Research map / Detection and segmentation / CNN features
OmniCT: Towards a Unified Slice-Volume LVLM for Comprehensive CT Analysis
Tianwei Lin, Zhongwei Qiu, Wenqiao Zhang, Jiang Liu, Yihan Xie, Mingjian Gao, Zhenxuan Fan, Zhaocheng Li, Sijing Li, Zhongle Xie, Peng LU, Yueting Zhuang and 3 more
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 |
|---|---|---|---|
| Add Spatial Consistency Enhancement (SCE) (ablation) | OmniCT · public 2D benchmarks | average accuracy: 78.68 -> 80.14 | Improved |
| Add SCE (ablation) | OmniCT · public 3D benchmarks | average score: 62.17 -> 63.68 | Improved |
| Add Organ-level Semantic Enhancement (OSE) (ablation) | OmniCT · public 3D benchmarks | average score: 62.17 -> 65.37 | Improved |
| Add OSE (ablation) | OmniCT · public 2D benchmarks | average accuracy: 78.68 -> 80.74 | Improved |
| Add OSE on top of SCE (ablation) | OmniCT · MedEval-CT-Bench | average score: 78.06 -> 79.62 | Improved |
| Add SCE on top of OSE (ablation) | OmniCT · MedEval-CT-Bench | average score: 78.62 -> 79.62 | Improved |
| Add SCE and OSE together (ablation) | OmniCT · public 3D benchmarks | average score: 62.17 -> 66.15 | Improved |
| Use 2D SigLIP encoder instead of 3D M3D-CLIP (ablation) | OmniCT · MedEval-CT-Bench native 3D | average accuracy: 80.4 -> 82.9 | Improved |
| Use SigLIP instead of VideoMAEv2 3D encoder (ablation) | OmniCT · MedEval-CT-Bench native 3D | average accuracy: 76.2 -> 82.9 | Improved |
| Use DINOv3 (0.61x tokens) instead of SigLIP (ablation) | OmniCT · MedEval-CT-Bench native 3D | average accuracy: 82.9 -> 81.9 | Mixed |
| Use Phi-4-mini instead of Qwen2.5-3B as LLM backbone (ablation) | OmniCT · MedEval-CT-Bench | average score: 75.9 -> 76.2 | No clear effect |
| Scale OmniCT from 3B to 7B | OmniCT · 2D CT benchmarks | average accuracy: 77.71 -> 81.45 | Improved |
Closest papers by meaning
- Enhancing vision-language models for medical imaging: bridging the 3D gap with innovative slice selection (NeurIPS 2024 · similarity 0.84)
- OmniMedVQA: A New Large-Scale Comprehensive Evaluation Benchmark for Medical LVLM (CVPR 2024 · similarity 0.84)
- SegVol: Universal and Interactive Volumetric Medical Image Segmentation (NeurIPS 2024 · similarity 0.84)
- 3DMedAgent: Unified Perception-to-Understanding for 3D Medical Analysis (ICML 2026 · similarity 0.84)
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.84)
- LiFT: Lifted Inter-slice Feature Trajectories for 3D Image Generation from 2D Generators (NeurIPS 2026 · 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.
