Incomplete Modality Disentangled Representation for Ophthalmic Disease Grading and Diagnosis
Authors: Chengzhi Liu, Zile Huang, Zhe Chen, Feilong Tang, Yu Tian, Zhongxing Xu, Zihong Luo, Yalin Zheng, Yanda Meng
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on four ophthalmology multimodal datasets demonstrate that the proposed IMDR outperforms the state-of-the-art methods significantly. |
| Researcher Affiliation | Academia | 1 Department of Computer Science, University of Exeter, UK 2 Department of Eye and Vision Sciences, University of Liverpool, UK 4 Center For AI And Data Science For Integrated Diagnostics, University of Pennsylvania, USA EMAIL |
| Pseudocode | No | The paper describes the proposed method, IMDR strategy, and JPL module using textual explanations and mathematical equations, but does not include any specific pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code or provide links to a code repository. |
| Open Datasets | Yes | We evaluate the proposed framework using four publicly available multimodal datasets: GAMMA dataset (Wu et al. 2023) and three subsets from Harvard-30k (Luo et al. 2024), including Harvard-30k AMD, Harvard-30k DR, and Harvard-30k Glaucoma, covering Age-related Macular Degeneration (AMD), Diabetic Retinopathy (DR), and Glaucoma. |
| Dataset Splits | Yes | To ensure reliable results, each dataset underwent five-fold cross-validation, and the model is assessed using four key metrics: Accuracy (ACC), F1 score (F1), Area Under the Curve (AUC), and Specificity (Spec), effectively predicting the severity and type of various ophthalmic diseases. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU models, CPU types, or memory used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers) needed to replicate the experiment. |
| Experiment Setup | Yes | LDistill = LCE + LFeat + LLogit, (3), where LCE is a cross-entropy loss for ophthalmic disease diagnosis task, allowing the same input to be associated with multiple classes. ... where ω1 and ω2 are weights that control the contribution of the mutual information and proxy losses, respectively. Figure 6: Ablation study of hyperparameters under the condition of missing OCT modality on Harvard-30k AMD test set. r: the ratio r of number of proxy Np to the number of samples in the training set N. τ: distillation temperature. |