Carleton University - School of Computer Science Honours Project
Fall 2022
Classifying scans in FAST ultrasound videos using Convolutional Neural Networks
SCS Honours Project Image
ABSTRACT
The Focus Assessment with Sonography for Trauma (FAST) procedure is a commonly used ultrasound examination for identifying peritoneal traumatic free fluid in specific regions of the body. This procedure scans one of four bodily regions, the Left Upper Quadrant, the Right Upper Quadrant, the Pericardium, and the Pelvic region. Despite the demonstrated benefits FAST provides, it is not without its limitations, namely with regards to assessing the skill of users. This thesis presents a research project focused on the development of a Convolutional Neural Network architecture, that may be used for region classification of data obtained from a FAST scan. The implementation of a suitable classifier may serve to further enable the deployment of an automated skills assessment pipeline for the FAST procedure. The proposed pipeline aims to enhance the training and competency of FAST operators and improve the practicality and applicability of the procedure in clinical settings through the use of convolutional neural networks (CNNs) to classify region-specific points of interest and assess operator performance using time-series motion data. The expected outcome of the project is a step towards the development of a fully-fledged, automated skills assessment pipeline for FAST, addressing the current challenges of training and assessment in the procedure.