Identification of radiomic biomarkers using ultrasound to diagnose sarcopenia in older adults

Systematic assessment of sarcopenia in routine clinical practice could lower functional decline and the associated socio-economic burden in the ageing population. Gold standard assessment tools are costly, and the heterogenous screening measures and protocols lack standardization and objectivity. The project aims to develop an automated, predictive model of sarcopenia based on a multimodal database including ultrasound radiomics and clinical data. At least n=65 elderly patients (75+ years) will be included from the geriatrics department of our partner hospital (CHUV). In addition, the usefulness (i.e. accuracy and reliability) of the model will be assessed as well as the feasibility to integrate explainable AI into the model. If the model proves useful, i.e. if the algorithm is accurate and reliable, we will apply for further funding allowing for the testing of a larger sample with the perspective to implement an open toolbox to guide clinical decision-making.

• Ribeiro Ricardo, Professeur ordinaire, HESAV
• Masset Jennifer, Professeur associé, HESAV

• HES-SO - Haute école spécialisée de Suisse occidentale - Rectorat

• CHUV - Centre Hôspitalier Universitaire Vaudois