Clip2clinic Mimic Pa

--- license: other license_name: physionet-credentialed-health-data license_link: https://physionet.org/content/mimic-cxr-jpg/view-dua/2.1.0/ language: - en library_name: pytorch pipeline_tag: zero-shot-image-classification tags: - medical-imaging - chest-xray - clip - vision-language - mimic-cxr - radiology - reproduction - dynamic-soft-labels - negation-aware base_model: - emilyalsentzer/Bio_ClinicalBERT - timm/swin_tiny_patch4_window7_224 datasets: - MIMIC-CXR-JPG metrics: - accuracy - auc - f1 model-index: - name: clip2clinic-mimic-pa-n_g_dtcg results: - task: type: zero-shot-image-text-retrieval name: CXR-Align Task A (Qwen-fair) dataset: type: mimic-cxr-jpg name: MIMIC-CXR-JPG PA+AP test metrics: - type: accuracy value: 0.955 name: Task A acc - task: type: zero-shot-image-text-retrieval name: CXR-Align Task B dataset: type: mimic-cxr-jpg name: MIMIC-CXR-JPG PA+AP test metrics: - type: accuracy value: 0.804 name: Task B acc --- # CLIP-to-Clinic, MIMIC-CXR Reproduction Weights for an academic re-implementation of **"Bringing CLIP to the Clinic: Dynamic Soft Labels and Negation-Aware Learning for Medical Analysis"** by Hanbin Ko & Chang-Min Park (CVPR 2025, [arXiv:2505.22079](https://arxiv.org/abs/2505.22079)). Produced for **METU CENG501 (Spring 2026) Deep Learning** ([course page](https://user.ceng.metu.edu.tr/~skalkan/DL/)), the term project where students reproduce a recent paper that ships no public code. Companion code: <https://github.com/alpozaydin/CENG501-Spring2026/tree/main/HorozOzaydin>. ## Headline result The full method (`n_g_dtcg`) recovers the paper's central CXR-Align numbers within **1 percentage point**: | Metric | Paper | Ours | |---|---:|---:| | CXR-Align Task A (Qwen-fair cross-generator) | 96.5 | **95.5** | | CXR-Align Task B | 80.1 | **80.4** | ## What's in this repo Eight ablation checkpoints, one per row of the paper's component decomposition. All trained for 10 epochs at batch size 64 on the same MIMIC-CXR PA+AP frontal split. | File | Soft labels | Graph stream | Hard negatives | What it isolates | |---|---|:-:|:-:|---| | `baseline_final.safetensors` | none (InfoNCE) | ✗ | ✗ | Paper's CLIP baseline | | `dt_final.safetensors` | textual | ✗ | ✗ | Textual soft labels alone | | `dt_c_final.safetensors` | textual + clinical | ✗ | ✗ | + CheXbert label similarity | | `n_final.safetenso...