A novel method for breast mass segmentation: from superpixel to subpixel segmentation
| Autor: | Tianming Zhan, Yi Chen, Fangqing Sheng, Shenghua Gu, Yunjie Chen |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Level set method
Computer science Physics::Medical Physics 02 engineering and technology 030218 nuclear medicine & medical imaging 03 medical and health sciences 0302 clinical medicine 0202 electrical engineering electronic engineering information engineering Segmentation skin and connective tissue diseases Cluster analysis business.industry Pattern recognition Distribution fitting Subpixel rendering Computer Science Applications Data set Hardware and Architecture Computer Science::Computer Vision and Pattern Recognition Pattern recognition (psychology) 020201 artificial intelligence & image processing Computer Vision and Pattern Recognition Artificial intelligence Noise (video) business Software |
| Zdroj: | Machine Vision and Applications. 30:1111-1122 |
| ISSN: | 1432-1769 0932-8092 |
| DOI: | 10.1007/s00138-019-01020-0 |
| Popis: | In this paper, an effective method is proposed for breast mass segmentation using a superpixel generation and curve evolution method. The simple linear iterative clustering method and density-based spatial clustering of applications with noise method are applied to generate superpixels in mammograms at first. Thereafter, a region of interesting (ROI) that contains the breast mass is built on the superpixel generation results. Finally, the image patch and the position of the manual labeled seed are used to build the prior knowledge for the level set method driven by the local Gaussian distribution fitting energy and evolve the curve to capture the edge of breast mass in ROI. Experimental results on mammogram data set demonstrate that the proposed method shows superior performance in contrast to some well-known methods in breast mass segmentation. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
| Nepřihlášeným uživatelům se plný text nezobrazuje | K zobrazení výsledku je třeba se přihlásit. |