinction between different colored pixels.(b) Highresolution microscopy image of transverse section used for "region-of-interest (ROI) quantification" at Â15.0 magnification showing increased clarity and segregation of colors between pixels.(c) Low-resolution microscopy image of longitudinal section used for WA quantification at Â1.5 magnification.(d) High-resolution microscopy of longitudinal section used for ROI quantification at Â15.0 magnification.Collagen in (a) and (b) appear as long segments within tunica media of transverse sectioning whereas in longitudinal sectioning of (c) and (d), they appear as round clusters.
acterize large-format histology sections.To understand the reproducibility of arterial tissue analysis we compared the quantitation of elastin, VSMCs and collagen within (1) smaller transverse sections versus larger longitudinal sections of arteries and within (2) highmagnification region-of-interest images versus low-magnification whole-slide images.Utilizing both transverse and longitudinal sections allowed the visualization of protein alignment and further understanding of how they contribute to the robustness of an artery.Transverse sectioning is most commonly used in literature to view arterial histology, however, it is unable to show changes in histochemical content along the length of an artery on a single section.Thus, longitudinal sectioning was incorporated to examine variations tissue composition and protein orientation.Comparisons of quantification between trans-