Research map / Vision-language and multimodal / Cross-modal learning
Joint Adaptation of Uni-modal Foundation Models for Multi-modal Alzheimer's Disease Diagnosis
Wentao Gu, Yuquan Li, XINYANG JIANG, Zilong Wang, Dongsheng Li, Zehui Li, Zijian Dong, Cairong Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use modality-anchored interaction instead of feature concatenation (ablation) | Ours · ADNI NC vs MCI (incomplete) | ACC: 0.894 -> 0.979 | Improved |
| use modality-anchored interaction instead of linear fusion (ablation) | Ours · ADNI NC vs AD (incomplete) | ACC: 0.899 -> 0.945 | Improved |
| use modality-anchored interaction instead of self-attention fusion (ablation) | Ours · ADNI NC vs MCI (incomplete) | ACC: 0.921 -> 0.979 | Improved |
| use BrainMVP instead of Sammed 3D for sMRI (ablation) | Ours · ADNI NC vs MCI | ACC: 0.961 -> 0.979 | Improved |
| use Brain-JePA instead of BrainLM for fMRI (ablation) | Ours · ADNI sMCI vs pMCI | ACC: 0.839 -> 0.846 | Improved |
| use Llama 2 instead of MedGemma for clinical records (ablation) | Ours · ADNI sMCI vs pMCI | ACC: 0.813 -> 0.846 | Improved |
| use NT-Transformer instead of DNA-Bert 2 for genetic data (ablation) | Ours · ADNI NC vs MCI | ACC: 0.960 -> 0.979 | Improved |
| add learnable cross-modal queries (16 queries vs 0) (ablation) | Modality-aware Q-former · ADNI | ACC: 0.881 -> 0.971 | Improved |
| fuse four modalities instead of the best single modality | Ours · ADNI NC vs MCI (complete) | ACC: 0.793 -> 0.871 | Improved |
| use multi-modal Ours instead of M4Survive | M4Survive · ADNI NC vs MCI (complete) | ACC: 0.827 -> 0.871 | Improved |
| transfer to PPMI Parkinson's NC vs PD with Ours instead of M4Survive | M4Survive · PPMI NC vs PD | ACC: 0.954 -> 0.967 | Improved |
| transfer to OASIS-3 with Ours instead of AI-diagnosis | AI-diagnosis · OASIS-3 NC vs AD | AUC: 0.688 -> 0.699 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- M3AE — M3AE: Multimodal Representation Learning for Brain Tumor Segmentation with Missing Modalities (AAAI 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Flex-MoE: Modeling Arbitrary Modality Combination via the Flexible Mixture-of-Experts (NeurIPS 2024 · similarity 0.87)
- EMAD: Evidence-Centric Grounded Multimodal Diagnosis for Alzheimer's Disease (CVPR 2026 · similarity 0.86)
- DECT: Harnessing LLM-assisted Fine-Grained Linguistic Knowledge and Label-Switched and Label-Preserved Data Generation for Diagnosis of Alzheimer’s Disease (AAAI 2025 · similarity 0.85)
- Jointly Modeling Inter- & Intra-Modality Dependencies for Multi-modal Learning (NeurIPS 2024 · similarity 0.85)
- CASL-VAE: Learning Structured Latent Variables from Unpaired Data for Semi-supervised Clustering and Paired Sample Generation (NeurIPS 2026 · similarity 0.84)
- Multi-modal Medical Diagnosis via Large-small Model Collaboration (CVPR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
