Développement d'une méthode d'analyse histomorphométrique de lames histologiques pour l'étude des maladies cardiovasculaires

Abstract

Assess the vulnerability of atheromatous plaques is an essential issue to prevent serious consequence of atherosclerosis. The vascular ultrasound elastography is a new imaging technique, which allows a mechanical characterization of atherosclerotic plaques and to infer the vulnerability of atheroma. This project is part of a general validation of this imaging modality. The overall objective of this study is to develop tools for transverse and longitudinal histomorphometric characterization for atherosclerotic arteries sections. The results of the histomorophometric characterization could be correlated with those of the endovascular elastography (EVE) and non-invasive elastography (NIVE). For this reason we have a database of 60 histological sections from carotid arteries of three pigs, such a serial sections. Several histological stains are used, depending on the histological parameter to highlight. Histomorphometric characterization algorithms of this project are developed on Matlab. It is divided as transverse characterization algorithms and a longitudinal characterization algorithm of the arterial wall. The first allows a segmentation of the histological regions of interest, using new segmentation technique: Anamorphosis. The quantitative results are recorded in an Excel file, while the qualitative results are recorded and used in the longitudinal characterization for simultaneous display of various regions. A rigid registration algorithm is developed as a bridge algorithm to the longitudinal characterization of serial sections, or to a simultaneous display of segmentation from different histological staining. Longitudinal characterization algorithm is based on the principle of 3D reconstruction by isosurface rendering. Finally, to facilitate user interaction, user interfaces are developed for each specific objective of the algorithm. With the aim of assessing the accuracy and repeatability of the algorithms in this project several validation tests are performed. The evaluations of the segmentation algorithm show that it is not sensitive to the choice of a priori form and allows a good characterization section. The evaluation of the registration algorithm and 3D reconstruction reveals that the algorithm is sensitive to visual interpretation of the user. Although there is still place for improvement, the histomorphometric characterization tools developed in this project allow correlating data with the results of endovascular elastography and non-invasive elastography.

Date	13 Sept 2012
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Jacques A. de Guise (Supervisor) & Sophie Lerouge (Co-supervisor)