Stage-aware Multimodal Brain Graph Contrastive Learning for Neurodegenerative Disorder Diagnosis
Organizations: School of Information and Electronic Engineering, Zhejiang Gongshang University, Hangzhou 310018, China · School of Computer and Network Security, Chengdu University of Technology, Chengdu, 610059, China · School of Computing Technologies, RMIT University, Melbourne, VIC 3000, Australia
Abstract
Neurodegenerative disorders are characterized by their progressive nature, with distinct stages that require precise identification and tracking of progression trajectories for accurate diagnosis. However, existing diagnostic methods often fail to fully capture the unique brain patterns associated with different stages, limiting their ability to effectively target specific stages of the disorders. To address this issue, we propose a Stage-aware Multimodal Brain Graph Contrastive Learning (smBGC) method. Specifically, smBGC introduces a stage-aware feature enhancement strategy to capture stage-specific features across all progression stages, enabling the generation of stage-aware multimodal brain graph representations. In addition, smBGC incorporates a stage-adaptive brain graph augmentation approach within a graph contrastive learning framework, facilitating multimodal fusion and improving the model's capability to distinguish between different disorder stages. Experiments on two real-world datasets demonstrate the superior performance of smBGC in various neurodegenerative disorder diagnosis tasks compared with state-of-the-art methods.
Figures & tables
| Notations | Description |
| Dataset of brain graphs | |
| , , | SBG, FBG and a brain graph from any modality of the i-th subject, respectively |
| The stage label of i-th subject | |
| Node set of i-th subject | |
| Edge set of , represents any modality | |
| The number of subjects in |
| Hyperparameter | ADNI-3 (three stages) | ADNI-3 (four stages) | PPMI |
| Functional magnitude | 0.6 | 0.6 | 0.7 |
| Structural magnitude | 0.6 | 0.5 | 0.3 |
| Dropping features | 20 | 20 | 30 |
| GNN layers | 1 | 1 | 1 |
| Dropout | 0.5 | 0.5 | 0.5 |
| Readout method | sum | sum | sum |
| Method | NC vs. MCI vs. AD | NC vs. EMCI vs. LMCI vs. AD | |||||||||
| ACC | F1 | AUC | SEN | SPE | ACC | F1 | AUC | SEN | SPE | ||
| GCN | 54.30 1.67 | 41.97 2.29 | 67.51 2.72 | 42.80 1.79 | 74.06 1.19 | 43.75 3.64 | 35.01 3.60 | 65.90 3.58 | 35.07 3.60 | 80.83 1.29 | |
| GAT | 50.78 2.44 | 41.65 3.48 | 65.22 2.99 | 41.79 2.86 | 72.07 0.88 | 46.88 7.23 | 37.07 5.31 | 69.95 5.03 | 39.17 4.77 | 81.38 1.87 | |
| GIN | 50.77 2.57 | 42.38 2.48 | 63.06 2.45 | 43.19 2.75 | 72.89 1.42 | 50.00 11.63 | 47.28 11.58 | 63.51 7.34 | 47.22 10.95 | 82.56 3.93 | |
| GraphSAGE | 56.95 4.13 | 47.74 5.15 | 69.47 3.56 | 47.84 5.02 | 75.55 2.87 | 43.75 6.12 | 39.29 5.14 | 61.55 4.87 | 40.84 4.46 | 81.26 2.47 | |
| BrainGB | 53.33 4.46 | 38.23 4.79 | 62.10 3.09 | 40.84 4.12 | 72.35 2.81 | 40.63 5.92 | 32.14 6.02 | 55.68 4.01 | 34.67 4.20 | 76.02 1.79 | |
| Method | NC vs. Prodromal vs. PD | |||||
| ACC | F1 | AUC | SEN | SPE | ||
| GCN | 51.92 3.57 | 49.55 3.19 | 66.26 4.91 | 50.34 2.23 | 74.66 0.66 | |
| GAT | 50.58 5.66 | 46.22 1.68 | 60.32 3.08 | 48.00 1.62 | 73.20 5.56 | |
| GIN | 50.52 5.10 | 49.24 4.23 | 68.13 3.94 | 49.83 4.42 | 74.25 2.43 | |
| GraphSAGE | 55.04 2.67 | 52.46 5.20 | 66.91 5.77 | 53.22 4.35 | 76.19 1.29 | |
| BrainGB | 48.10 5.39 | 42.80 4.26 | 60.82 2.62 | 45.51 2.99 | 72.29 1.75 | |
| Metric | Fold 1 | Fold 2 | Fold 3 | Fold 4 | Fold 5 | Mean Std. | 95% CI |
| ACC | 80.12 | 82.68 | 82.90 | 82.11 | 80.74 | [80.19, 83.23] | |
| F1 | 78.43 | 79.12 | 79.78 | 79.25 | 78.52 | [78.33, 79.71] | |
| AUC | 96.55 | 96.28 | 95.73 | 96.51 | 95.63 | [95.60, 96.68] | |
| SEN | 62.42 | 67.21 | 63.85 | 67.54 | 63.83 | [62.15, 67.79] | |
| SPE | 96.72 | 95.81 | 96.65 | 95.54 | 96.48 | [95.58, 96.90] |
| Metric | smBGC (%) | MHGF (%) | Improvement | 95% CI | Adjusted | |
| ACC | 81.71 | 57.56 | 24.15 | [20.71, 27.59] | 19.47 | |
| F1 | 79.02 | 47.41 | 31.61 | [27.84, 35.37] | 23.29 | |
| AUC | 96.14 | 69.98 | 26.16 | [19.65, 32.67] | 11.16 | |
| SEN | 64.97 | 50.78 | 14.19 | [11.50, 16.87] | 14.65 | |
| SPE | 96.24 | 85.23 | 11.01 | [9.75, 12.28] | 24.22 |
| Methods | NC vs. Prodromal vs. PD | |||||
| ACC | F1 | AUC | SEN | SPE | ||
| w/o CFD & ASFE | 41.73 5.26 | 36.97 5.87 | 56.50 4.08 | 39.53 3.85 | 69.09 5.43 | |
| w/o ASFE | 57.44 5.29 | 50.15 1.59 | 70.94 6.76 | 52.24 5.63 | 80.88 2.94 | |
| w/o CFD | 83.24 2.29 | 78.28 3.61 | 97.47 1.19 | 82.96 2.64 | 91.73 5.91 | |
| smBGC (ours) | 91.15 3.19 | 91.24 3.13 | 96.27 0.91 | 91.77 4.33 | 95.61 3.52 | |
Medical Diagnosis
Medical diagnosis uses clinical data, tests, and images to identify diseases, assess risk, and guide care.
See all 274 papers