BibTeX:

@article{IJIRSTV2I2050,
     title={Detection of Malignant Tissues by Segmentation of Histology Images using Histograms of Color and Filter Responses},
     author={Akhila E and Preethymol B},
     journal={International Journal for Innovative Research in Science & Technology},
     volume={2},
     number={2},
     pages={162--168},
     year={},
     url={http://www.ijirst.org/articles/IJIRSTV2I2050.pdf},
     publisher={IJIRST (International Journal for Innovative Research in Science & Technology)},
}

In medical science, digital image processing has relevance as several techniques such as MRI, CT-scan, laparoscopic and endoscopic surgeries and cancer diagnosis tools are being used currently. We present an approach for segmentation of images which are having vague boundaries between regions. The traditional segmentation techniques exhibit weak performance on edgeless images such as histology images. In contrast, the recently proposed segmentation framework exposed better results on histology dataset. It modeled images as the occlusions of realizations of textures. This concept directed us to suggest a variety of this framework. Our method achieves segmentation through convolution, factorization and deconvolution using the histograms of filter responses instead of color distributions. Based on the theoretical study, the system reveals the occlusion of textures in histology images and attains the segmentation in a faster manner. Also, we introduce a method to diagnose cancer by looking for abnormal mitosis and mutations. The histology image is classified as normal and cancer affected, for the diagnosis.

Filter Response, Convolution, Deconvolution, Occlusion of Textures, Mitosis, Malignant Tissues