TY - JOUR AU - Cristina Sanchez Montes AU - F. Javier Sanchez AU - Jorge Bernal AU - Henry Cordova AU - Maria Lopez Ceron AU - Miriam Cuatrecasas AU - Cristina Rodriguez de Miguel AU - Ana Garcia Rodriguez AU - Rodrigo Garces Duran AU - Maria Pellise AU - Josep Llach AU - Gloria Fernandez Esparrach PY - 2019// TI - Computer-aided Prediction of Polyp Histology on White-Light Colonoscopy using Surface Pattern Analysis T2 - END JO - Endoscopy SP - 261 EP - 265 VL - 51 IS - 3 N2 - Background and study aims: To evaluate a new computational histology prediction system based on colorectal polyp textural surface patterns using high definition white light images.Patients and methods: Textural elements (textons) were characterized according to their contrast with respect to the surface, shape and number of bifurcations, assuming that dysplastic polyps are associated with highly contrasted, large tubular patterns with some degree of bifurcation. Computer-aided diagnosis (CAD) was compared with pathological diagnosis and the diagnosis by the endoscopists using Kudo and NICE classification.Results: Images of 225 polyps were evaluated (142 dysplastic and 83 non-dysplastic). CAD system correctly classified 205 (91.1%) polyps, 131/142 (92.3%) dysplastic and 74/83 (89.2%) non-dysplastic. For the subgroup of 100 diminutive (<5 mm) polyps, CAD correctly classified 87 (87%) polyps, 43/50 (86%) dysplastic and 44/50 (88%) non-dysplastic. There were not statistically significant differences in polyp histology prediction based on CAD system and on endoscopist assessment.Conclusion: A computer vision system based on the characterization of the polyp surface in the white light accurately predicts colorectal polyp histology. L1 - http://158.109.8.37/files/SSB2018b.pdf UR - http://dx.doi.org/10.1055/a-0732-5250 N1 - MV; 600.096; 600.119; 600.075 ID - Cristina Sanchez Montes2019 ER -