Lesion Segmentation in Whole-Body Multi-Tracer PET-CT Images; a Contribution to AutoPET 2024 Challenge
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866929509784092672 |
|---|---|
| author | Astaraki, Mehdi Bendazzoli, Simone |
| author_facet | Astaraki, Mehdi Bendazzoli, Simone |
| contents | The automatic segmentation of pathological regions within whole-body PET-CT volumes has the potential to streamline various clinical applications such as diagno-sis, prognosis, and treatment planning. This study aims to address this challenge by contributing to the AutoPET MICCAI 2024 challenge through a proposed workflow that incorporates image preprocessing, tracer classification, and lesion segmentation steps. The implementation of this pipeline led to a significant enhancement in the segmentation accuracy of the models. This improvement is evidenced by an average overall Dice score of 0.548 across 1611 training subjects, 0.631 and 0.559 for classi-fied FDG and PSMA subjects of the training set, and 0.792 on the preliminary testing phase dataset. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_14475 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Lesion Segmentation in Whole-Body Multi-Tracer PET-CT Images; a Contribution to AutoPET 2024 Challenge Astaraki, Mehdi Bendazzoli, Simone Image and Video Processing Computer Vision and Pattern Recognition The automatic segmentation of pathological regions within whole-body PET-CT volumes has the potential to streamline various clinical applications such as diagno-sis, prognosis, and treatment planning. This study aims to address this challenge by contributing to the AutoPET MICCAI 2024 challenge through a proposed workflow that incorporates image preprocessing, tracer classification, and lesion segmentation steps. The implementation of this pipeline led to a significant enhancement in the segmentation accuracy of the models. This improvement is evidenced by an average overall Dice score of 0.548 across 1611 training subjects, 0.631 and 0.559 for classi-fied FDG and PSMA subjects of the training set, and 0.792 on the preliminary testing phase dataset. |
| title | Lesion Segmentation in Whole-Body Multi-Tracer PET-CT Images; a Contribution to AutoPET 2024 Challenge |
| topic | Image and Video Processing Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2409.14475 |