1. Introduction

Inflammatory bowel disease (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), has emerged as a public health challenge worldwide with increasing incidence [1]. In North America and Europe, over 1.5 million and 2 million people are affected by these diseases, respectively [2,3], who are at an increased risk of developing colorectal cancer [4]. IBD treatments are still limited, and existing drug therapy is characterized by undesirable side effects and poor long-term efficacy [5,6]. Consequently, the detection and monitoring of IBD through colonoscopic imaging and biopsy is critically important to prognosis and treatment.

Currently, the standard surveillance guideline for assessing IBD and dysplasia [5,6] is to perform colonoscopic biopsies at multiple sites followed by histopathology analyses [7,8]. This procedure is inherently subject to sampling errors. While the increasing use of chromoendoscopy provides enhanced visual guidance for targeted biopsy [9,10], it is still sub-optimal, in particular on small and flat lesions (which are difficult to discern even under colonoscopy with enhanced contrast) and on deep-mucosal lesions (which are endoscopically invisible).

IBD surveillance and biopsy guidance can potentially be improved by employing colonoscopic OCT, which enables in situ and real-time visualization of the 3-dimensional (3D) microstructures of the colonic wall [11–17]. However, conventional colonoscopic OCT operating at 1300 nm with resolution of ∼6.5 µm in tissue is unable to detect tiny lesions or differentiate subtle microscopic features [12,13,16], and its ability to detect precancerous or early neoplastic lesions associated with IBD is limited [18]. There remains a critical need for the development of colonoscopic OCT with a higher resolving power capable of delineating the fine dysplastic changes in colonic mucosa [14,19–22]. And because the lesions are typically diffuse and multifocal in IBD patients [18], there is an urgent need for a method of directly visualizing and quantifying mucosal lesions (both endoscopically visible and invisible) over a large area in real-time, to enable precise guidance of colonoscopic biopsies.

In this study, we utilized UHR (∼1.7 µm in tissue) colonoscopic OCT for intraluminal imaging of the colonic microstructures in 2,4,6-trinitrobenzene sulfonic acid (TNBS)-induced colitis in live rats. We established a method for identifying characteristic microscopic features of the normal and inflamed colon in OCT images which were confirmed with corresponding histopathological images. In addition, quantitative optical attenuation maps generated from intensity-based OCT images showed a promising potential for discerning diseased colonic mucosa in a rat model of IBD. Our study demonstrates a promising potential of UHR colonoscopic OCT for enabling a far more accurate IBD pathological analysis than conventional colonoscopy.

2. Materials and methods

2.1 UHR colonoscopic OCT

OCT is a high-speed, nonionizing modality capable of 3D imaging of biological tissue microanatomy at microscale resolution in vivo [23]. In our study, the UHR colonoscopic OCT platform consisted of a homemade 800-nm spectral-domain OCT (SD-OCT) engine and a miniature achromatic OCT catheter (Fig. 1(a)) [20–22]. In the SD-OCT system, a homemade broadband rotary joint was used [24]. In the catheter, a diffractive lens was introduced in the distal end to minimize the chromatic aberration over the broad spectrum. The basic distal end optics design was similar to the one reported in Ref. [20]. In essence, the achromatic catheter has an outer diameter of 1.3 mm (Fig. 1(b)), and offers an axial resolution of ∼2.4 µm in air (about 1.7 µm in tissue) and a transversal resolution of ∼6.8 µm. This catheter provides a full circumferential view (without blocking struts as in distal scanning catheter) and a proper diameter for imaging rat colon. In vivo volumetric imaging was performed by pullback of a circumferentially rotating OCT catheter within a stationary plastic sheath with a computerized linear translation stage at the proximal end of the probe. The pullback speed could be manipulated to control the separation (i.e., pullback pitch) between two adjacent circumferential images; 20 µm was used in our study. The catheter was inserted about 7 cm in depth from the anus and a 3-cm long colon was then imaged for each rat, which took about 75 seconds at an imaging speed of 20 frames per second.

 

Fig. 1. UHR colonoscopic OCT platform. (a) Schematic of the OCT system, which consists of a homemade Ti:sapphire laser (with a center wavelength around 820 nm and a 3-dB bandwidth of ∼150 nm), an ultra-broadband fiber coupler (Gould Fiber Optics), and a home-built linear-k spectrometer. The homemade broadband rotary joint was mounted on a linear translational stage, enabling pullback during circumferential beam scanning for performing 3D volumetric imaging. (b) Photograph of the miniature achromatic OCT catheter (outer diameter: 1.3 mm) encased in a transparent plastic sheath (outer diameter: 1.8 mm). The imaging laser was focused at about 200 µm out of the plastic sheath.

