Emerging Topics in Artificial Intelligence (ETAI) 2022

Link to full Oral

AUTHORS: Agnès Pérez-Millan, Laia Borrell,José Contador,Mircea Balasa, Albert Lladó, Raquel Sanchez-Valle, Roser Sala-Llonch.

TITLE: Classification between Early Onset Alzheimer’s disease and frontotemporal dementia using a single neuroimaging feature.

CONFERENCE: Emerging Topics in Artificial Intelligence (ETAI) 2022

PLACE: San Diego, Unitated States of America

DATES: August 21 - 25, 2022

ABSTRACT:

INTRODUCTION: Early Onset Alzheimer’s Disease (EOAD) and Frontotemporal Dementia (FTD) are common forms of early-onset dementia. Therefore, there is a need to establish accurate diagnosis and to obtain markers for disease tracking. We combined supervised and unsupervised machine learning (ML) to discriminate between EOAD and FTD patients.

METHODS: We included 3T-T1 MRI of 203 subjects under 65 years: 66 healthy controls (CTR, age: 55.0 ± 8.4 years), 85 EOAD patients (age: 57.3 ± 6.1 years) and 52 FTD patients (age: 57.9 ± 4.8 years). We obtained subcortical gray matter volumes and cortical thickness (CTh) using FreeSurfer. For ML, we performed a principal component analysis (PCA) of all volumes and Cth values. Then, the first principal component (PC) was introduced into a Support Vector Machine (SVM). Overall performance was assessed using k-fold cross-validation, in which test data was not included in any of the procedures within the pipeline (PCA and SVM).

RESULTS: Our algorithm had an accuracy of 87.2 ± 14.2 % in the CTR vs EOAD classification, 80.8 ± 20.4 % for CTR vs FTD, 66.5 ± 12.9 % for EOAD vs FTD and 65.2 ± 10.6 % when discriminating the 3 groups. We used the weights of the first PC to create disease-specific patterns.

CONCLUSION: By using a single feature that combines information from CTh and subcortical volumes, our algorithm classifies CTR, EOAD and FTD with good accuracy. We suggest that this approach can be used as a feature reduction strategy in ML algorithms while providing interpretable atrophy patterns.

Nifty tech tag lists from Wouter Beeftink