Correspondence: Yasuharu Maeda Digestive Disease Center, Showa Medical University Northern Yokohama Hospital, 35-1 Chigasaki-chuo, Tsuzuki, Yokohama 224-8503, Japan E-mail: yasuharu@med.showa-u.ac.jp

Received 2025 June 11; Revised 2025 August 8; Accepted 2025 August 12.

Abstract

Ulcerative colitis, a chronic inflammatory bowel disease, is characterized by subtle microvascular alterations that play a critical role in disease perpetuation and mucosal injury. Recent advances in image-enhanced endoscopy and ultrahigh-magnification endoscopy have improved the real-time visualization of these vascular changes while highlighting their diagnostic value. Artificial intelligence (AI)-enabled endoscopic systems provide automated, reproducible vascular assessments. Emerging data suggest that AI-based vascular healing correlates with clinical remission and may alter histological scores, enabling the prediction of sustained remission. Despite these promising advances, challenges remain, such as standardizing vascular healing definitions, addressing interobserver variability, and validating AI-driven platforms in real-world settings. Integrating microvascular-targeted therapies and advanced imaging has the potential to transform the management of ulcerative colitis, facilitating sustained remission and improving the quality of life. This review examined the evolving role of microvascular assessment in ulcerative colitis, the potential of AI in refining endoscopic evaluation, and the prospects of incorporating vascular healing as a therapeutic target.

Keywords: Artificial intelligence; Endoscopy; Microvessels; Prognosis; Ulcerative colitis

INTRODUCTION

Ulcerative colitis (UC), a form of inflammatory bowel disease (IBD) that primarily affects the colonic mucosa, can profoundly impair quality of life. In recent decades, the global incidence of UC has increased, posing considerable challenges for healthcare systems.1

Traditional endoscopic assessments of UC have focused on visible mucosal features, such as erythema, ulceration, and bleeding, to determine disease activity.2 However, subtle changes in the microvasculature that precede or accompany mucosal injury have not been investigated. Recent studies have emphasized the critical role of microvasculature in the pathogenesis and perpetuation of mucosal inflammation. The intestinal microvasculature delivers immune cells and nutrients and modulates inflammatory mediators, thereby sustaining chronic inflammation in UC.3,4

Advances in imaging technologies, such as image-enhanced endoscopy (IEE) and ultrahigh-magnification endoscopy, currently allow for detailed visualization of the microvasculature5 and enable refined discrimination between inflamed and non-inflamed mucosa.6 Simultaneously, artificial intelligence (AI) is poised to transform endoscopic diagnosis and management. Deep learning and convolutional neural networks have demonstrated remarkable proficiency in object recognition, image segmentation, and predictive analytics.7,8 Moreover, the emerging concept of AI-enabled vascular healing, which directly targets the microvasculature as a treatment goal, represents a novel time- and cost-effective marker for predicting prognoses.9,10

This review synthesizes recent insights into the pathogenesis of UC, with a focus on microvascular alterations, and examines how IEE and AI-based tools contribute to a more precise assessment of the disease than other tools (Fig. 1). We also discuss an emerging strategy targeting vascular structures as a novel therapeutic avenue. Integrating advanced imaging with microvascular-focused therapies could redefine the management of UC, offering new hope for durable remission and an improved quality of life.

Fig. 1.

Representative tools for assessing vascular healing in patients with ulcerative colitis. AI, artificial intelligence; VCE, virtual chromoendoscopy.

DIFFERENT HEALING TARGETS IN UC

Previously, clinical remission was regarded as the primary treatment goal. Currently, endoscopic remission is recognized as a critical prognostic marker in the management of patients with UC,11 and is capable of predicting sustained clinical remission and resection-free survival.12 Nevertheless, in some cases of endoscopic remission, evidence of ongoing histological activity in mucosal biopsies has been observed. Consequently, in UC, a deeper form of mucosal healing, termed histological healing, has been defined, which requires the complete absence of active inflammation in mucosal biopsies.13,14 Functional barrier healing has also been proposed as an initial event in the resolution of bowel inflammation in UC.15-17 However, the necessity of repeated biopsies and the complexity of assessment methods have limited their widespread adoption in routine clinical practice. Advanced endoscopy and AI-enabled vascular healing are gaining attention as indicators that can be evaluated noninvasively and in real time without additional costs, such as those associated with biopsies.

In this review, the term “vascular healing” is used to describe the restoration of normal colonic microvascular architecture; specifically, it is a fully visible, evenly distributed capillary network without evidence of dilatation, tortuosity, or dropout. In some cases, vascular healing may refer to areas of scarred mucosa where the capillary network fails to regenerate, resulting in attenuated and barely discernible vascular structures. Vascular healing is conceptually and operationally distinct from macroscopic mucosal healing, which is defined as the absence of ulcers or erosions regardless of the microvascular status, and histological healing, which is characterized by the absence of neutrophilic infiltration on biopsy specimens (Table 1).18,19

Table 1.

Multilevel assessment of healing in ulcerative colitis: definitions, diagnostic modalities, and prognostic implications

COLONIC MICROVASCULATURE IN UC

Normal microvascular anatomy

The mucosal capillary plexus of the large intestine is organized in a regular hexagonal honeycomb pattern, encircling the mucosal glands. This intricate network is supplied by arterial branches that originate within the submucosa and extend to form subepithelial capillaries. Venous drainage occurs via venules that emerge immediately beneath the mucosal surface and subsequently converge into submucosal veins (Fig. 2).19 This fundamental vascular architecture is consistently observed throughout the large intestine, from the cecum to the rectum. However, there is some variability in the number of capillary layers surrounding each crypt within different colonic segments and even among individual specimens.20 In healthy colonic tissue, the vascular structure remains uniform and meticulously organized, with minimal evidence of leakage or inflammatory activity. These vessels play crucial roles in delivering oxygen, nutrients, and immune cells to the colonic mucosa.

Fig. 2.

In the normal mucosa, capillaries (yellow arrows) are arranged in a honeycomb-like pattern around the crypts. Ascending arterioles (red arrows) from the submucosa feed this plexus, which is drained by parallel descending veins (green arrow). Modified and adapted from Konerding et al. Br J Cancer 2001;84:1354–1362.19

Microvascular alterations in UC

A range of distinct microvascular alterations has been extensively reported in UC. The chronic inflammatory characteristic of UC drives the production of angiogenic factors, notably vascular endothelial growth factor, which stimulates angiogenesis and neovascularization.3,21 These newly formed vessels are frequently tortuous and fragile, predisposing patients to bleeding manifestations such as hematochezia. Simultaneously, the proinflammatory cytokine milieu and disruptions in endothelial junction integrity result in increased vascular permeability, manifesting as tissue edema, plasma protein extravasation, and infiltration of inflammatory cells into the mucosa. Furthermore, endothelial cells in UC display upregulated expression of adhesion molecules, including intercellular adhesion molecule-1 and vascular cell adhesion molecule-1,4,22,23 facilitating leukocyte adhesion and transmigration across the endothelial barrier.24 Sustained endothelial activation perpetuates the inflammatory cascade and contributes to progressive tissue damage.25 Additionally, the observation of microthrombi in some patients with active UC suggests that hypercoagulability and microvascular occlusion may compromise mucosal perfusion and oxygenation, thereby exacerbating local hypoxia and injury.26 Collectively, these microvascular abnormalities not only reflect disease severity but also play an active role in the pathogenesis of UC by amplifying inflammation and perpetuating mucosal damage. Disruption of the normal vascular architecture and function in UC has several downstream effects. Inadequate perfusion can lead to localized tissue hypoxia, which, in turn, exacerbates inflammation by stabilizing hypoxia-inducible factors and perpetuating cytokine production.27,28 Additionally, the enhanced expression of adhesion molecules and chemokines promotes the recruitment and accumulation of leukocytes, further intensifying mucosal injury. This compromised vascular environment also impairs wound healing, contributing to the chronic and relapsing nature of UC lesions. Understanding these microvascular alterations is important to investigate new therapeutic avenues. Strategies that modify endothelial function, inhibit pathological angiogenesis, and improve microcirculatory flow can mitigate inflammation and support mucosal healing.

