Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2503.16780 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916658882281472 |
|---|---|
| author | Cho, Uihyun Kim, Namhun |
| author_facet | Cho, Uihyun Kim, Namhun |
| contents | Recent advances in deep-learning based denoising methods have improved Low-Dose CT image quality. However, due to distinct HU distributions and diverse anatomical characteristics, a single model often struggles to generalize across multiple anatomies. To address this limitation, we introduce \textbf{Agent-Integrated Denoising Experts (A-IDE)} framework, which integrates three anatomical region-specialized RED-CNN models under the management of decision-making LLM agent. The agent analyzes semantic cues from BiomedCLIP to dynamically route incoming LDCT scans to the most appropriate expert model. We highlight three major advantages of our approach. A-IDE excels in heterogeneous, data-scarce environments. The framework automatically prevents overfitting by distributing tasks among multiple experts. Finally, our LLM-driven agentic pipeline eliminates the need for manual interventions. Experimental evaluations on the Mayo-2016 dataset confirm that A-IDE achieves superior performance in RMSE, PSNR, and SSIM compared to a single unified denoiser. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_16780 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A-IDE : Agent-Integrated Denoising Experts Cho, Uihyun Kim, Namhun Computer Vision and Pattern Recognition Recent advances in deep-learning based denoising methods have improved Low-Dose CT image quality. However, due to distinct HU distributions and diverse anatomical characteristics, a single model often struggles to generalize across multiple anatomies. To address this limitation, we introduce \textbf{Agent-Integrated Denoising Experts (A-IDE)} framework, which integrates three anatomical region-specialized RED-CNN models under the management of decision-making LLM agent. The agent analyzes semantic cues from BiomedCLIP to dynamically route incoming LDCT scans to the most appropriate expert model. We highlight three major advantages of our approach. A-IDE excels in heterogeneous, data-scarce environments. The framework automatically prevents overfitting by distributing tasks among multiple experts. Finally, our LLM-driven agentic pipeline eliminates the need for manual interventions. Experimental evaluations on the Mayo-2016 dataset confirm that A-IDE achieves superior performance in RMSE, PSNR, and SSIM compared to a single unified denoiser. |
| title | A-IDE : Agent-Integrated Denoising Experts |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.16780 |