Abstract
Alzheimer’s disease (AD) is increasingly recognized as a systemic disorder involving interactions between the brain and heart. Heart FDG-PET may provide complementary metabolic information for early AD diagnosis, yet paired brain and heart PET data are not easily available in routine clinical practice. Here, we propose LEAP-AD, a two-stage framework that encodes brain–heart knowledge from paired PET data into a large language model (LLM) and uses the adapted LLM to enhance brain-only representations for early AD diagnosis. In Stage 1, heart PET is used to guide brain representation learning from paired PET data to capture brain–heart knowledge. Class-level diagnostic prompts and contrastive learning are then used to encode this knowledge into the LoRA adapters of a frozen LLM. In Stage 2, brain features are passed through the LoRA-adapted LLM to incorporate complementary information learned from heart PET, and are then adaptively fused with the original brain features for diagnosis. Extensive experiments show that LEAP-AD outperforms representative baselines for NC vs. MCI classification, supporting the effectiveness of the proposed framework.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MLMI_059.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=y21WCt6llW
BibTex
@InProceedings{ZhoWei_LEAPAD_MICCAISAT2026,
author = { Zhou, Weilin AND Li, Fan AND Zhao, Shilun AND Bai, Shuwei AND Liang, Zhichao AND Yang, Xinnian AND Sun, Kaicong AND Shen, Dinggang},
title = { { LEAP-AD: An LLM-Encoded Brain-Heart Axis Prior Injection Framework for Brain-Only Early Alzheimer’s Disease Diagnosis } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17272},
month = {pending},
page = {pending}
}