ADVANCED ENDOSCOPY FOR ASSESSING MICROVASCULAR FINDINGS

Global uptake of high-definition white-light endoscopy (HD-WLE) is the standard of care for the routine endoscopic evaluation of patients with IBD. However, over the past two decades, advanced imaging modalities such as virtual chromoendoscopy (VCE) and ultrahigh-magnification endoscopy have been used to identify microvascular findings in the gastrointestinal tract. These modalities have shown promising accuracy for gauging disease activity and predicting clinical outcomes in IBD.29

VCE

VCE has emerged as a prevalent modality for detecting and characterizing early intestinal neoplasia.30 The benefits of VCE in evaluating IBD-related disease activity have been widely reported. Narrow-band imaging (NBI; Olympus), optical enhancement iSCAN (Pentax), and blue laser imaging (Fujifilm) highlight capillaries in the superficial mucosa, whereas red dichromatic imaging (Olympus) highlights blood vessels in the deeper mucosal layers.

In 2009, Kudo et al.30 appeared to be the first to report the correlation between mucosal vascular patterns and histological inflammation using NBI. Sasanuma et al.31 classified the mucosal vascular findings observed with magnified NBI of the colonic mucosa in patients with UC into three categories: honeycomb-like blood vessels (BV-H), blood vessels shaped like bare branches (BV-BB), and blood vessels shaped like vines. In 52 patients with UC, the percentage of remitted mucosa with BV-BB was 37%, while that of mucosa with scars and BV-H was 35%. BV-H and BV-BB showed no histological evidence of inflammation (12/292, 4.1% and 8/299, 2.7%, respectively), whereas vine-shaped blood vessels were strongly associated with histological inflammation (27/33, 81.8%). There was a significant correlation between magnified NBI findings and histological findings (p<0.01). Interestingly, the odds ratio for inflammatory activity at the 1-year follow-up was 14.2 for BV-BB compared with BV-H (95% confidence interval, 3.3–60.9). Guo et al.32 showed that the mucosal vascular pattern observed using NBI was predictive of potent angiogenic activity, as assessed by immunohistochemical staining for vascular endothelial growth factor, and increased microvascular density, as determined by CD31 staining. The iSCAN system has also demonstrated promising results. Neumann et al.33 reported that iSCAN significantly improved the prediction of inflammatory activity and extent, with agreement rates of 53.9% and 48.7% (p<0.01) in the HD-WLE group, and 89.7% and 92.3% in the iSCAN group (p=0.01). Iacucci et al.34-36 developed the first VCE endoscopic scoring system using iSCAN to assess inflammation in UC, and introduced endoscopic markers for mucosal and vascular healing. Among patients with a Mayo endoscopic subscore of 0, 30.4% showed an abnormal mucosal pattern, and 73.9% displayed an abnormal vascular pattern on iSCAN.

The Paddington international virtual chromoendoscopy score (PICaSSO) is a well-validated VCE-based scoring system for UC. This scoring system was designed by international experts in optical diagnosis to redefine endoscopic markers of mucosal and vascular healing.37 The PICaSSO system more accurately defines mucosal healing by integrating mucosal and vascular features and demonstrates a stronger correlation with histological scores than assessment by WLE.38 A real-world multicenter study confirmed the PICaSSO’s capacity to predict well-defined clinical outcomes at 6- and 12-month follow-ups, and it was similar to histological assessments.39 Red dichromatic imaging enhances the visualization of deep mucosal vessels in patients with mild-to-moderate UC.40 In a prospective study of 34 patients with UC, red dichromatic imaging scores showed a stronger correlation with histological activity (Spearman’s r=0.63, p<0.01) than the Mayo endoscopic subscore (r=0.48) and UC endoscopic index of severity (r=0.51).41

VCE-visualized microvascular findings are strongly correlated with histological inflammatory activity in UC and demonstrate an impressive capacity to predict clinical outcomes. However, the optimal selection and integration of these VCE modalities in routine clinical practice remain unresolved challenges.

Ultrahigh-magnification endoscopy

Ultrahigh-magnification endoscopy has emerged as a pivotal technique for “optical biopsy,” exploiting light-based principles to facilitate real-time in vivo visualization of histological structures. Two primary modalities have gained clinical acceptance: endocytoscopy (EC, CF-H290ECI; Olympus) and probe-based confocal laser endomicroscopy (pCLE, Cellvizio; Mauna Kea Technologies).28,35 EC uses a lens with a magnification of ×520 affixed to the flexible tip of an endoscope and generates color images following topical staining. In contrast, pCLE uses a probe with a magnification of ×1,000, which is inserted through the endoscope’s working channel and produces monochromatic images.42,43

A major advantage of ultrahigh-magnification endoscopy is its ability to visualize cellular and subcellular structures that remain beyond the scope of conventional IEE, such as the dynamic flow of blood cells within microvessels.44 Furthermore, EC enables detailed visualization of nuclear structures and goblet cells,45-47 whereas pCLE permits the observation of fluorescein leakage, cellular exfoliation, crypt architecture, and cellular infiltration.15,48 These advanced imaging capabilities support the comprehensive evaluation of mucosal healing and provide useful insights into the pursuit of complete remission.

EC

EC requires pre-staining with 1.0% methylene blue (±0.05% crystal violet). Several scoring systems have been developed to standardize the assessment of mucosal healing. In 2011, Bessho et al.48 introduced the endocytoscopy system score (ECSS) to evaluate glandular shape, interglandular distance, and vascular visibility, producing scores ranging from 0 to 6. The ECSS was strongly correlated with histological activity according to Matts’ score.49 During a 3-year follow-up, patients with an ECSS of zero had a relapse rate of 7.4% (2/27) versus 21.6% (8/37) for an ECSS ≥1.50

Maeda et al.50 focused on the microvascular findings observed with NBI as a simple, stain-free indicator of endoscopic activity. In 52 patients, there was a strong correlation between the microvascular patterns visualized using EC-NBI and the histological Geboes index (r=0.871, p<0.01). The sensitivity, specificity, and accuracy of EC-NBI for diagnosing acute inflammation were 84.0%, 100%, and 92.3%, respectively. With the advent of second-generation NBI, which enhances light intensity and vessel visibility, researchers have developed a more nuanced classification system using the intramucosal capillary/crypt index (Fig. 3) to evaluate vascular atypia and glandular changes. During a median follow-up of 16 months in 208 patients with UC in remission, those with an intramucosal capillary/crypt score ≤1 had a relapse rate of 5.6% compared with 30.5% in patients with an intramucosal capillary/crypt score ≥2.52 Notably, Maeda et al.51 also showed the real-time efficiency of acquiring NBI images compared with histological sampling: NBI-based EC significantly reduced the examination time from an average of 130 seconds for five stained images to just 19 seconds for five NBI images (p<0.01).52

Fig. 3.

