1. A Hybrid Approach for Pap-Smear Cell Nucleus Extraction
- Author
-
Cosmin Mihai, Hichem Sahli, Maykel Orozco-Monteagudo, Alberto Taboada-Crispi, Carrasco-ochoa, J. Ariel, Audio Visual Signal Processing, Electronics and Informatics, and Multidimensional signal processing and communication
- Subjects
Ground truth ,Computer science ,business.industry ,segmentation ,ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION ,Smear cell ,Pattern recognition ,SVM classification ,Hybrid approach ,cell images ,Accurate segmentation ,Support vector machine ,medicine.anatomical_structure ,medicine ,Segmentation ,Artificial intelligence ,business ,Nucleus ,Classifier (UML) - Abstract
This paper, proposes a two-phases approach for a computerassisted screening system that aims at early diagnosis of cervical cancer in Pap smear images and accurate segmentation of nuclei. The first phase uses spectral, shape as well as the class membership to produce a nested hierarchical partition (hierarchy of segmentations). The second phase, selects the best hierarchical level based on an unsupervised criterion, and refines the obtained segmentation by classifying the individual regions using a Support Vector Machine (SVM) classifier followed by merging adjacent regions belonging to the same class. The effectiveness of the proposed approach for producing a better separation of nucleus regions and cytoplasm areas is demonstrated using both ground truth data, being manually segmented images by pathologist experts, and comparison with state-of-art methods.
- Published
- 2011