Recurrent Neural Networks in Aiding the Diagnosis of Amyotrophic Lateral Sclerosis Using Facial Expressions

• Grant number: 22/16090-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: April 01, 2023
• End date: December 31, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: João Paulo Papa
• Grantee: Arissa Yoshida
• Host Institution: Faculdade de Ciências (FC). Universidade Estadual Paulista (UNESP). Campus de Bauru. Bauru , SP, Brazil

Abstract

Early diagnosis of Amyotrophic Lateral Sclerosis (ALS) is essential for determining the start of treatments, helping to increase life expectancy and improve the quality of its patients. Facial movement analysis provides important information for recognizing the initial symptoms of ALS; however, detecting these signals is not an easy task. With the advent of computer vision and machine learning models, computational methods to aid the diagnosis of neurodegenerative diseases through videos have become a reality, generating different approaches for the detection of ALS biomarkers. By using interpretive methods, most studies end up not exploring the temporal dimension during the classification process. This work proposes to introduce sequential models of Recurrent Neural Networks (RNN) in multimedia data (videos) in such a way that the relevance of the temporal dynamics of the action units (AUs) in the identification of ALS is investigated. The proposal also contemplates the study of RNN variants as well as a BEPE internship period at the Royal Melbourne Institute of Technology University, Australia.