Representative endoscopic images illustrating “vascular healing” and “vascular active” microvascular patterns. Upper panels: magnifying (×100) narrow-band imaging views; lower panels: corresponding Ultrahigh magnifying (×500) narrow-band imaging. Vascular healing (left block): normal mucosa and scarred mucosa in remission show a regular, honeycomb-like capillary network formed by uniformly thin, evenly distributed intramucosal vessels without congestion or leakage. Vascular active (right block): mildly inflamed mucosa displays an irregular, dilated, and tortuous vascular pattern with heterogeneous vessel density, focal blurring, and loss of the normal reticular architecture.

pCLE

The pCLE technique requires intravenous fluorescein administration and allows dynamic assessment and in vivo visualization of the colonic microvasculature. Under pCLE, colonic blood vessels appear as bright, fluorescein-enhanced structures with a branching or honeycomb-like architecture, depending on the mucosal pattern and inflammation status.53 Tian et al.52 reported that the capillary diameter was significantly greater and the functional capillary density area was significantly smaller in patients with UC than in healthy individuals (both p<0.01). Buda et al.53 also reported that, in 14 patients with UC and histologically inactive and quiescent disease, vascular structures showed increased tortuosity and irregularity, accompanied by a marked elevation in vascular density.54 By contrast, the remaining five patients with histologically quiescent disease displayed a vascular pattern analogous to that of the healthy control group. They found a significant difference in the median fluorescence between patients with UC and controls (3,451 vs. 2,404 pixels, p<0.01). Furthermore, 12 patients with UC and inactive disease showed significantly higher pericryptic fluorescence values than healthy controls (median fluorescence: 3,888 vs. 2,404 pixels, p<0.01).

In summary, ultrahigh-magnification endoscopy facilitates a comprehensive, real-time, in vivo assessment of disease activity and treatment responses. This technique is promising for predicting outcomes and personalizing the management of IBD. Nevertheless, the current use of ultrahigh-magnification endoscopy is largely confined to expert centers, and there are challenges such as the need for specific expertise, intravenous fluorescein administration, and the cost of pCLE probes, limiting wider clinical implementation.

Although all three platforms (VCE, EC, and pCLE) provide better visualization of the microvasculature than HD-WLE, they have various advantages and disadvantages. VCE (NBI/iSCAN/blue LASER imaging) enhances superficial capillary contrast over a wide field (approximately 50–100-µm depth) with no contact or dyes, making it ideal for whole-colon scouting and validated scores such as the PICaSSO. EC achieves optical biopsy-level detail (×520) in color after topical methylene blue application and uniquely depicts nuclei, goblet cells, and capillary loops; however, it is limited to contact points and requires brief staining and training. The highest magnification (×1,000) and dynamic fluorescein-labelled flow is yielded by pCLE, capturing barrier leakage and crypt permeability, but requires intravenous fluorescein injection, which is a disposable probe with a high cost. Therefore, VCE excels in rapid disease mapping, EC excels in non-invasive histological surrogate scoring, and pCLE in functional vascular and barrier imaging when resources permit (Table 2).30-34,37-40,48,50,51,53

Table 2.

Summary of studies that assessed microvascular findings with advanced endoscopy

Current unmet needs of advanced endoscopic techniques

Recent advancements in endoscopic imaging have enabled a more detailed and refined visualization of vascular structures in the intestinal mucosa, revealing subtle features indicative of early inflammation. Nevertheless, the assessment of these disease activity markers and their integration into existing scoring systems remains hampered by considerable subjectivity, high interobserver variability, and gaps in operator expertise and training.55 Currently, the application of these advanced modalities and their corresponding scoring systems is predominantly confined to experienced practitioners. Although advanced endoscopic diagnostics hold promise, a standardized approach to training and proficiency development is essential for their broader adoption in routine practice.53 Additionally, the cost-effectiveness of these sophisticated techniques is yet to be fully determined, raising questions about their long-term feasibility for widespread clinical use. Rigorous studies are required to confirm that advanced endoscopy-based microvascular assessments can streamline workflow without compromising diagnostic accuracy or economic viability.

In summary, advanced endoscopic technology represents a pivotal advancement in the comprehension and management of UC, enabling the visualization of vascular patterns that were previously obscured or challenging to delineate. The subsequent logical progression of this technology involves the integration of these enhanced visual insights with computational techniques, as discussed in the following section.

AI FOR ASSESSING MICROVASCULAR FINDINGS

AI-enabled endoscopy

AI encompasses a range of computational approaches designed to replicate or augment human cognitive processes. The integration of AI with advanced endoscopic technologies has the potential to incorporate the concept of vascular healing9,56 into routine clinical practice.

A computer-aided diagnosis (CAD) system known as EndoBRAIN-UC (Cybernet Systems Corp.) has received regulatory approval in Japan and has been commercially available since February 2021. This system was developed for predicting histological healing on the basis of microvascular findings observed by a 520-fold ultra-magnifying contact microscope (Endocyto: CF-H290EC; Olympus) under NBI.57 Using this system, Maeda et al.57 collected 525 validation sets of 525 independent segments from 100 patients with UC. When histological remission was defined as a Geboes score <3.1, this system achieved a sensitivity of 74%, specificity of 97%, and overall accuracy of 91% for validation samples. Subsequently, the research team implemented real-time AI analysis during colonoscopy in patients with UC in clinical remission and followed them prospectively for 12 months (n=134). Patients were stratified into two cohorts: those with AI-identified active disease and those with AI-identified healing. Clinical relapse, defined as a partial Mayo score ≥3, occurred in 28.4% (21/74) in the active disease group compared with only 4.9% (3/61) in the healing group.58 As external validation, Omori et al.58 assessed the commercially available version of EndoBRAIN-UC in a real-world clinical setting, and showed 74.2% sensitivity and 93.8% specificity in the diagnosis of a Geboes histological score <3.1.59 These external validation results were consistent with preliminary findings.57 However, the use of EndoBRAIN-UC remains limited because of the specialized knowledge required to operate an ultrahigh-magnification colonoscope. Finally, Kuroki et al.9 introduced an AI platform compatible with standard magnifying endoscopes that provides an objective binary output and classifies patients as showing “AI-based vascular healing” or “AI-based vascular activity.” They found that the 12-month post-colonoscopy clinical relapse rate was significantly higher in the AI-based vascular activity group (23.9%, 16/67) than in the vascular healing group (3.0%, 1/33; p=0.01). Interestingly, although they found a divergence between AI-based vascular healing and histological inflammatory activity, their predictive ability to maintain remission at 1-year post-endoscopy was similar. Furthermore, these investigators evaluated the clinical utility of integrating white-light imaging (WLI)- and NBI-based AI models during surveillance colonoscopy in patients with UC in clinical remission.10 Compared with the WLI model alone, this combined AI approach significantly improved specificity (from 42.2% to 61.5%, p=0.013), while it maintained sensitivity for predicting the maintenance of clinical remission over the subsequent 12 months.

The concept of endovascular healing, which is an integrated assessment of endoscopic remission and vascular healing, could be a potential treatment target for UC, similar to the concept of histo-endoscopic remission.

