Histological Segmentation of the Superficial Femoral Artery From Microscan to CT Using Artificial Intelligence: a Feasibility Study

What this Trial Is About

Researchers are investigating the technical feasibility of using artificial intelligence to perform histological segmentation of the superficial femoral artery from CT images. This study focuses on the femoropopliteal artery segment, which undergoes various movements and is affected by peripheral arterial disease. Different types of arterial plaques, such as calcified, fibrous, and lipidic, are important to characterize for treatment planning. Current imaging methods cannot non-invasively differentiate these lesions accurately along this artery segment, but high-resolution scanners offer near-histological tissue analysis. The study involves subjects scheduled for transfemoral amputation as part of their standard vascular surgery care. During the procedure, the femoropopliteal artery segment is harvested from amputated limbs and analyzed using a microscanner. Artificial intelligence algorithms, specifically convolutional neural networks, are trained and validated to automatically segment these artery images and correlate them with histological sections. The goal is to evaluate the algorithm's performance in segmenting artery tissue from preoperative CT scans. Participants undergo routine endovascular surgery and CT as part of their standard care before amputation. The study measures the feasibility of histological segmentation on preoperative microscanner images using AI within a one-hour timeframe. Researchers collect and compare microscanner images with histology to train and validate the algorithm. The trial is sponsored by the University Hospital in Strasbourg, France, and results will guide future larger studies to assess diagnostic capabilities of automated segmentation.

Histological Segmentation of the Superficial Femoral Artery From Microscan to CT Using Artificial Intelligence