A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation
Despite the tremendous success of deep neural networks in medical image segmentation, they typically require a large amount of costly, expert-level annotated data. Few-shot segmentation approaches address this issue by learning to transfer knowledge from limited quantities of labeled examples. Incor...
Na minha lista:
| Autores principales: | , , , , |
|---|---|
| Formato: | Artigo |
| Publicado em: |
2021
|
| Assuntos: | |
| Acesso em linha: | https://arxiv.org/abs/2103.10178 http://arxiv.org/abs/2103.10178 |
| Tags: |
Sem tags, seja o primeiro a adicionar uma tag!
|
| _version_ | 1879172852166426624 |
|---|---|
| author | Yu, Qinji Dang, Kang Tajbakhsh, Nima Terzopoulos, Demetri Ding, Xiaowei |
| author_browse | Dang, Kang Ding, Xiaowei Tajbakhsh, Nima Terzopoulos, Demetri Yu, Qinji |
| author_facet | Yu, Qinji Dang, Kang Tajbakhsh, Nima Terzopoulos, Demetri Ding, Xiaowei |
| author_sort | Yu, Qinji |
| building | arXiv |
| collection | Acceso abierto |
| description | Despite the tremendous success of deep neural networks in medical image segmentation, they typically require a large amount of costly, expert-level annotated data. Few-shot segmentation approaches address this issue by learning to transfer knowledge from limited quantities of labeled examples. Incorporating appropriate prior knowledge is critical in designing high-performance few-shot segmentation algorithms. Since strong spatial priors exist in many medical imaging modalities, we propose a prototype-based method -- namely, the location-sensitive local prototype network -- that leverages spatial priors to perform few-shot medical image segmentation. Our approach divides the difficult problem of segmenting the entire image with global prototypes into easily solvable subproblems of local region segmentation with local prototypes. For organ segmentation experiments on the VISCERAL CT image dataset, our method outperforms the state-of-the-art approaches by 10% in the mean Dice coefficient. Extensive ablation studies demonstrate the substantial benefits of incorporating spatial information and confirm the effectiveness of our approach. |
| first_indexed | 2026-10-15T23:47:59Z |
| format | Article |
| fullrecord | <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ https://www.openarchives.org/OAI/2.0/oai_dc.xsd"><identifier>arxiv-2103.10178</identifier><datestamp>2021-03-19</datestamp>
<dc:title>A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation</dc:title>
<dc:creator>Yu, Qinji</dc:creator>
<dc:creator>Dang, Kang</dc:creator>
<dc:creator>Tajbakhsh, Nima</dc:creator>
<dc:creator>Terzopoulos, Demetri</dc:creator>
<dc:creator>Ding, Xiaowei</dc:creator>
<dc:subject>Computer Vision and Pattern Recognition</dc:subject>
<dc:description>Despite the tremendous success of deep neural networks in medical image segmentation, they typically require a large amount of costly, expert-level annotated data. Few-shot segmentation approaches address this issue by learning to transfer knowledge from limited quantities of labeled examples. Incorporating appropriate prior knowledge is critical in designing high-performance few-shot segmentation algorithms. Since strong spatial priors exist in many medical imaging modalities, we propose a prototype-based method -- namely, the location-sensitive local prototype network -- that leverages spatial priors to perform few-shot medical image segmentation. Our approach divides the difficult problem of segmenting the entire image with global prototypes into easily solvable subproblems of local region segmentation with local prototypes. For organ segmentation experiments on the VISCERAL CT image dataset, our method outperforms the state-of-the-art approaches by 10% in the mean Dice coefficient. Extensive ablation studies demonstrate the substantial benefits of incorporating spatial information and confirm the effectiveness of our approach. </dc:description>
<dc:description>ISBI2021 accepted</dc:description>
<dc:date>2021-03-18</dc:date>
<dc:type>text</dc:type>
<dc:identifier>https://arxiv.org/abs/2103.10178</dc:identifier>
<dc:identifier>http://arxiv.org/abs/2103.10178</dc:identifier>
</oai_dc:dc> |
| id | arxiv-2103.10178 |
| institution | arXiv |
| last_indexed | 2026-10-15T23:47:59Z |
| publishDate | 2021 |
| publishDateRange | 2021 |
| publishDateSort | 2021 |
| record_format | oai_dc |
| spelling | arxiv-2103.101782021-03-19 A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation Yu, Qinji Dang, Kang Tajbakhsh, Nima Terzopoulos, Demetri Ding, Xiaowei Computer Vision and Pattern Recognition Despite the tremendous success of deep neural networks in medical image segmentation, they typically require a large amount of costly, expert-level annotated data. Few-shot segmentation approaches address this issue by learning to transfer knowledge from limited quantities of labeled examples. Incorporating appropriate prior knowledge is critical in designing high-performance few-shot segmentation algorithms. Since strong spatial priors exist in many medical imaging modalities, we propose a prototype-based method -- namely, the location-sensitive local prototype network -- that leverages spatial priors to perform few-shot medical image segmentation. Our approach divides the difficult problem of segmenting the entire image with global prototypes into easily solvable subproblems of local region segmentation with local prototypes. For organ segmentation experiments on the VISCERAL CT image dataset, our method outperforms the state-of-the-art approaches by 10% in the mean Dice coefficient. Extensive ablation studies demonstrate the substantial benefits of incorporating spatial information and confirm the effectiveness of our approach. ISBI2021 accepted 2021-03-18 text https://arxiv.org/abs/2103.10178 http://arxiv.org/abs/2103.10178 |
| spellingShingle | Computer Vision and Pattern Recognition Yu, Qinji Dang, Kang Tajbakhsh, Nima Terzopoulos, Demetri Ding, Xiaowei A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation |
| title | A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation |
| title_full | A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation |
| title_fullStr | A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation |
| title_full_unstemmed | A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation |
| title_short | A Location-Sensitive Local Prototype Network for Few-Shot Medical Image Segmentation |
| title_sort | location sensitive local prototype network for few shot medical image segmentation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2103.10178 http://arxiv.org/abs/2103.10178 |
| work_keys_str_mv | AT yuqinji alocationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT dangkang alocationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT tajbakhshnima alocationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT terzopoulosdemetri alocationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT dingxiaowei alocationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT yuqinji locationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT dangkang locationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT tajbakhshnima locationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT terzopoulosdemetri locationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation AT dingxiaowei locationsensitivelocalprototypenetworkforfewshotmedicalimagesegmentation |