Bossuyt et al.59 conducted a pilot investigation into a CAD system using a single short-wavelength monochromatic light-emitting diode light source (Fujifilm) for magnifying endoscopic imaging.60 This platform enabled real-time visualization of mucosal architecture, such as crypt structures, peri-cryptal capillaries, and active bleeding sites. Consequently, the system achieved high diagnostic precision for identifying histological remission, which was defined as a Geboes score <2B.1, with a sensitivity of 79% and a specificity of 90%. When applied to non-magnifying imaging, this algorithm showed a diagnostic accuracy of 95.2% for detecting histological remission.60 This research team then aimed at improving the accuracy and applying CAD to non-magnifying images. Their new model on the entire dataset (112 patients) resulted in an accuracy of 95.2%, sensitivity of 96.4%, and specificity of 92.9% in a segment-based assessment.61

Iacucci et al.62 developed an AI-driven system to generate the PICaSSO using short video segments captured via the iSCAN platform. Their system was calibrated against histological remission benchmarks, including a Robarts histopathology index ≤3, a Nancy histologic index ≤1, and a PICaSSO histologic remission index of 0, and showed an accuracy of 83%, 81%, and 83%, respectively. Hazard ratios for adverse clinical events, such as UC-related hospitalizations, colectomy, and treatment modifications following relapse, were 2.9 for the HD-WLE model and 4.0 for the iSCAN-based model. This research team developed a novel image-processing system capable of generating iSCAN and NBI images using WLI. This system integrates AI with advanced image-processing techniques to analyze information from multiple imaging modalities simultaneously. Their multi-image analysis models, combining WLI with iScan2 and iScan3, as well as WLI with NBI, showed a superior performance to that of their respective unimodal counterparts in predicting endoscopic remission, histological remission, and future disease flare-ups.63

In summary, these data are hypothesis-generating: AI-based vascular healing offers a novel prognostic marker independent of good prognosis.64 Robust external validation across multiple centers, endoscope vendors, and treatment algorithms is still lacking, and no guidelines have endorsed vascular healing as an independent treatment-to-target endpoint. Consequently, vascular healing should currently be viewed as an exploratory biomarker that complements but does not replace endoscopic and histological remission,14,65 pending multicenter trials with consensus-based definitions.

AI-enabled digital pathology

Microvascular findings, even in histopathological specimens, may be useful in predicting treatment resistance in UC. A histopathological analysis shows the characteristics of intestinal lesions at the microscopic level, providing an explanation for the biological behavior of the disease.66,67 While histological remission is currently a focus of attention, the histological activity score does not include a microvascular index.19 However, intramucosal microvascular dilatation, as well as increased eosinophil counts and interleukin-1β expression in the intestinal mucosa, is associated with a poor response to glucocorticoid treatment.68,69 The integration of AI and digital pathology has considerably advanced the field, leading to the application of computer vision methods.70,71 These methods assist pathologists in accurately and reproducibly scoring histology, enabling precise quantification of clinically relevant features.72-75 Zhang et al.75 developed a deep learning-based digital pathology model to predict the prognosis of glucocorticoid treatment in active UC. Their validation study showed that increased numbers of inflammatory cells and intestinal mucosal microvascular dilatation were associated with a worse response to glucocorticoid therapy.76

In summary, AI facilitates the objective and accurate evaluation of microvasculature visualized through advanced endoscopy and digital histopathology. This technology enhances interobserver agreement among clinicians and supports the clinical implementation of vascular healing (Table 3).9,10,56-62,75

Table 3.

Summary of studies that assessed microvascular findings with AI-enabled endoscopy

CONCLUSIONS

With the recent widespread treat-to-target strategy, the incremental prognostic value of vascular healing needs to be confirmed in prospective multicenter studies before its incorporation into the STRIDE or other consensus frameworks. While histological remission is invasive, time-consuming, and resource-intensive, limiting its feasibility for routine monitoring in every patient,77 vascular healing can be assessed in real time during endoscopy, without requiring acquisition of a biopsy, pathologists’ labor, or associated costs.13 By contrast, barrier healing remains hampered by the high cost of pCLE probes and the potential risk of fluorescein-induced allergic reactions, posing obstacles to its clinical implementation.78,79 Notably, vascular healing is evaluated using endoscopic technologies that are already widely available and commonly used in clinical practice, without the need for supplementary staining or drug administration. However, the use of these advanced endoscopic technologies to achieve high diagnostic precision requires specialized training. In this context, integrating AI support systems could facilitate the efficient adoption of this therapeutic target.9 Advancements in IEE and AI herald an era of unparalleled diagnostic precision for UC. However, several challenges remain. First, vascular healing does not have universally accepted definitions or scoring systems readily applicable in routine practice. Although IEE has greatly improved the detection and characterization of microvascular changes, standardized classification systems across different endoscopic platforms are lacking. Establishing consensus-based scoring criteria and developing automated quantification tools can mitigate interobserver variability and enhance clinical workflow. Second, there is no clear consensus or established therapeutic algorithm that guides treatment adjustments based on vascular healing status. Therefore, clinicians are uncertain how to incorporate these findings into routine decision-making processes. To address these challenges, large-scale randomized controlled studies are required to validate the clinical utility of AI-based vascular healing in diverse patient populations. Although proof-of-concept trials have shown high accuracy in detecting inflammation and stratifying future flare-ups, real-world adoption demands a rigorous evaluation of its reproducibility and cost-effectiveness. Despite promising validation metrics, most AI-based endoscopic tools still depend on high-quality, artifact-free video inputs. Suboptimal bowel preparation, halation, or out-of-focus frames can reduce the diagnostic accuracy by as much as 20% in external validation studies. Additionally, the majority of existing models are trained using data from a single endoscope vendor and processor, limiting their performance when applied to images from different optical platforms, which is a persistent issue of cross-platform generalizability.8 Another major challenge is the “black-box” nature of deep learning classifiers. Only a minority of models provide interpretable outputs such as heat maps or attention plots, which are essential for endoscopists to verify the vascular features that influence AI predictions. Finally, model performance is susceptible to domain shifts across institutions arising from differences in patient populations, imaging protocols, or disease prevalence not represented in the original training data. These challenges indicate the urgent need for federated learning frameworks and rigorous multicenter validation. Ethical and regulatory considerations also govern the implementation of AI in clinical decision-making. Explainable AI models that generate interpretable outputs are crucial for fostering clinicians’ trust and ensuring patient safety. As AI-driven diagnostics extend into community settings, robust data governance frameworks are essential to protect patient privacy while enabling the refinement of algorithms through shared data.

Therapeutic strategies that target angiogenesis, endothelial function, or microcirculatory flow may have considerable potential, particularly in patients who do not respond to existing immunomodulatory and biological therapies. Although substantial challenges remain regarding the adoption and integration of these emerging modalities, continued research and interdisciplinary collaborations are likely to expedite their translation into routine clinical practice. Ultimately, harnessing the synergy between advanced imaging, AI-driven analytics, and targeted vascular interventions could redefine the standard of care for patients with UC.

Despite these hurdles, the combined potential of advanced endoscopy and AI-enabled vascular-targeted therapies represents a promising frontier in the management of UC. Through coordinated efforts among gastroenterologists, computer scientists, and drug developers, the next decade may witness transformative progress in achieving deeper and more durable remission in patients with UC.

Notes

Conflicts of Interest

The authors have no potential conflicts of interest.

Funding

This work was supported by a grant from JSPS KAKENHI (Grant No. JP23K09537) and the Takeda Science Foundation.

