Artificial intelligence‐assisted endoscopy changes


The relevance of endoscopic monitoring of ulcerative colitis (UC) has been translated into the new concept of “mucosal healing (MH)” as the therapeutic goal to achieve because a large amount of scientific data have revealed the favorable prognostic value of a healed mucosa in determining the clinical outcome of UC. Recent interest in MH has skewed toward not only endoscopic remission but also histological improvement (so called histological MH). Recently, several AI‐assisted endoscopic systems have been developed for assessment of MH in UC. In the future, the development of a new endoscopic scoring system based on AI might standardize the definition of MH. Therefore, “The road to an exact definition of MH in the treatment of UC has begun only now”.

Mucosal healing in UC – the subjective endoscopic score as the main stay of MH assessment: The concept of MH was introduced approximately 50 years ago. In a clinical practice, physicians have accepted the definition of MH as “the complete resolution of the visible alterations or lesions, regardless of their severity and/or type at baseline colonoscopy”. Although there have been 50 years of clinical trials and endoscopic scoring systems with different designs have been developed (Baron score, Mayo score, Sutherland, Powell‐Tuck and Rachmilewitz indices, among others), the definitions and scoring methods of these instruments have never been prospectively validated. Two new scoring systems were developed and prospectively validated: the Ulcerative Colitis Endoscopic Index of Severity (UCEIS) and the Ulcerative Colitis Colonoscopic Index of Severity. Nevertheless, no validated endoscopic score for the evaluation of UC activity evaluation reflects the complexity.

Histological MH is this ideal therapeutic goal: The necessity of the histological evaluation of the colonic mucosa of UC patients has been emphasized. Theory reported that the presence of basal plasmacytosis predicts UC clinical relapse in patients with complete MH. Peyrin‐Biroulet indicated that data indicating a prognostically relevant role for histologic activity in the mucosa of UC patients, and that finding, in addition to the macroscopic activity, has opened the door to the concept of “histological MH”, with the complete absence of clinical, laboratory, endoscopic and histological features of active inflammation. However, it remains unclear whether the histological results of rectal biopsy reflect inflammatory conditions in the entire colonic mucosa of patients with extensive UC who have achieved endoscopic remission.

Limitation of AI for the assessment of mucosal inflammation in UC: Recognize that a deeper model with a large number of parameters tends to be more accurate, and accuracy is also improved by combining the predictions of multiple models (ensemble) rather than using the predictions made by a single model. Based on these principles, a large number of images with an accurate diagnosis regarding MH are required. As mentioned earlier, many researchers have utilized a large number of endoscopic pictures evaluated by the MES, and UCEIS and tagged with the histologic results to develop an AI model for the assessment of intestinal inflammation in UC. However, there is still a problem because MH in UC lacks an exact, correct definition. Since DL is automatic, the content and direction of the learning change considerably depending on the data provided. Therefore, it is essential to select the data to be read carefully. In this regard, it remains unclear whether the present endoscopic images, endoscopic index and histologic score are suitable parameters for DL. At present, if we establish a more accurate AI model for the assessment of MH, endoscopic images combine with biological parameters such as the levels of fecal calprotectin and leucine‐rich α2 glycoprotein should be used. The development of a graphic processing unit can contribute to the improvement of the recognition of details of endoscopic images; however, this may not be enough. The addition of molecular analysis data, such as cytokine and epithelial regenerative molecule expression in the intestinal mucosa of patients with long‐term quiescent UC, as parameters for DL in the future may enable the precise identification of MH.

You can send your manuscript at  

Media Contact:

Lina James

Managing Editor

Mail Id: 

American Journal of Computer Science and Engineering Survey

Whatsapp number: + 1-504-608-2390.