Weakly Supervised Object Detection in Chest X-Rays with Differentiable ROI Proposal Networks and Soft ROI Pooling
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909498446184448 |
|---|---|
| author | Müller, Philip Meissen, Felix Kaissis, Georgios Rueckert, Daniel |
| author_facet | Müller, Philip Meissen, Felix Kaissis, Georgios Rueckert, Daniel |
| contents | Weakly supervised object detection (WSup-OD) increases the usefulness and interpretability of image classification algorithms without requiring additional supervision. The successes of multiple instance learning in this task for natural images, however, do not translate well to medical images due to the very different characteristics of their objects (i.e. pathologies). In this work, we propose Weakly Supervised ROI Proposal Networks (WSRPN), a new method for generating bounding box proposals on the fly using a specialized region of interest-attention (ROI-attention) module. WSRPN integrates well with classic backbone-head classification algorithms and is end-to-end trainable with only image-label supervision. We experimentally demonstrate that our new method outperforms existing methods in the challenging task of disease localization in chest X-ray images. Code: https://github.com/philip-mueller/wsrpn |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_11985 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Weakly Supervised Object Detection in Chest X-Rays with Differentiable ROI Proposal Networks and Soft ROI Pooling Müller, Philip Meissen, Felix Kaissis, Georgios Rueckert, Daniel Computer Vision and Pattern Recognition Machine Learning Weakly supervised object detection (WSup-OD) increases the usefulness and interpretability of image classification algorithms without requiring additional supervision. The successes of multiple instance learning in this task for natural images, however, do not translate well to medical images due to the very different characteristics of their objects (i.e. pathologies). In this work, we propose Weakly Supervised ROI Proposal Networks (WSRPN), a new method for generating bounding box proposals on the fly using a specialized region of interest-attention (ROI-attention) module. WSRPN integrates well with classic backbone-head classification algorithms and is end-to-end trainable with only image-label supervision. We experimentally demonstrate that our new method outperforms existing methods in the challenging task of disease localization in chest X-ray images. Code: https://github.com/philip-mueller/wsrpn |
| title | Weakly Supervised Object Detection in Chest X-Rays with Differentiable ROI Proposal Networks and Soft ROI Pooling |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2402.11985 |