Acknowledgments

We thank Ellen Knapp, Ph.D., from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.

Author Contributions

Conceptualization: YM, TK; Data collection: YM, TK; Supervision: SK, KO; Writing–original draft: YM, TK; Writing–review & editing: YM, SK, YK, KI, JO, NO, KO, MM.

References

1. Gros B, Kaplan GG. Ulcerative colitis in adults: a review. JAMA 2023;330:951–965. 10.1001/jama.2023.15389. 37698559.

2. Limdi JK, Picco M, Farraye FA. A review of endoscopic scoring systems and their importance in a treat-to-target approach in inflammatory bowel disease (with videos). Gastrointest Endosc 2020;91:733–745. 10.1016/j.gie.2019.11.032. 31786161.

3. Li Z, Zeng L, Huang W, et al. Angiogenic factors and inflammatory bowel diseases. Biomedicines 2025;13:1154. 10.3390/biomedicines13051154. 40426981.

4. Carloni S, Bertocchi A, Mancinelli S, et al. Identification of a choroid plexus vascular barrier closing during intestinal inflammation. Science 2021;374:439–448. 10.1126/science.abc6108. 34672740.

5. Maeda Y, Kudo SE, Ogata N, et al. Use of advanced endoscopic technology for optical characterization of neoplasia in patients with ulcerative colitis: systematic review. Dig Endosc 2022;34:1297–1310. 10.1111/den.14335. 35445457.

6. Santacroce G, Zammarchi I, Tan CK, et al. Present and future of endoscopy precision for inflammatory bowel disease. Dig Endosc 2024;36:292–304. 10.1111/den.14672. 37643635.

7. Iacucci M, Santacroce G, Zammarchi I, et al. Artificial intelligence and endo-histo-omics: new dimensions of precision endoscopy and histology in inflammatory bowel disease. Lancet Gastroenterol Hepatol 2024;9:758–772. 10.1016/s2468-1253(24)00053-0. 38759661.

8. Iacucci M, Santacroce G, Yasuharu M, et al. Artificial intelligence-driven personalized medicine: transforming clinical practice in inflammatory bowel disease. Gastroenterology 2025;169:416–431. 10.1053/j.gastro.2025.03.005. 40074186.

9. Kuroki T, Maeda Y, Kudo SE, et al. A novel artificial intelligence-assisted "vascular healing" diagnosis for prediction of future clinical relapse in patients with ulcerative colitis: a prospective cohort study (with video). Gastrointest Endosc 2024;100:97–108. 10.1016/j.gie.2024.01.010. 38215859.

10. Kuroki T, Maeda Y, Kudo SE, et al. Combination of white-light imaging-based and narrow-band imaging-based artificial intelligence models during colonoscopy in patients with ulcerative colitis. J Crohns Colitis 2025;19:jjaf014. 10.1093/ecco-jcc/jjaf014. 39888722.

11. Turner D, Ricciuto A, Lewis A, et al. STRIDE-II: an update on the Selecting Therapeutic Targets in Inflammatory Bowel Disease (STRIDE) Initiative of the International Organization for the Study of IBD (IOIBD): determining therapeutic goals for treat-to-target strategies in IBD. Gastroenterology 2021;160:1570–1583. 10.1053/j.gastro.2020.12.031. 33359090.

12. Barreiro-de Acosta M, Vallejo N, de la Iglesia D, et al. Evaluation of the risk of relapse in ulcerative colitis according to the degree of mucosal healing (Mayo 0 vs 1): a longitudinal cohort study. J Crohns Colitis 2016;10:13–19. 10.1093/ecco-jcc/jjv158. 26351390.

13. Nardone OM, Iacucci M, Villanacci V, et al. Real-world use of endoscopic and histological indices in ulcerative colitis: results of a global survey. United European Gastroenterol J 2023;11:514–519. 10.1002/ueg2.12423. 37323091.

14. Li K, Marano C, Zhang H, et al. Relationship between combined histologic and endoscopic endpoints and efficacy of ustekinumab treatment in patients with ulcerative colitis. Gastroenterology 2020;159:2052–2064. 10.1053/j.gastro.2020.08.037. 32853634.

15. Rath T, Atreya R, Bodenschatz J, et al. Intestinal barrier healing is superior to endoscopic and histologic remission for predicting major adverse outcomes in inflammatory bowel disease: the prospective ERIca trial. Gastroenterology 2023;164:241–255. 10.1053/j.gastro.2022.10.014. 36279923.

16. Majumder S, Santacroce G, Maeda Y, et al. Endocytoscopy with automated multispectral intestinal barrier pathology imaging for assessment of deep healing to predict outcomes in ulcerative colitis. Gut 2024;73:1603–1606. 10.1136/gutjnl-2024-332894. 39107087.

17. Iacucci M, Maeda Y, Santacroce G, et al. AI-enabled 'endo-histo-omics': breaking down intestinal barriers in IBD. Nat Rev Gastroenterol Hepatol 2025;22:362–363. 10.1038/s41575-025-01051-1. 40033062.

18. Feakins R, Borralho Nunes P, Driessen A, et al. Definitions of histological abnormalities in inflammatory bowel disease: an ECCO position paper. J Crohns Colitis 2024;18:175–191. 10.1093/ecco-jcc/jjad142. 37607017.

19. Konerding MA, Fait E, Gaumann A. 3D microvascular architecture of pre-cancerous lesions and invasive carcinomas of the colon. Br J Cancer 2001;84:1354–1362. 10.1054/bjoc.2001.1809. 11355947.

20. Mateescu RB, Bastian AE, Nichita L, et al. Vascular endothelial growth factor: key mediator of angiogenesis and promising therapeutical target in ulcerative colitis. Rom J Morphol Embryol 2017;58:1339–1345. 29556626.

21. Takagi T, Uchiyama K, Asaeda K, et al. Association of vascular cell adhesion molecule-1 expression in colonic mucosa with mucosal inflammation and subsequent relapse in patients with ulcerative colitis. J Gastroenterol Hepatol 2025;40:1719–1727. 10.1111/jgh.17000.

22. Van den Berghe N, Verstockt B, Gils A, et al. Tissue exposure does not explain non-response in ulcerative colitis patients with adequate serum vedolizumab concentrations. J Crohns Colitis 2021;15:988–993. 10.1093/ecco-jcc/jjaa239. 33245363.

23. Iacucci M, Santacroce G, Majumder S, et al. Opening the doors of precision medicine: novel tools to assess intestinal barrier in inflammatory bowel disease and colitis-associated neoplasia. Gut 2024;73:1749–1762. 10.1136/gutjnl-2023-331579. 38851294.

24. Kruschewski M, Busch C, Dörner A, et al. Angio-architecture of the colon in Crohn disease and ulcerative colitis: light microscopy and scanning electron microscopy studies with reference to the morphology of the healthy large intestine. Langenbecks Arch Chir 1995;380:253–259. 10.1007/bf00184098. 7500795.

25. He G, Ouyang Q, Chen D, et al. The microvascular thrombi of colonic tissue in ulcerative colitis. Dig Dis Sci 2007;52:2236–2240. 10.1007/s10620-006-9158-5. 17429731.

26. Yin J, Ren Y, Yang K, et al. The role of hypoxia-inducible factor 1-alpha in inflammatory bowel disease. Cell Biol Int 2022;46:46–51. 10.1002/cbin.11712. 34658125.

