UK-Based AI Researcher Develops Vision Transformer System to Detect Early Alzheimer’s from Brain Scans


New research combines advanced artificial intelligence and explainable AI to improve early detection of Alzheimer’s disease using MRI imaging.

A UK-based artificial intelligence researcher Bassey Riman has developed a machine learning system that could help improve early detection of Alzheimer’s disease by analysing brain MRI scans.

Bassey Riman, an AI engineer and recent MSc Artificial Intelligence graduate from Teesside University, United Kingdom, has designed a system called AlzDetect, which uses a deep learning architecture known as a Vision Transformer (ViT) to classify stages of Alzheimer’s disease from neuroimaging data.

Alzheimer’s disease remains one of the most pressing neurological health challenges worldwide.

Early diagnosis is particularly difficult because the structural brain changes associated with the disease can be subtle during the initial stages.

Researchers have increasingly turned to artificial intelligence to detect patterns in medical images that may be difficult for the human eye to identify.

Riman’s research focuses on combining transformer-based machine learning models with explainable AI techniques, allowing clinicians to not only receive predictions from the system but also visual insights into how the model reached its conclusions.

“The goal is not just to build a high-accuracy model. In healthcare, it’s equally important that AI systems provide interpretable explanations so that clinicians can understand why a decision was made,” Riman said.

The model was trained using MRI data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), a large international research dataset widely used in neurological research.

The system classifies scans into four diagnostic stages: Non-Demented, Very Mild Demented, Mild Demented, and Moderate Demented.

To demonstrate how the technology could be applied in practice, Riman also developed a web-based prototype platform that allows users to upload MRI images and receive automated classifications along with attention-based visual explanations highlighting regions of the brain that influenced the model’s prediction.

The platform is intended as a research prototype, illustrating how transformer-based AI models could eventually be integrated into clinical decision-support tools.

The work forms the basis of a research manuscript currently being prepared for submission to a peer-reviewed journal with supervision from researchers at Teesside University.

Experts say the use of transformer architectures originally developed for natural language processing represents a growing trend in medical imaging research, where global spatial relationships within images can be critical for identifying disease patterns.

Beyond the research itself, Riman hopes the project demonstrates how emerging AI technologies can contribute to healthcare innovation.

“Artificial intelligence has enormous potential to assist clinicians and accelerate medical discovery. My focus is on building systems that combine technical performance with transparency so they can be trusted in real-world applications,” he said.

As interest in AI-driven healthcare solutions continues to grow, projects like AlzDetect highlight the expanding role of machine learning in addressing complex medical challenges.