Automated L3-based sarcopenia quantification in CT scans

Othmane Laousy, Guillaume Chassagnon, Nikos Paragios, Marie-Pierre Revel, Maria Vakalopoulou

Keywords: sarcopenia, muscle loss, deep reinforcement learning, deep learning

TL;DR: End-to-end automated sarcopenia quantification using deep learning based techniques to facilitate clinical practice.

Abstract:

Sarcopenia refers to a skeletal muscle disorder that results in gradual and widespread muscle loss. Single-slice sarcopenia quantification is done with the localization of a vertebra first, followed by muscle segmentation.
In this paper, we present a fully automated sarcopenia assessment pipeline for CT scans, relying on powerful deep learning based techniques. Our framework consists of two steps, one that solves the detection of the appropriate CT slice using reinforcement learning and one that addresses the segmentation of the abdominal muscle mass together with the total quantification of its area.
Our pipeline has been evaluated on 100 patients, including different CT scan protocols, and reports an overall quantification performance smaller than the interobserver, indicating its potential for clinical practice.

Registration: I acknowledge that acceptance of this work at MIDL requires at least one of the authors to register and present the work during the conference.

Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.

Paper Type: novel methodological ideas without extensive validation

Primary Subject Area: Detection and Diagnosis

Secondary Subject Area: Application: Radiology

Confidentiality And Author Instructions: I read the call for papers and author instructions. I acknowledge that exceeding the page limit and/or altering the latex template can result in desk rejection.