27. Giatromanolaki A, Sivridis E, Maltezos E, et al. Hypoxia inducible factor 1alpha and 2alpha overexpression in inflammatory bowel disease. J Clin Pathol 2003;56:209–213. 10.1136/jcp.56.3.209. 12610101.

28. Yang YJ. Current status of image-enhanced endoscopy in inflammatory bowel disease. Clin Endosc 2023;56:563–577. 10.5946/ce.2023.070. 37793436.

29. Wanders LK, East JE, Uitentuis SE, et al. Diagnostic performance of narrowed spectrum endoscopy, autofluorescence imaging, and confocal laser endomicroscopy for optical diagnosis of colonic polyps: a meta-analysis. Lancet Oncol 2013;14:1337–1347. 10.1016/s1470-2045(13)70509-6. 24239209.

30. Kudo T, Matsumoto T, Esaki M, et al. Mucosal vascular pattern in ulcerative colitis: observations using narrow band imaging colonoscopy with special reference to histologic inflammation. Int J Colorectal Dis 2009;24:495–501. 10.1007/s00384-008-0631-9. 19145441.

31. Sasanuma S, Ohtsuka K, Kudo SE, et al. Narrow band imaging efficiency in evaluation of mucosal healing/relapse of ulcerative colitis. Endosc Int Open 2018;6:E518–E523. 10.1055/s-0044-102297. 29713677.

32. Guo T, Qian JM, Yang AM, et al. Narrow-band imaging colonoscopy to assess mucosal angiogenesis in ulcerative colitis. Gastroenterol Res Pract 2019;2019:8312624. 10.1155/2019/8312624. 30918516.

33. Neumann H, Vieth M, Günther C, et al. Virtual chromoendoscopy for prediction of severity and disease extent in patients with inflammatory bowel disease: a randomized controlled study. Inflamm Bowel Dis 2013;19:1935–1942. 10.1097/mib.0b013e318290550e. 23839228.

34. Iacucci M, Fort Gasia M, Hassan C, et al. Complete mucosal healing defined by endoscopic Mayo subscore still demonstrates abnormalities by novel high definition colonoscopy and refined histological gradings. Endoscopy 2015;47:726–734. 10.1055/s-0034-1391863. 25826276.

35. Iacucci M, Kiesslich R, Gui X, et al. Beyond white light: optical enhancement in conjunction with magnification colonoscopy for the assessment of mucosal healing in ulcerative colitis. Endoscopy 2017;49:553–559. 10.1055/s-0042-124363. 28315280.

36. Iacucci M, Daperno M, Lazarev M, et al. Development and reliability of the new endoscopic virtual chromoendoscopy score: the PICaSSO (Paddington International Virtual ChromoendoScopy ScOre) in ulcerative colitis. Gastrointest Endosc 2017;86:1118–1127. 10.1016/j.gie.2017.03.012. 28322774.

37. Trivedi PJ, Kiesslich R, Hodson J, et al. The Paddington International Virtual Chromoendoscopy Score in ulcerative colitis exhibits very good inter-rater agreement after computerized module training: a multicenter study across academic and community practice (with video). Gastrointest Endosc 2018;88:95–106. 10.1016/j.gie.2018.02.044. 29548940.

38. Iacucci M, Smith SCL, Bazarova A, et al. An international multicenter real-life prospective study of electronic chromoendoscopy score PICaSSO in ulcerative colitis. Gastroenterology 2021;160:1558–1569. 10.1053/j.gastro.2020.12.024. 33347880.

39. Naganuma M, Yahagi N, Bessho R, et al. Evaluation of the severity of ulcerative colitis using endoscopic dual red imaging targeting deep vessels. Endosc Int Open 2017;5:E76–E82. 10.1055/s-0042-119393. 28180151.

40. Hashimoto Y, Kuribayashi S, Sato K, et al. Validation of red dichromatic imaging score (RDI score) to evaluate the severity of ulcerative colitis. Surg Endosc 2023;37:3627–3633. 10.1007/s00464-022-09852-0. 36627537.

41. Maeda Y, Kudo SE, Iacucci M. Ultrahigh magnification endoscopy in inflammatory bowel disease: how do we bridge the gap between research and practice? Dig Endosc 2024;36:290–291. 10.1111/den.14678. 37737479.

42. Maeda Y, Ditonno I, Puga-Tejada M, et al. Artificial intelligence-enabled advanced endoscopic imaging to assess deep healing in inflammatory bowel disease. eGastroenterology 2024;2e100090. 10.1136/egastro-2024-100090. 39944363.

43. Kudo SE, Misawa M, Wada Y, et al. Endocytoscopic microvasculature evaluation is a reliable new diagnostic method for colorectal lesions (with video). Gastrointest Endosc 2015;82:912–923. 10.1016/j.gie.2015.04.039. 26071058.

44. Kudo T, Suzuki K, Mori Y, et al. Endocytoscopy for the differential diagnosis of colorectal low-grade adenoma: a novel possibility for the "resect and discard" strategy. Gastrointest Endosc 2020;91:676–683. 10.1016/j.gie.2019.11.029. 31785276.

45. Kudo SE, Maeda Y, Ogata N, et al. Combined endocytoscopy with pit pattern diagnosis in ulcerative colitis-associated neoplasia: pilot study. Dig Endosc 2022;34:133–143. 10.1111/den.13964. 33641190.

46. Takishima K, Maeda Y, Ogata N, et al. Beyond complete endoscopic healing: goblet appearance using an endocytoscope to predict future sustained clinical remission in ulcerative colitis. Dig Endosc 2022;34:1030–1039. 10.1111/den.14202.

47. Hundorfean G, Chiriac MT, Mihai S, et al. Development and validation of a confocal laser endomicroscopy-based score for in vivo assessment of mucosal healing in ulcerative colitis patients. Inflamm Bowel Dis 2017;24:35–44. 10.1093/ibd/izx012. 29272480.

48. Bessho R, Kanai T, Hosoe N, et al. Correlation between endocytoscopy and conventional histopathology in microstructural features of ulcerative colitis. J Gastroenterol 2011;46:1197–1202. 10.1007/s00535-011-0439-1. 21805068.

49. Nakazato Y, Naganuma M, Sugimoto S, et al. Endocytoscopy can be used to assess histological healing in ulcerative colitis. Endoscopy 2017;49:560–563. 10.1055/s-0043-106556. 28472831.

50. Maeda Y, Ohtsuka K, Kudo SE, et al. Endocytoscopic narrow-band imaging efficiency for evaluation of inflammatory activity in ulcerative colitis. World J Gastroenterol 2015;21:2108–2115. 10.3748/wjg.v21.i7.2108. 25717245.

51. Maeda Y, Kudo SE, Ogata N, et al. Endocytoscopic intramucosal capillary network changes and crypt architecture abnormalities can predict relapse in patients with an ulcerative colitis Mayo endoscopic score of 1. Dig Endosc 2020;32:1082–1091. 10.1111/den.13655. 32073691.

52. Tian Y, Zheng Y, Teng G, et al. Imbalanced mucosal microcirculation in the remission stage of ulcerative colitis using probe-based confocal laser endomicroscopy. BMC Gastroenterol 2019;19:114. 10.1186/s12876-019-1037-6. 31262270.

53. Buda A, Hatem G, Neumann H, et al. Confocal laser endomicroscopy for prediction of disease relapse in ulcerative colitis: a pilot study. J Crohns Colitis 2014;8:304–311. 10.1016/j.crohns.2013.09.005. 24094597.

