Archive des Bibliothèques universitaires de Nantes

New endomicroscopic computer-based algorithm for the differential diagnosis of inflammatory bowel diseases

Par : Grégoire David

Document archivé le : 20/04/2018

Introduction : unclassified inflammatory bowel disease (UIBD) remains a contreversial subject and up to 10% of IBD patients may still not receive an optimal therapy. In this contexte, endomicroscopy appears as a promising tool; but despite an increasing number indications, a great heterogeneity of use persists and it remains an expert technique. Objectives : to define new computer-based quantitative and objective endomicroscopic bio-markers and assess the positive and differential diagnosis of IBD. Methods : We analyzed a total of 409 movies recorded among 59 patients (9 controls, 23 Crohn’s disease (CD) and 27 ulcerative colitis(UC)). The comparison between groups was done using thirteen architectural parameters: the perimeter, the sphericity, the roundness, the elongation factor, the maximal Feret diameter, the Ma/ma ratio, the density, the average and minimal inter-cryptic distance (AICD and MICD), the wall thickness, the mean area, length and diameter of the vessels and a functionnal parameter: the fluorescein leakage through the colonic mucosa (FLCM). Results : using the parameters described here and computer-based analysis, we developed two scoring systems : the IBDiag score that allowed the positive diagnosis of IBD among control patients in 100% of cases and the IBDif score for the differential diagnosis between UC and CD in remission with a sensitivity of 92.3% and a specificity of 91.3%. Conclusion : quantitative and objective parameters can diagnose IBD and differentiate UC from CD. This observation confirm the potential of endomicroscopy in this indication and the possibility to overcome the observer expertise by computerized measures. 17NANT035M


Fichier(s) associé(s) au document :
davidMED17.pdf