PreprintArticleVersion 1Preserved in Portico This version is not peer-reviewed
Exploring Autism Spectrum Disorder: A Comparative Study of Traditional Classifiers and Deep Learning Techniques to Analyze Functional Connectivity Measures from a Multicentric Dataset
Version 1
: Received: 10 July 2024 / Approved: 10 July 2024 / Online: 10 July 2024 (13:35:21 CEST)
How to cite:
Mainas, F.; Golosio, B.; Retico, A.; Oliva, P. Exploring Autism Spectrum Disorder: A Comparative Study of Traditional Classifiers and Deep Learning Techniques to Analyze Functional Connectivity Measures from a Multicentric Dataset. Preprints2024, 2024070867. https://doi.org/10.20944/preprints202407.0867.v1
Mainas, F.; Golosio, B.; Retico, A.; Oliva, P. Exploring Autism Spectrum Disorder: A Comparative Study of Traditional Classifiers and Deep Learning Techniques to Analyze Functional Connectivity Measures from a Multicentric Dataset. Preprints 2024, 2024070867. https://doi.org/10.20944/preprints202407.0867.v1
Mainas, F.; Golosio, B.; Retico, A.; Oliva, P. Exploring Autism Spectrum Disorder: A Comparative Study of Traditional Classifiers and Deep Learning Techniques to Analyze Functional Connectivity Measures from a Multicentric Dataset. Preprints2024, 2024070867. https://doi.org/10.20944/preprints202407.0867.v1
APA Style
Mainas, F., Golosio, B., Retico, A., & Oliva, P. (2024). Exploring Autism Spectrum Disorder: A Comparative Study of Traditional Classifiers and Deep Learning Techniques to Analyze Functional Connectivity Measures from a Multicentric Dataset. Preprints. https://doi.org/10.20944/preprints202407.0867.v1
Chicago/Turabian Style
Mainas, F., Alessandra Retico and Piernicola Oliva. 2024 "Exploring Autism Spectrum Disorder: A Comparative Study of Traditional Classifiers and Deep Learning Techniques to Analyze Functional Connectivity Measures from a Multicentric Dataset" Preprints. https://doi.org/10.20944/preprints202407.0867.v1
Abstract
Investigation of functional magnetic resonance imaging (fMRI) data with machine learning (ML) techniques, including also deep learning (DL) methods, have been widely used to study Autism Spectrum Disorder (ASD). This disorder is characterized by symptoms that affect the individual’s behavioral aspects and social relationships. Early diagnosis is crucial for intervention, but the complexity of ASD poses challenges for treatment development. This study compares traditional ML techniques with deep learning (DL) methods in the analysis of functional connectivity measures obtained from the time series of multicentric ABIDE dataset. Specifically, Support Vector Machines (SVM) classifiers, with both linear and Radial Basis Function (RBF) kernels, as well as eXtreme Gradient Boosting (XGBoost) classifiers, are compared against the TabNet classifier, which is a DL architecture customized for tabular data analysis and a Multi Layer Perceptron (MLP). The findings suggest that DL classifiers may not be optimal for the type of data analyzed, as their performance trails behind that of standard classifiers. SVMs achieve performances, in terms of AUC, around 75%, compared to the best TabNet and MLP results, which are 65% and 71%, respectively. Additionally, this work investigates the brain regions that contribute most to the classification task, which are found to be those primarily responsible for sensory and spatial perception, as well as attention modulation, known to be altered in ASD.
Keywords
ABIDE; Multi-site Data; Deep Learning; Machine Learning; Autism Spectrum Disorder
Subject
Physical Sciences, Applied Physics
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.