Automated Classification of Centroblast Cells using Morphological and Texture Features

Sandhya B.S, Chayadevi M. L, Anitha P
2013 International Journal of Computer Applications  
The early detection of lymphoma in patients can greatly increase the probability of recovery. Centroblasts of lymphoma are conventionally identified under microscopes by expert doctor's manually. cytogenetics and immunopheno typing are currently used. A novel method has been proposed to recognize centroblast (CB) cells from non centroblast (non-CB) cells for computer assisted evaluation of lymphocyte centroblasts. The novel aspects is to identify the centroblast cells with prior information.
more » ... ior information. Geometric and texture features are extracted from the input image for identification and classification. Important geometric features like area, boundingbox, convex area, perimeter, major-axis, minor-axis, solidity are extracted. Texture features are extracted using Log-Gabor filter. Combined features of texture and morphology improve the performance of the system significantly. Supervised support vector machine (SVM) classifier is used to classify centroblast (CB) and non-CB cell.
doi:10.5120/12474-8869 fatcat:cprfzu3l5zc5nks2obgveg2tye