A Pilot Study on an Enhanced Deep Learning Framework for Automated Amyotrophic Lateral Sclerosis Diagnosis

Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disease that poses considerable diagnostic difficulties owing to its diverse symptoms and similarities with other neurological disorders. A prompt and accurate diagnosis is necessary to enable prompt interventions and improve patient outcomes. Conventional diagnostic methods depend significantly on clinical assessment and manual analysis of MRI and other neuroimaging techniques, which are time-consuming and susceptible to interobserver variability. Recent breakthroughs in artificial intelligence (AI) and Deep Learning (DL) have shown promise for automating neuroimaging data analysis; however, standalone models often lack robustness when applied across varying datasets and modalities. To address these limitations, we propose an enhanced AI-driven deep learning framework for automated ALS diagnosis using multi-modal imaging data, such as structural MRI, DTI, and fMRI. The framework uses an ensemble design that combines Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformer-based vision models to take advantage of the complimentary characteristics of each modality. A fusion mechanism based on adaptive weighting is used to combine outputs from individual networks, improving diagnostic reliability and performance. Preprocessing steps such as intensity normalization, motion correction, and modality-specific augmentation are applied to ensure consistency across modalities. Experimental results on a curated multi-modal ALS imaging dataset demonstrate the superiority of the proposed ensemble model, achieving a classification accuracy of 97.8%, along with enhanced sensitivity and specificity compared to traditional CNN approaches. This robust framework holds promise for assisting clinicians in the early detection of ALS, ultimately contributing to more effective patient management and personalized treatment planning.