54. Travis SP, Schnell D, Krzeski P, et al. Reliability and initial validation of the ulcerative colitis endoscopic index of severity. Gastroenterology 2013;145:987–995. 10.1053/j.gastro.2013.07.024. 23891974.

55. Gregorio V, Maeda Y, Kudo SE, et al. Evolving role of artificial intelligence in endoscopic management of inflammatory bowel disease: diagnosis, surveillance, and assessment. Dig Endosc 2025;37:1148–1161. 10.1111/den.15081.

56. Maeda Y, Kudo SE, Mori Y, et al. Fully automated diagnostic system with artificial intelligence using endocytoscopy to identify the presence of histologic inflammation associated with ulcerative colitis (with video). Gastrointest Endosc 2019;89:408–415. 10.1016/j.gie.2018.09.024. 30268542.

57. Maeda Y, Kudo SE, Ogata N, et al. Evaluation in real-time use of artificial intelligence during colonoscopy to predict relapse of ulcerative colitis: a prospective study. Gastrointest Endosc 2022;95:747–756. 10.1016/j.gie.2021.10.019. 34695422.

58. Omori T, Yamamoto T, Murasugi S, et al. Comparison of endoscopic and artificial intelligence diagnoses for predicting the histological healing of ulcerative colitis in a real-world clinical setting. Crohns Colitis 360 2024;6:otae005. 10.1093/crocol/otae005. 38419859.

59. Bossuyt P, De Hertogh G, Eelbode T, et al. Computer-aided diagnosis with monochromatic light endoscopy for scoring histologic remission in ulcerative colitis. Gastroenterology 2021;160:23–25. 10.1053/j.gastro.2020.09.053. 33058863.

60. Sinonquel P, Lenfant M, Eelbode T, et al. Development of an automated tool for the estimation of histological remission in ulcerative colitis using single-wavelength endoscopy technology. J Crohns Colitis 2025;19:jjae180. 10.1093/ecco-jcc/jjae180. 39602814.

61. Iacucci M, Cannatelli R, Parigi TL, et al. A virtual chromoendoscopy artificial intelligence system to detect endoscopic and histologic activity/remission and predict clinical outcomes in ulcerative colitis. Endoscopy 2023;55:332–341. 10.1055/a-1960-3645. 36228649.

62. Iacucci M, Zammarchi I, Santacroce G, et al. A novel switching of artificial intelligence to generate simultaneously multimodal images to assess inflammation and predict outcomes in ulcerative colitis-(with video). Dig Endosc 2025;37:1078–1088. 10.1111/den.15067.

63. Maeda Y, Kudo SE, Santacroce G, et al. Artificial intelligence-assisted colonoscopy to identify histologic remission and predict the outcomes of patients with ulcerative colitis: a systematic review. Dig Liver Dis 2024;56:1119–1125. 10.1016/j.dld.2024.04.005. 38643020.

64. Maeda Y, Kudo SE, Takishima K, et al. Letter: the combination of histologic remission and Mayo endoscopic score 1 as a suitable therapeutic target in ulcerative colitis. Aliment Pharmacol Ther 2021;53:955–956. 10.1111/apt.16302. 33745170.

65. Christensen B, Hanauer SB, Erlich J, et al. Histologic normalization occurs in ulcerative colitis and is associated with improved clinical outcomes. Clin Gastroenterol Hepatol 2017;15:1557–1564. 10.1016/j.cgh.2017.02.016. 28238954.

66. Yoon H, Jangi S, Dulai PS, et al. Incremental benefit of achieving endoscopic and histologic remission in patients with ulcerative colitis: a systematic review and meta-analysis. Gastroenterology 2020;159:1262–1275. 10.1053/j.gastro.2020.06.043. 32585306.

67. Zezos P, Patsiaoura K, Nakos A, et al. Severe eosinophilic infiltration in colonic biopsies predicts patients with ulcerative colitis not responding to medical therapy. Colorectal Dis 2014;16:O420–O430. 10.1111/codi.12725. 25040651.

68. Hanabata N, Sasaki Y, Tanaka M, et al. Vascular endothelial growth factor expression and microvessel parameters of colonic mucosa correlate with sensitivity to steroid in patients with ulcerative colitis. Scand J Gastroenterol 2005;40:188–193. 10.1080/00365520410010580. 15764150.

69. Puga-Tejada M, Majumder S, Maeda Y, et al. Artificial intelligence-enabled histology exhibits comparable accuracy to pathologists in assessing histological remission in ulcerative colitis: a systematic review, meta-analysis, and meta-regression. J Crohns Colitis 2025;19:jjae198. 10.1093/ecco-jcc/jjae198. 39742395.

70. Iacucci M, Maeda Y, Ghosh S. Artificial intelligence enabled histological scoring in ulcerative colitis: are we ready yet? United European Gastroenterol J 2024;12:1000–1001. 10.1002/ueg2.12579. 38761369.

71. Iacucci M, Parigi TL, Del Amor R, et al. Artificial intelligence enabled histological prediction of remission or activity and clinical outcomes in ulcerative colitis. Gastroenterology 2023;164:1180–1188. 10.1053/j.gastro.2023.02.031. 36871598.

72. Iacucci M, Maeda Y, Ghosh S. A baby step or a real giant stride: histomic enabled by artificial intelligence to predict treatment response in pediatric patients with ulcerative colitis. Gastroenterology 2024;166:730–732. 10.1053/j.gastro.2024.03.004. 38460608.

73. Ohara J, Nemoto T, Maeda Y, et al. Deep learning-based automated quantification of goblet cell mucus using histological images as a predictor of clinical relapse of ulcerative colitis with endoscopic remission. J Gastroenterol 2022;57:962–970. 10.1007/s00535-022-01924-1. 36184701.

74. Ohara J, Maeda Y, Ogata N, et al. Automated neutrophil quantification and histological score estimation in ulcerative colitis. Clin Gastroenterol Hepatol 2025;23:846–854. 10.1016/j.cgh.2024.06.040. 39059545.

75. Zhang N, Huang Y, Peng B, et al. AI-assisted glucocorticoid treatment response prediction of active ulcerative active patients. J Gastroenterol Hepatol 2025;40:1754–1762. 10.1111/jgh.16961. 40304464.

76. Plevris N, Lees CW. Disease monitoring in inflammatory bowel disease: evolving principles and possibilities. Gastroenterology 2022;162:1456–1475. 10.1053/j.gastro.2022.01.024. 35101422.

77. Neurath MF, Vieth M. Different levels of healing in inflammatory bowel diseases: mucosal, histological, transmural, barrier and complete healing. Gut 2023;72:2164–2183. 10.1136/gutjnl-2023-329964. 37640443.

78. Zammarchi I, Santacroce G, Puga-Tejada M, et al. Epithelial neutrophil localization and tight junction Claudin-2 expression are innovative outcome predictors in inflammatory bowel disease. United European Gastroenterol J 2024;12:1155–1166. 10.1002/ueg2.12677. 39361538.

79. Jahagirdar V, Bapaye J, Chandan S, et al. Diagnostic accuracy of convolutional neural network-based machine learning algorithms in endoscopic severity prediction of ulcerative colitis: a systematic review and meta-analysis. Gastrointest Endosc 2023;98:145–154. 10.1016/j.gie.2023.04.2074. 37094691.

