Novel software for automated morphometric analysis of stented arteries

Abstract

Morphometric analysis of explanted arteries remains the gold standard in assessing pathological mechanisms of arterial disease, and the efficacy of therapeutic interventions. Software currently used for morphometric analysis (ImageJ, ImagePro, etc.) requires manual tracing of each region of interest by the user to obtain direct and derived data needed for analysis. Manual segmentation of borders between differently stained arterial compartments is time-consuming and prone to bias, even when the operator is blind to the treatment type. Here, we report newly designed morphometric software, floodSeg, which greatly aids analysis through semiautomated segmentation, followed by automated computation and data entry. The program includes tools for both rapid component selection, and precision user-controlled contour correction. In practice, segmentation occurs following the selection of seed points and corresponding threshold values, for each desired component. The flood fill algorithm is used to map out components, followed by Sklanskys convex-hull algorithm to obtain the outer contours. If necessary, convexity defects can be overcome through manual point placement on top of existing points, and regeneration of the contour. floodSeg was tested using a set of non-uniformly stained stented rat arteries, and compared against manually obtained results. The accuracy of the resulting measurements was within the expected limit based on repeated manual measurements by the same operator, and did not exceed 3%. Most notably, the duration for data acquisition using floodSeg was less than 20% of the time required for manual measurements by an experienced operator. Thus, our contribution is an improvement on widely used software, with significant potential for application in a multitude of areas of pathology practice.