# AdaptAD: Foundation Model Adaptation with Missing Modality Robustness for Alzheimer's Disease Staging Across the Disease Continuum

DevFeed: [AdaptAD: Foundation Model Adaptation with Missing Modality Robustness for Alzheimer's Disease Staging Across the Disease Continuum](<https://devfeed.tech/articles/adaptad-foundation-model-adaptation-with-missing-modality-robustness-for-alzheimer-s-disease-staging-across-the-disease-continuum-114337.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.00084>)

Author: Dawa Chyophel Lepcha, Sophie A. Martin, Pierrick Coup\\'e, Shabbir Syed Abdul

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [lora](<https://devfeed.tech/topics/lora.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [ImageNet](<https://devfeed.tech/topics/imagenet.md>)

Tags: [3-d](<https://devfeed.tech/tags/3-d.md>), [cdr](<https://devfeed.tech/tags/cdr.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [disease](<https://devfeed.tech/tags/disease.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [foundation-model](<https://devfeed.tech/tags/foundation-model.md>), [lora](<https://devfeed.tech/tags/lora.md>), [model](<https://devfeed.tech/tags/model.md>), [oasis](<https://devfeed.tech/tags/oasis.md>), [robustness](<https://devfeed.tech/tags/robustness.md>), [staging](<https://devfeed.tech/tags/staging.md>)

## AI overview

AdaptAD stages cognitively normal, MCI, and Alzheimer's dementia from registered T1 MRI using 16 medial-temporal 2.5-D slices and a frozen ImageNet ViT-B/16 adapted with LoRA; morphometry is optional, with a learned missing token and a stacker to fuse available inputs. In five-fold ADNI evaluation (597 participants, 29% lacking FreeSurfer), it achieved macro-AUC 0.745 and significantly exceeded full fine-tuning, ShaSpec, and a 3-D CNN. On a locked OASIS-3 evaluation, macro-AUC was 0.780; the CDR-based label score was 0.804 and significantly surpassed the compared 3-D and image-tabular methods. LoRA matched full fine-tuning with 25-fold fewer trainable parameters, and omitting morphometry changed external macro-AUC by less than 0.003. The reported disease score tracked cognitive measures and stratified MCI-to-dementia progression; these findings are from the described cohorts and protocol.

## Source excerpt

arXiv:2611.00084v1 Announce Type: cross Abstract: Staging Alzheimer's disease (AD) across its continuum, from cognitively normal (CN) through mild cognitive impairment (MCI) to dementia from a single structural MRI remains difficult because 3-D networks overfit small cohorts, and FreeSurfer morphometry is often unavailable. We present AdaptAD, which reads a registered T1-weighted scan as 16 anatomically placed 2.5-D slices through the medial temporal lobe and adapts a frozen ImageNet-pretrained ViT-B/16 with low-rank adaptation (LoRA), training 3.5 M of 89 M parameters. Morphometry enters as tokens when present, or as a learned [MISSING] token, and a missing-aware stacker fuses both experts. Under a pre-registered protocol, ten methods were compared for three-class CN/MCI/AD staging on identical 5-fold splits of 597 ADNI participants, 29% without FreeSurfer, with Holm-corrected paired bootstrap tests. The locked model was tested once on 993 OASIS-3 participants (988 scored by every method). On ADNI, AdaptAD achieved a three-class macro-AUC of 0.745 (95% CI 0.709 to 0.781), ranking first and significantly exceeding full fine-tuning, ShaSpec and a 3-D CNN. On OASIS-3 it reached 0.780 (0.748 to 0.814), and on the CDR-based label (0.804) it significantly outperformed every 3-D and image-tabular method. LoRA matched full fine-tuning with 25 times fewer trainable parameters, and removing morphometry changed external macro-AUC by less than 0.003. The disease score tracked MMSE and CDR-SB within diagnostic groups and stratified MCI-to-dementia progression (log-rank p < 0.001), while saliency concentrated in medial temporal and periventricular regions. A parameter-efficient 2-D foundation model can therefore stage the full AD continuum from a routine T1 scan without morphometry, offering a scalable route to memory-clinic deployment.