A team from LNMIIT comprising of undergraduate students Manohar Kuse and Tanuj Sharma led by faculty Mr Sudhir Gupta have won Pattern Recognition in Histology Images (PR in HIMA) contest held during 20th International Conference on Pattern Recognition, Istanbul, Turkey between 21st August and 26th August, 2010.

Related publication will be published in the upcoming Springer Lecture Notes in Computer Science (LNCS). Manohar Kuse also received a travel grant from Microsoft Research, India to present his work in Istanbul.

Abstract of the work: A technique for automating the detection of lymphocytes in histopathological images is presented. The proposed system takes Hematoxylin and Eosin (H&E) stained digital color images to identify lymphocytes. The process involves segmentation of cells from extracellular matrix, feature extraction, classification and overlap resolution. Extra- cellular matrix segmentation is a two step process carried out on the HSV-equivalent of the image, using mean shift based clustering for color approximation followed by thresholding in the HSV space. Texture features extracted from the cells are used to train a SVM classifier that is used to classify lymphocytes and non-lymphocytes.