Automatic textual description of colorectal polyp features: explainable artificial intelligence

Abstract

Computer-aided diagnosis systems (CADx) can improve colorectal polyp (CRP) optical diagnosis. For integration into clinical practice, better understanding of artificial intelligence (AI) by endoscopists is needed. We aimed to develop an explainable AI CADx capable of automatically generating textual descriptions of CRPs. For training and testing of this CADx, textual descriptions of CRP size and features according to the Blue Light Imaging (BLI) Adenoma Serrated International Classification (BASIC) were used, describing CRP surface, pit pattern, and vessels. CADx was tested using BLI images of 55 CRPs. Reference descriptions with agreement by at least five out of six expert endoscopists were used as gold standard. CADx performance was analyzed by calculating agreement between the CADx generated descriptions and reference descriptions. CADx development for automatic textual description of CRP features succeeded. Gwet’s AC1 values comparing the reference and generated descriptions per CRP feature were: size 0.496, surface-mucus 0.930, surface-regularity 0.926, surface-depression 0.940, pits-features 0.921, pits-type 0.957, pits-distribution 0.167, and vessels 0.778. CADx performance differed per CRP feature and was particularly high for surface descriptors while size and pits-distribution description need improvement. Explainable AI can help comprehend reasoning behind CADx diagnoses and therefore facilitate integration into clinical practice and increase trust in AI.

Publication History

Received: 01 February 2023

Accepted after revision: 06 April 2023

Accepted Manuscript online:
11 April 2023

Article published online:
17 May 2023

• References

• 1 Subramaniam S, Hayee B, Aepli P. et al. Optical diagnosis of colorectal polyps with Blue Light Imaging using a new international classification. United European Gastroenterol J 2019; 7: 316-325
• 2 Rondonotti E, Hassan C, Andrealli A. et al. Clinical validation of BASIC classification for the resect and discard strategy for diminutive colorectal polyps. Clin Gastroenterol Hepatol 2020; 18: 2357-2365
• 3 Neumann H, Vieth M, Fry LC. et al. Learning curve of virtual chromoendoscopy for the prediction of hyperplastic and adenomatous colorectal lesions: a prospective 2-center study. Gastrointest Endosc 2013; 78: 115-120
• 4 Poon AIF, Sung JJY. Opening the black box of AI-Medicine. J Gastroenterol Hepatol 2021; 36: 581-584
• 5 Houwen B, Hassan C, Coupe VMH. et al. Definition of competence standards for optical diagnosis of diminutive colorectal polyps: European Society of Gastrointestinal Endoscopy (ESGE) Position Statement. Endoscopy 2022; 54: 88-99
• 6 van der Zander QEW, Schreuder RM, Fonolla R. et al. Optical diagnosis of colorectal polyp images using a newly developed computer-aided diagnosis system (CADx) compared with intuitive optical diagnosis. Endoscopy 2021; 53: 1219-1226
• 7 Fonolla R, van der Zander QEW, Schreuder RM. et al. Automatic image and text-based description for colorectal polyps using BASIC classification. Artif Intell Med 2021; 121: 102178
• 8 Gwet K. Kappa statistic is not satisfactory for assessing the extent of agreement between raters. Statistical methods for inter-rater reliability assessment 2002; 1: 1-6
• 9 Beaulieu D, Barkun A, Martel M. Quality audit of colonoscopy reports amongst patients screened or surveilled for colorectal neoplasia. World J Gastroenterol 2012; 18: 3551
• 10 Messmann H, Ebigbo A, Hassan C. et al. How to integrate artificial intelligence in gastrointestinal practice. Gastroenterology 2022; 162: 1583-1586