Download Full Size | PDF

2.2 TNBS-induced IBD model in rats and experimental design

Male Sprague-Dawley (SD) rats of the same age (10 weeks) were used for this study. 11 SD rats were randomly divided into two groups: 4 rats in the control group and 7 rats in the IBD group (Table S1 in Supplement 1). TNBS-based IBD model in rodents is well-established and has been widely studied [25–31]. Following the classical protocol for IBD rodent models, colitis was induced at day-0 via intrarectal administration of TNBS (Sigma-Aldrich, Saint-Louis, USA) at a dose of 80 mg/kg in 40% ethanol (0.75 ml), at an insertion depth of ∼7 cm from the anus. Four rats were injected with 0.75 ml ethanol, serving as the control group. Within a 14-day study period, OCT imaging was performed to assess the development of IBD at 4 different time points, i.e., day 0, 3, 8 and 13, with at least one rat sacrificed at each time point after imaging (Table S1). For the IBD group, seven rats with TNBS-induced colitis were imaged at day 3, five rats were imaged at day 8, and four rats were imaged at day 13. The Disease Activity Index (DAI) of the rats was scored on a daily basis from day 0 to day 13. DAI is the combined score of weight loss, stool consistency, and rectal bleeding (Table S2), which was found in previous studies to correlate well with pathological findings in the TNBS-induced IBD model [28,32].

Before each imaging procedure, a standard rat colon cleansing procedure was performed, including (A) the relocation of rats to wired-bottom cages, (B) a 24-hour fasting period with free access only to water, and (C) a manual procedure to remove rectal fecal pellets (if needed) to eliminate all fecal residues. Prior to OCT imaging, rats were first anesthetized by inhalation of 1%-2% isoflurane-oxygen mixture. The OCT catheter was then inserted from the rat’s anus to the left part of the colon with a total insertion depth of ∼7 cm. During imaging, anesthesia was maintained via continuous isoflurane inhalation. After imaging, rats were returned to the housing facility, and to their normal diet and cages, until the next imaging procedure. At least one rat was sacrificed at each time point after imaging in order to obtain the corresponding histology. The rat IBD model and in vivo OCT imaging protocol were approved by the Animal Care and Use Committee of the Johns Hopkins University.

2.3 Imaging registration and histological correlation

After OCT imaging, the plastic sheath was kept in colon to register the imaged tissues while the imaging catheter was retracted from the sheath. The rat was then sacrificed, and the imaged colon was harvested. The tissue specimen with the plastic sheath inside the colon was placed in 10% neutral buffered formalin overnight. Afterwards, the plastic sheath was removed, and the fixed colon specimen was longitudinally divided into 3-4 small sections for histological processing and correlation with OCT images (Fig. S1 in Supplement 1). To match the OCT cross-sectional images, the corresponding colon tissue was orientated and sectioned accordingly. Each slide was sectioned at 5-µm in thickness followed by standard hematoxylin and eosin (H&E) staining.

2.4 Depth-resolved optical attenuation

The conventional depth-resolved optical attenuation algorithm has been previously described and widely adopted to extract optical properties of tissues from OCT intensity images to help differentiate the diseased tissues from the normal ones [33]. In order to remove the influence of the depth-dependent effects of a focused imaging beam profile on the calculation results, we adopted a previously proposed method to normalize the OCT signal with a tissue phantom of a known attenuation coefficient [34,35]. By following a well-established procedure [36], the phantom was made of titanium dioxide (TiO₂) nanoparticles (32 mg dissolved in acetone) embedded in room-temperature-vulcanizing silicone (40 ml base and 4 ml catalyst). The resulted phantom had a measured attenuation coefficient of ∼0.75 mm^-1, which was consistent with previously reported experimental results [36]. OCT imaging of both the phantom and tissue was then performed under the same experimental conditions.

After normalization of the tissue OCT signals with the phantom OCT signals [34,35,37], the depth-resolved attenuation coefficients for each A-line were calculated with the following equation:

2.5 Statistical analysis

To correlate the measurements of mucosal thickness based on OCT and histology, we performed a two-sample, two-tailed Student’s t-test of unequal variance with the hypotheses that both groups provide close measurements. For intergroup analyses on the normalized mucosal thickness and attenuation values of the IBD group, we performed a two-sample, two-tailed Student’s t-test of unequal variance with the hypotheses that the colitis activities result in different mucosal thickness and tissue optical properties. To correlate the OCT metrics with DAI scores, we performed a two-sample, two-tailed Student’s t-test of unequal variance with the hypotheses that both OCT metrics and DAI scores provide comparable evaluation on the colitis activity in colon. A linear-regression model was used to compare OCT metrics with DAI scores and the mucosal thickness measured from OCT images with those from histological images.

