Truace
Subscribe

DTI-based AI diagnosis of Alzheimer's disease across the AD continuum

Source article: A review on diffusion tensor imaging-based comprehensive intelligent diagnosis of Alzheimer's disease

AI-assisted early diagnosis of Alzheimer's disease (AD) has substantial clinical value, and diffusion tensor imaging (DTI), which captures white matter microstructural alterations, has considerable potential across the AD continuum. However, major barriers to clinical translation remain. This review systematically evaluated the evolution of algorithmic paradigms, the effectiveness of multimodal fusion, and barriers to clinical generalization in DTI-based AD diagnosis. Using an "Input-Model-Fusion" evaluation fra…

By the Truvace desk1 min read1 sourcePeer-reviewedTRV-2026-1244· cite & verify
A review on diffusion tensor imaging-based comprehensive intelligent diagnosis of Alzheimer's disease
Hospital Universitario Doctor Peset, Valencia 01 by 19Tarrestnom65. CC BY-SA 4.0 · https://creativecommons.org/licenses/by-sa/4.0
Trace impact readingContested
P 71The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 71The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

A 2026 review in Reviews in the Neurosciences examined 98 studies from 2010 to 2026 on diffusion tensor imaging-based AI for Alzheimer's disease, using an Input-Model-Fusion framework to assess modality choice, algorithm paradigms, and fusion strategies.

The synthesis matters because it shows promising discriminative regions and task-specific utility but also documents degraded performance outside single-center settings, leaving uncertainty about standardized acquisition, representative sampling, and interpretable models needed for clinical adoption.

Main points

  1. Review analyzed 98 studies from 2010 to 2026 using an Input-Model-Fusion framework comparing sample size, task, and validation design.
  2. Conventional machine learning like support vector machines dominated small-sample studies while CNNs, graph convolutional networks, and Transformers were used with large datasets.
  3. Hippocampal pathways, corpus callosum, fornix, and cingulum showed relatively stable discriminative value across studies.

The gain

AI models using diffusion tensor imaging to detect white matter changes show potential for early Alzheimer's diagnosis with clinical value.

The problem

DTI-based AI models for Alzheimer's diagnosis suffer substantial performance degradation in multicenter or external validation, limiting clinical translation.

The rundown

The review systematically evaluated 98 DTI-based studies from 2010 to 2026, finding modality selection should be task-driven and that algorithm choice tracked sample size, with conventional methods in small studies and deep learning in large ones.

It identified persistent translation barriers including heterogeneous acquisition protocols, software pipelines, and validation strategies, and called for future work to prioritize standardized DTI processing, external validation, and neurobiological interpretability over accuracy competition.

What this doesn’t fix

Findings are constrained by heterogeneity in data sources, acquisition protocols, and processing pipelines that limit comparability and generalizability, with performance dropping in multicenter validation.

Sources

  1. Peer-reviewedReviews in the Neurosciences2026-10-01

The debate