Diffusion Based Ambiguous Image Segmentation
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866913994711760896 |
|---|---|
| author | Christensen, Jakob Lønborg Hannemose, Morten Rieger Dahl, Anders Bjorholm Dahl, Vedrana Andersen |
| author_facet | Christensen, Jakob Lønborg Hannemose, Morten Rieger Dahl, Anders Bjorholm Dahl, Vedrana Andersen |
| contents | Medical image segmentation often involves inherent uncertainty due to variations in expert annotations. Capturing this uncertainty is an important goal and previous works have used various generative image models for the purpose of representing the full distribution of plausible expert ground truths. In this work, we explore the design space of diffusion models for generative segmentation, investigating the impact of noise schedules, prediction types, and loss weightings. Notably, we find that making the noise schedule harder with input scaling significantly improves performance. We conclude that x- and v-prediction outperform epsilon-prediction, likely because the diffusion process is in the discrete segmentation domain. Many loss weightings achieve similar performance as long as they give enough weight to the end of the diffusion process. We base our experiments on the LIDC-IDRI lung lesion dataset and obtain state-of-the-art (SOTA) performance. Additionally, we introduce a randomly cropped variant of the LIDC-IDRI dataset that is better suited for uncertainty in image segmentation. Our model also achieves SOTA in this harder setting. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_05977 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Diffusion Based Ambiguous Image Segmentation Christensen, Jakob Lønborg Hannemose, Morten Rieger Dahl, Anders Bjorholm Dahl, Vedrana Andersen Computer Vision and Pattern Recognition Medical image segmentation often involves inherent uncertainty due to variations in expert annotations. Capturing this uncertainty is an important goal and previous works have used various generative image models for the purpose of representing the full distribution of plausible expert ground truths. In this work, we explore the design space of diffusion models for generative segmentation, investigating the impact of noise schedules, prediction types, and loss weightings. Notably, we find that making the noise schedule harder with input scaling significantly improves performance. We conclude that x- and v-prediction outperform epsilon-prediction, likely because the diffusion process is in the discrete segmentation domain. Many loss weightings achieve similar performance as long as they give enough weight to the end of the diffusion process. We base our experiments on the LIDC-IDRI lung lesion dataset and obtain state-of-the-art (SOTA) performance. Additionally, we introduce a randomly cropped variant of the LIDC-IDRI dataset that is better suited for uncertainty in image segmentation. Our model also achieves SOTA in this harder setting. |
| title | Diffusion Based Ambiguous Image Segmentation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.05977 |