3. Results

3.1 Imaging normal colonic microstructures with UHR colonoscopic OCT

The performance of UHR colonoscopic OCT for in vivo volumetric imaging of normal rat colonic tissues was demonstrated in Fig. 2 with a 10-mm long cutaway view. An en face intensity projection OCT image, which is similar to that obtained with a magnifying endoscope, can be generated from the 3D OCT dataset by using the sum of OCT axial (z direction) intensity along with the imaging depth. Figure 2(a) is an en face image overlaid on the inner surface of the 3D colon image, where groups of colonic crypts and normal vascular pattern (red arrowheads) are visible. As shown in the 2X enlarged en face image (Fig. 2(b)), the well-organized, tightly packed cryptic structure was clearly delineated. Crypts visualized as bright regions between dark bands (lamina propria) in OCT were well correlated with corresponding histological images. Crypt size differed slightly in the corresponding OCT and histological images, which could be due to differences between crypt orientation in the OCT image and histological planes. Sample fixation during histological processing may also have caused slight shrinkage in tissue size (Fig. S2 in Supplement 1).

 

Fig. 2. OCT images of normal rat colon with correlated histology. (a) Cutaway view of the 3D-rendered 10-mm long normal rat colon with an en face OCT image overlaid on the inner surface. The en face image was reconstructed by axially summing the OCT intensity over a 600-µm depth starting from the colon surface. (b) 2X enlarged view of the en face image (corresponding to the blue dashed box in (a)) and its correlated histology (Note: two black lines in histology are artifacts introduced during the tissue processing). Red arrowheads indicate the blood vessels. (c) A longitudinal OCT image (corresponding to the red dashed box in (a)) and its correlated H&E histology. Inset: 2X enlarged views (black and yellow boxes). (d) A representative cross-sectional OCT image (corresponding to the green dashed line in (a)) with a 2X enlarged view and the associated histology. Yellow arrowhead indicates the crypt. Blue arrowhead indicates the surface of the epithelium. CM: colonic mucosa; MM: muscularis mucosa; SM: submucosa; MP: muscularis propria.

Download Full Size | PDF

Figure 2(c) shows a longitudinal-sectional OCT image and Fig. 2(d) shows a cross-sectional OCT image of the colonic wall. Both of them reveal 4 layers of microstructures, including colonic mucosa (CM, corresponding to histological colonic epithelium, crypts, and lamina propria), muscularis mucosa (MM), submucosa (SM, demonstrated between MM and muscularis propria), and muscularis propria (MP). In the layer of CM, colonic crypts were distinguished as white regions surrounded by black column-like bands of lamina propria, consistent with previous in vivo studies on small animals [14,15,24,38,39]. In addition, we could appreciate the characteristic tailing patterns (i.e., shaft-like shadowing effects) of crypts in the cross-sectional image, which manifest the well-organized cryptic structure and serve as a good indicator of the integrity of colonic mucosa [40]. Both 1X and 2X longitudinal-sectional images (Fig. 2(c)) and 2X cross-sectional image (Fig. 2(d)) show a strong correlation between OCT images and corresponding histological images. Moreover, excellent correlation (r = 0.98, p < 0.001) was also observed between OCT images and histological images, in the form of mucosal thickness measurements (Fig. S2 in Supplement 1).

3.2 Imaging and identifying the characteristic pathological features of IBD

To demonstrate the capability of UHR colonoscopic OCT to image pathologic colonic changes in IBD, a rat model was used (see Materials and methods). In our study, DAI was scored according to Table S2 in Supplement 1 [32,41], and disease development in the IBD group as indicated by DAI scores (Fig. S3) was consistent with previous studies of TNBS-induced colitis models in small animals [27,28,32]. Compared with normal colons (Fig. 3(a)), we observed remarkable microstructural changes in the en face OCT images of colons with IBD (Figs. 3(b) and 3(c)), including disruption and loss of vascular patterns, disorganized crypts with irregular size, and patchy loss of crypts (denoted by red asterisks).

 

Fig. 3. Comparison of representative en face OCT images between normal (a) and IBD (b, c) rat colon. OCT images with respective enlarged views (red dashed box) and cross-sectional views (blue dashed line and box) were acquired from the same rat at day 0 (a), day 3 (b), and day 8 (c) after the induction of IBD with TNBS, respectively. Axial summation of the OCT intensity over a 600-µm depth starting from the colon surface was used to generate the en face images. Red asterisks indicate the patchy loss of crypts. Red arrowheads indicate the ulceration-like abnormal tissues. Red arrows indicate tissue folding.

