Ismail Bilgen

The proliferation of non-invasive neuroimaging datasets acquired from different modalities has driven advancements in machine learning models for diagnosing brain disorders. While prior studies have primarily focused on feature engineering and model architecture improvements, they often neglect the impact of low-quality samples in training datasets, which can significantly hinder diagnostic performance. To address this, we introduce a novel sample selection framework, Influence-based Detection of Opponent Samples (IDOS), which estimates sample quality using influences approximated by the change in loss relative to a reference point. We utilized Graph Convolutional Networks (GCN) and Differentiable Graph Pooling Modules (DIFFPOOL) in IDOS using an architecture that leverages whole-brain graphs. Excluding low-quality samples identified by IDOS significantly enhanced both models’ performance compared to the baseline, yielding average improvements of 6.89% and 7.15% across accuracy, precision, recall, and specificity, for GCN and DIFFPOOL, respectively. The proposed framework offers a generalizable solution for mitigating the impact of suboptimal samples.

Autism spectrum disorder (ASD) affects the brain connectivity at different levels. Nonetheless, non-invasively distinguishing such effects using magnetic resonance imaging (MRI) remains very challenging to machine learning diagnostic frameworks due to ASD heterogeneity. So far, existing network neuroscience works mainly focused on functional (derived from functional MRI) and structural (derived from diffusion MRI) brain connectivity, which might not capture relational morphological changes between brain regions. Indeed, machine learning (ML) studies for ASD diagnosis using morphological brain networks derived from conventional T1-weighted MRI are very scarce. To fill this gap, we leverage crowdsourcing by organizing a Kaggle competition to build a pool of machine learning pipelines for neurological disorder diagnosis with application to ASD diagnosis using cortical morphological networks derived from T1-weighted MRI. During the competition, participants were provided with a training dataset and only allowed to check their performance on a public test data. The final evaluation was performed on both public and hidden test datasets based on accuracy, sensitivity, and specificity metrics. Teams were ranked using each performance metric separately and the final ranking was determined based on the mean of all rankings. The first-ranked team achieved 70% accuracy, 72.5% sensitivity, and 67.5% specificity, while the second-ranked team achieved 63.8%, 62.5%, 65% respectively. Leveraging participants to design ML diagnostic methods within a competitive machine learning setting has allowed the exploration and benchmarking of wide spectrum of ML methods for ASD diagnosis using cortical morphological networks.