TriFusion-ADFormer: a deep learning framework for early Alzheimer’s disease detection using MRI and cognitive metrics

Description

S Sabari Vasan and P Jayalakshmi build a deep learning system that sorts people into Alzheimer disease, mild cognitive impairment, or normal cognition by fusing three data kinds. Their 2026 framework, TriFusion-ADFormer, reads structural MRI scans, turns the scan measures into a clinical text summary, and adds cognitive test scores such as memory and daily function scales.

The fused system reached 86.0 percent accuracy across the three classes, with a macro accuracy curve score of 0.93 and an F1 score of 86.0 percent. The text summaries matched known Alzheimer patterns, such as diffuse brain shrinkage, which helps doctors read why the model calls a case. The authors argue the three way fusion beats any single stream and suits early detection, when treatment can still change the course. Wider validation is the stated next step.

Catalog record

Source: https://europepmc.org/article/MED/PMC13454240