Download Full Size | PDF

With an increased resolution, UHR colonoscopic OCT can help resolve and identify the layered colonic structure and lymphoid follicle. The latter is demonstrated by a reflective (dark), clearly defined boundary area immediately beneath the MM layer in OCT images (Fig. 4(a)) and confirmed by the corresponding histology (Fig. 4(b)). In addition, our study demonstrated that UHR colonoscopic OCT can identify some key pathological features of IBD in vivo (Figs. 4(c)-(l)), compared to the OCT images from the normal colon (Fig. S4 in Supplement 1). As shown in Fig. 4(c) and its inset, the disorganized, irregularly enlarged and delimited cryptic areas within CM in the OCT image are correlated with crypt distortions and dropouts in corresponding histological micrographs (Fig. 4(d)).

 

Fig. 4. Comparison of cross-sectional OCT images between normal (a and b) and IBD (c - l) rat colon with corresponding histology. Normal colonic structures in OCT (a) with the correlated histology (b) versus the typical microscopic features of IBD pathologies observed in OCT (c, e, g, i, k) and the correlated H&E histology (d, f, h, j, l), including distortion of crypt architecture (c, d), dilated crypt and microabscess (e, f, g, h), ulceration and inflammation (i, j, k, l). CM: colonic mucosa; MM: muscularis mucosa; SM: submucosa; MP: muscularis propria.

Download Full Size | PDF

Other pathological features of IBD, such as dilated crypts, could also be clearly identified in OCT images as characteristic empty void structures in CM (Figs. 4(e) and 4 g). Moreover, Figs. 4(e) and 4 g along with the insets also show hyper-reflective structures in the voids. The enlarged views correlate well with the histological images in Figs. 4(f) and 4 h, which show dilated cryptic lumens with multiple neutrophils, i.e., micro-abscesses, a key pathological change in IBD. As the severity of IBD increased, ulcerations with dense inflammatory infiltration manifesting as hyper-reflective surfaces followed by dark bounded areas in CM could be found in OCT images (Figs. 4 g, 4i, 4k, and the insets). These OCT features of ulcerations and inflammation are consistent with previous studies of the human colon and confirmed in the correlated histology (Figs. 4(j) and 4 l) [17].

3.3 In vivo assessment of IBD pathology with quantitative optical attenuation map

Complementary to the conventional OCT intensity image, tissue optical property (i.e., optical attenuation coefficient) can be directly derived from the same OCT image dataset by using an established depth-resolved attenuation algorithm (see Materials and methods) [33,42]. A method for performing real-time acquisition and visualization of OCT attenuation maps for intraoperative tissue differentiation was demonstrated in our previous studies [35,37]. As shown in Fig. 5(a), a representative color-coded en face attenuation map for a rat colon with IBD (overlaid on the unwrapped 3D intensity image) demonstrates a heterogeneous attenuation distribution. This attenuation map of colonic mucosa (Fig. 5(b)) provides a detailed delineation of lesions, such as continuous ulcerations (red asterisks), which is comparable to those observed under a magnifying colonoscopy with discernable contrast. It was found that continuous ulcerations exhibited a dark color with an attenuation value higher than 5 mm^-1. As verified in the OCT cross-sectional image (Fig. 5(c)), the dark red dots in Fig. 5(b) (indicated by the blue arrow in the 4X magnified view of the red dashed box area) are dilated crypts located beneath the mucosa surface. The dilated crypts exhibited a lower attenuation coefficient (about 1.5 mm^-1) than that of the surrounding normal mucosa (about 3.3 mm^-1). Notably, crypt distortion and micro-abscesses can also be found in both the OCT image (Fig. 5(c)) and its correlated histology (Fig. 5(d)). Furthermore, the correlated OCT and histology pair in Figs. 5(e) and 5(f) verified the congestions and erosions in mucosa (indicated by red arrowheads) in the attenuation map (see the magnified view of the black dashed box area in Fig. 5(b)), showing an attenuation property (about 4.3 mm^-1) higher than that of the surrounding normal tissue.

