L2GNet: Optimal Local-to-Global Representation of Anatomical Structures for Generalized Medical Image Segmentation
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916604243083264 |
|---|---|
| author | Gorade, Vandan Mittal, Sparsh Dasu, Neethi Singhal, Rekha Santosh, KC Jha, Debesh |
| author_facet | Gorade, Vandan Mittal, Sparsh Dasu, Neethi Singhal, Rekha Santosh, KC Jha, Debesh |
| contents | Continuous Latent Space (CLS) and Discrete Latent Space (DLS) models, like AttnUNet and VQUNet, have excelled in medical image segmentation. In contrast, Synergistic Continuous and Discrete Latent Space (CDLS) models show promise in handling fine and coarse-grained information. However, they struggle with modeling long-range dependencies. CLS or CDLS-based models, such as TransUNet or SynergyNet are adept at capturing long-range dependencies. Since they rely heavily on feature pooling or aggregation using self-attention, they may capture dependencies among redundant regions. This hinders comprehension of anatomical structure content, poses challenges in modeling intra-class and inter-class dependencies, increases false negatives and compromises generalization. Addressing these issues, we propose L2GNet, which learns global dependencies by relating discrete codes obtained from DLS using optimal transport and aligning codes on a trainable reference. L2GNet achieves discriminative on-the-fly representation learning without an additional weight matrix in self-attention models, making it computationally efficient for medical applications. Extensive experiments on multi-organ segmentation and cardiac datasets demonstrate L2GNet's superiority over state-of-the-art methods, including the CDLS method SynergyNet, offering an novel approach to enhance deep learning models' performance in medical image analysis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_05229 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | L2GNet: Optimal Local-to-Global Representation of Anatomical Structures for Generalized Medical Image Segmentation Gorade, Vandan Mittal, Sparsh Dasu, Neethi Singhal, Rekha Santosh, KC Jha, Debesh Computer Vision and Pattern Recognition Continuous Latent Space (CLS) and Discrete Latent Space (DLS) models, like AttnUNet and VQUNet, have excelled in medical image segmentation. In contrast, Synergistic Continuous and Discrete Latent Space (CDLS) models show promise in handling fine and coarse-grained information. However, they struggle with modeling long-range dependencies. CLS or CDLS-based models, such as TransUNet or SynergyNet are adept at capturing long-range dependencies. Since they rely heavily on feature pooling or aggregation using self-attention, they may capture dependencies among redundant regions. This hinders comprehension of anatomical structure content, poses challenges in modeling intra-class and inter-class dependencies, increases false negatives and compromises generalization. Addressing these issues, we propose L2GNet, which learns global dependencies by relating discrete codes obtained from DLS using optimal transport and aligning codes on a trainable reference. L2GNet achieves discriminative on-the-fly representation learning without an additional weight matrix in self-attention models, making it computationally efficient for medical applications. Extensive experiments on multi-organ segmentation and cardiac datasets demonstrate L2GNet's superiority over state-of-the-art methods, including the CDLS method SynergyNet, offering an novel approach to enhance deep learning models' performance in medical image analysis. |
| title | L2GNet: Optimal Local-to-Global Representation of Anatomical Structures for Generalized Medical Image Segmentation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2502.05229 |