Article information Continued

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Level of healing	Hallmark	Typical tool	Prognostic value
Clinical	Symptom resolution	PROs	Limited
Mucosal healing	No ulcers/erosions (MES, 0–1)	WLE	Relapse rates are reduced, but 15% to 30% still activity for 1 year
Vascular healing (proposed)	Restoration of the capillary network or the presence of barely discernible vascular structures within scarred mucosa	Image-enhanced endoscopy/AI	Further reduction in relapse has been reported
Histological healing	No neutrophils (Geboes score, <2; RHI, ≤3)	Biopsy	Best long-term outcomes, but invasive
Barrier healing	No fluorescein leakage	Probe CLE	Deepest but experimental

Study	Year	Study design	Modality	No. of validation samples	Outcome measures	Results
Kudo et al.30	2009	Retrospective study	NBI	157 Segments from 30 patients	Histological inflammation	Obscure MVP was correlated with higher inflammation (p<0.01)
Sasanuma et al.31	2018	Retrospective study	Magnifying NBI	112 Segments from 53 patients	Histological activity and relapse	NBI findings predicted histological activity and the future outcome
Guo et al.32	2019	Retrospective study	NBI	35 Patients with UC and 10 controls	Mucosal angiogenesis score, correlation with VEGF, and microvessel density	The NBI score was correlated with mucosal VEGF (r=0.71, p<0.01) and MVD (r=0.76, p<0.01), and the UC relapse rate was higher in patients with a high angiogenesis score
Neumann et al.33	2013	Prospective study	iSCAN	78 Patients with UC	Agreement between endoscopic and histologic assessments of the severity and extent of disease	iSCAN assessed the severity and extent of disease more accurately than WLE (p<0.01 and p=0.01, respectively)
Iacucci et al.34	2015	Prospective study	iSCAN	129 Patients	Endoscopic remission (MES=0) was compared with histological activity, while angiogenesis was evaluated with the PICaSSO	Histological activity was present in 54% of patients with an MES of 0, and residual abnormalities were detected by the PICaSSO
Trivedi et al.37	2018	Prospective study	iSCAN	30 Cases (MES, UCEIS, and PICaSSO)	Interobserver agreement	Using the PICaSSO, the ICC improved from 0.75 to 0.85 for mucosal features and from 0.62 to 0.75 for vascular features
Iacucci et al.38	2021	Prospective study	iSCAN	307 Patients	Future outcome prediction	A PICaSSO ≤3 was associated with better clinical outcomes at 6–12 months.
Naganuma et al.39	2017	Retrospective study	DRI	112 segments from 43 patients with UC	Correlations with the VAS, histological score, interobserver agreement, and prediction of relapse	The DRI was strongly correlated with the VAS (r=0.96) and histological scores (r=0.72–0.80). Interobserver agreement was substantial to almost perfect (κ=0.63–0.88). A lower DRI score predicted a longer remission (p<0.01).
Hashimoto et al.40	2023	Prospective study	DRI	191 Sites from 34 patients	Correlations of the DRI score with histology (Nancy index), UCEIS, and MES	DRI–Nancy r=0.63; UCEIS r=0.74; MES r=0.78. The DRI outperformed WLI in the histological correlation.
Bessho et al.48	2011	Retrospective study	EC	76 EC–histology pairs from 55 patients with UC	Correlation of the ECSS with Matts’ histopathological grade	The correlation was ρ=0.713 (p<0.001)
Maeda et al.50	2015	Retrospective study	EC-NBI	52 Patients with UC	Correlation with histological activity; diagnostic accuracy	The correlation was r=0.871 (p<0.01)
Maeda et al.51	2020	Retrospective study	EC-NBI	218 Patients with UC	The difference in clinical relapse-free rates between the active group and inactive patients with an MES of 1	In patients with an MES of 1, 30.5% (25/82) relapsed in the active group and 5.6% (3/54) relapsed in the inactive group
Buda et al.53	2014	Prospective study	pCLE	19 Patients with UC in remission and 19 controls	Assessment of crypt and microvascular architecture and function	Peri-crypt fluorescence (p<0.01) and the crypt diameter (p< 0.05), but not the inter-crypt distance (p=0.07), were significantly higher in patients with UC than in healthy controls

Study	Year	Study design	Modality	No. of training samples	No. of validation samples	Outcome measures	Results
Maeda et al.56	2019	Retrospective study	EC-NBI	12,900 EC images from 87 patients	525 Segments from 100 patients	Histological inflammation (Geboes score ≥3.1)	Sensitivity, 74%; specificity, 97%; accuracy, 91%
Maeda et al.57	2022	Prospective cohort study	EC-NBI	N/A	135 Patients	12-month clinical relapse (partial Mayo score ≥2)	Relapse: 28.4% (AI-active) vs. 4.9% (AI-healing) (p<0.001)
Omori et al.58	2024	Retrospective study	EC-NBI	N/A	191 Segments from 52 patients	Histological healing (Geboes score <3.1)	Sensitivity, 74.2%; specificity, 93.8%; accuracy, 77.5%
Kuroki et al.9	2024	Prospective study	NBI endoscopy	8,853 Images from 167 patients	6 Segments from 104 patients	Clinical relapse over 12 months	The relapse rate was 23.9% (16/67) in the vascular-active group vs. 3.0% (1/33) in the vascular-healing group (p=0.01); area under the curve: 0.70 vs. 0.65
Kuroki et al.10	2025	Prospective study	White-light and NBI endoscopy	N/A	102 Patients	Clinical relapse over 12 months	Specificity improved from 42.2% to 61.5% (p=0.013) with preserved sensitivity when white-light imaging and NBI AI models were combined
Bossuyt et al.59	2021	Prospective study	Monochromatic LED	N/A	113 Segments from 58 patients	Histological remission (Geboes score <2B.1)	Sensitivity, 79%; specificity, 90%, accuracy, 86%
Sinonquel et al.60	2025	Prospective study	Single-wavelength endoscopy	42 Patients (initial) and then 112 patients (final)	The same 112 patients	Histological remission (Geboes score ≤2B.0)	Sensitivity, 96.4%; specificity, 92.9%; accuracy, 95.2% (final model)
Iacucci et al.61	2023	Prospective study	White-light endoscopy and iSCAN	1,090 Videos and 67,280 frames from 283 patients	283 Patients	ER prediction, histological prediction of remission, and prediction of the clinical relapse risk	ER (UCEIS ≤1) in WLE videos showed 72% sensitivity, 87% specificity, and an AUROC of 0.85. ER was detected in VCE videos (PICaSSO ≤3), with a sensitivity of 79%, specificity of 95%, and AUROC of 0.94.
Iacucci et al.62	2025	Retrospective study	WLI, iScan2, iScan3, and NBI	2,535 Frames from 169 videos of iSCAN	iSCAN (72 videos) and NBI (51 videos)	Endoscopic/histological activity and outcome prediction	Endoscopic and histological remission, especially with different modalities combined with iSCAN (accuracy, 81.3% and 89.6%; AUROC, 0.92 and 0.89 by UCEIS and PICaSSO, respectively)
Zhang et al.75	2025	Retrospective, multicenter study	Whole-slide images	80% of 485 slides from 212 patients with UC	20% of 485 biopsy slides from 212 patients	Glucocorticoid therapy response	Areas under the curve were 0.826 (training), 0.731 (validation) and 0.725 (external). Inflammatory cells and vascular dilation were related to a poor response.