 

Fig. 5. Quantitative optical attenuation map of IBD rat colon imaged at D8 of a DAI score 4.6. (a) The color-coded en face attenuation map overlaid on an unwrapped 3D intensity image of a 30-mm long rat colon with IBD. (b) The en face attenuation map is generated by summing the attenuation values along the depth over a 210-µm thick mucosa layer. The enlarged view of the red box area shows the region with dilated crypts, while the enlarged view of the black box area illustrates the mucosa of congestions and erosions. Red asterisks indicate continuous ulceration, blue arrows indicate dilated crypts, and red arrowheads mark the congestions and erosions in mucosa. (c) OCT cross-sectional image (black dashed line in (b)) shows the ulceration (red asterisks) with inflammation and dilated crypts (blue arrow), and distorted crypts and microabscesses, verified in the enlarged view and correlated histology (d). (e) The OCT image (blue dashed line in (b)) shows the multifocal congestions and erosions in mucosa (red arrowhead), confirmed by the enlarged view and associated histology micrograph (f).

Download Full Size | PDF

4. Discussion and conclusions

It was demonstrated that colonoscopic OCT operating at 1300 nm could provide 3D depth-resolved imaging of colon in IBD patients and potentially help differentiate normal from pathologic colorectal tissues in vivo [12,13,16,17]. However, this technology is still suboptimal for the accurate delineation of the microstructures and subtle pathological changes in colonic mucosa mainly due to its limited resolution (∼6.5 µm in tissue), while it is well-known that pathological changes induced by IBD affect primarily colonic mucosa. As a result, a pertinent, non-invasive means of evaluating IBD disease activity/progression and developing effective treatment strategies remains hindered [5,6,43].

With an improved imaging resolution and contrast, our study demonstrates the capability of 800-nm UHR colonoscopic OCT to provide an accurate delineation of the fine microstructures of the rat colon, such as mucosa (crypts, lamina propria, and muscularis mucosa), submucosa, and muscularis propria. In conjunction with a rat IBD model, we further demonstrated that this technology can identify the key pathophysiological characteristics of IBD in vivo, such as mucosal erosion, congestion, crypt distortion, dilation, microabscess, and ulcerations [7,12,13,16,27]. In addition, we have shown that our advanced high-resolution imaging technique enables direct visualization and quantification of the pathologic changes of colonic mucosa over a large field of view (FOV) in vivo with a color-coded optical attenuation map. Moreover, our pilot study demonstrates that the disease activity of IBD (as indicated with DAI scores) in rats was closely related with the colonic mucosa thickness (r = 0.92, p < 0.001) and the mucosal attenuation coefficient (r = 0.93, p < 0.001), respectively (Fig. S5 in Supplement 1).

The capability of UHR colonoscopic OCT to resolve subtle pathological changes in colon and evaluate the disease activity has the potential to enable the study of IBD in vivo, providing an intravital imaging tool for effectively elucidating IBD pathogenesis and evaluating patient responses to treatment. Further, while conventional colonoscopy is limited to examination of the surface features of mucosa [5,6], volumetric visualization and quantification of subsurface pathologies of colon with UHR colonoscopic OCT provides a direct visual cue of mucosal lesions over a large area, which may enables precisely guided, more efficacious colonoscopic biopsies. Patients with IBD face an increased lifetime risk of developing colorectal cancer [4,18]. Accurate detection of subsurface pathologies can potentially advance disease surveillance well beyond what is possible through conventional colonoscopic screening, ultimately reducing the possibility of tumor pathogenesis in IBD patients [18].

The current study only involved a small animal (rodent) model, which demonstrated the operational feasibility of the technology for assessing IBD associated features in vivo. A systematic study of IBD patients is imperative to validate the clinical potential of using UHR colonoscopic OCT to assess the pathology of IBD, alongside the histopathology of biopsy samples. For future clinical use, UHR colonoscopic OCT will benefit from several technological advances. First, a faster imaging speed is more desirable in order to image a much longer segment of the human colon and to minimize motion artifacts. In principle, the imaging speed of SD-OCT system can be improved by developing a fast fiber-optic rotary joint (not commercially available) for proximal scanning catheter or employing a fast micromotor for distal scanning catheter, in addition to a fast spectroscopic OCT detector (i.e., imaging spectrometer). Second, the development of a balloon catheter is needed for clinical use to dilate the folded colonic tract and help stabilize the imaging probe in colon. Third, an accurate and automated image assessment method is highly desirable to help process the large volume of diagnostic OCT images [40], which will further enhance the clinical viability of UHR colonoscopic OCT in performing objective evaluation of IBD in human. Last but not the least, as patients with IBD always exhibit visible bleeding upon colonoscopy examination and the exist of blood adversely affects the OCT imaging quality, a rinsing procedure with saline before the OCT imaging may help mitigate the problem.