Automated characterization and counting of Ki-67 protein for breast cancer prognosis

摘要: Hybrid clustering using fuzzy C-means and k-means to segment Ki-67 nuclei.Characterization of protein for assessing proliferation index.Validation methodology from the ground truth information.Qualitative assessment proposed F-measure. expression plays an important role in predicting proliferative status tumour cells deciding future course therapy breast cancer. Immunohistochemical (IHC) determination score or labelling index, by estimating fraction Ki67 positively stained cells, is most widely practiced method assess (Dowsett etal. 2011). Accurate manual counting these (specifically nuclei) due complex dense distribution therefore, becomes critical presents a major challenge pathologists. In this paper, we suggest hybrid algorithm quantify index cancer based on automated nuclei. The initially pre-processes IHC images slides RGB are converted grey, L*a*b*, HSI, YCbCr, YIQ XYZ colour space. All then characterized two stage segmentation process. Fuzzy quantifies all as one cluster. blue channel first output given input algorithm, which provides separate cluster positive negative cells. count nuclei used calculate F-measure each A comparative study our work with expert opinion studied evaluate error rate. detection results spaces compared validation calculated. L*a*b* space (0.8847) best statistical result Further, carried out manually automatically average rate 6.84% significant. L*a*b*colour technique. Computerized evaluation can aid pathologist severity. methodology, further, has potential advantage saving time assisting decision making over present procedure could evolve assistive pathological support system.