Name
Title | ||
|---|---|---|
Automated segmentation of stromal tissue in histology images using a voting Bayesian model. |
Abstract | ||
|---|---|---|
Over the past two decades, digital histology has been clinically approved for the various cancer diagnosis and prognosis tasks including proliferation rate estimation (PRE). Histology images contain two types of regions: epithelial and stromal. PRE is clinically restricted to epithelial tissue because stromal cells do not become cancerous. PRE has very high inter- and intra-pathologist variability and especially among juniors. The major cause of this variability is the stromal area. In this paper, we digitally segment out all stromal areas and present the pathologist with only epithelial areas for PRE. This reduces inter- and intra-pathologist variability. To that end, we propose a Bayesian voting-based model for removal of stromal cells utilizing cells texture and color. Our results on fifty clinical images show that pathologists’ PRE become more accurate and reproducible. Furthermore, PRE of expert pathologists shows very high inter-observer reliability after our fully automated segmentation. We validate our proposed model by testing three aspects and we find: (i) the effect of our segmentation on the clinical decision is the same before and after our segmentation. (ii) the segmentation similarity dice measure is 86.78 % which is a high similarity level. (iii) the time reduction of the pathologist is, on average, over 39 % which also supports the clinical benefit of our proposed work. |
Year | DOI | Venue |
|---|---|---|
2013 | 10.1007/s11760-012-0393-2 | Signal, Image and Video Processing |
Keywords | Field | DocType |
Image segmentation, Digital pathology, Breast cancer prognosis, Proliferation rate estimation, Computer aided diagnosis | Stromal cell,Computer vision,Bayesian inference,Pattern recognition,Segmentation,Computer-aided diagnosis,Image segmentation,Digital pathology,Artificial intelligence,Mathematics,Histology,Bayesian probability | Journal |
Volume | Issue | ISSN |
7 | 6 | 1863-1711 |
Citations | PageRank | References |
3 | 0.46 | 8 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hazem Hiary | 1 | 46 | 7.68 |
| Raja' S. Alomari | 2 | 167 | 16.03 |
| Maha Saadah | 3 | 3 | 0.46 |
| Vipin Chaudhary | 4 | 838 | 83.24 |