Open Access||Research ArticleSchizophrenia Diagnosis with a Transparent Multimodal MRI Fusion Approach for Early Detection and Clinical Trust
Mohammad Shohelur Rahman1
, Ahnaf Tajwar2, Tawaz Rahman3
1 Department of Computer Science and Engineering, Barkal Ragisb Rabeya College, Barkal, Rangamati, 4570, Chittagong, Bangladesh
2 Department of Computer Science and Engineering, Brac University, Badda, Dhaka, 1212, Dhaka, Bangladesh
3 Department of Computer Science and Engineering, American International University, Khilkhet, Dhaka, 1229, Dhaka, Bangladesh
Abstract
Schizophrenia is a disorder that affects the brain. We can find it earlier and more accurately via special brain scans and computer tools. However, it is difficult to understand the information from these scans, and it is not clear how the computers make their decisions. In our study, we made a new way to use two types of brain imaging, structural magnetic resonance imaging (sMRI) and functional magnetic resonance imaging (fMRI), to better find signs of schizophrenia. Our method looks for important changes in the brain, especially how different parts of the brain work together and their structure. To make it easier to understand how our computer model works, we use explainable AI(XAI) techniques that present which parts of the brain are most important for predictions. We tested our method on a public dataset, and it worked better than the other methods did. The XAI tools provided clear pictures that show how the model makes decisions, which helps doctors and researchers trust and use our results.
Keywords
Graphical Abstract
Novelty Statement
A multimodal sMRI-fMRI and XAI framework improves schizophrenia detection by identifying interpretable structural and functional brain patterns.