Identification of Abnormalities in Mammograms using CUDA Platform
Asad Parvez (604013)
Maharashtra Institute of Technology (M.I.T), Pune. M.E. Electronics (Digital Systems) Email id: asad.apms@gmail.com Mobile No: 9028177296 Abstract Digital mammogram has become the most effective technique for early breast cancer detection. The aim of this study is to develop an algorithm suitable for automated system for assisting the analysis of digital mammograms. Many image processing techniques have been developed over the past two decades to help radiologists in diagnosing breast cancer. At the same time, many studies have proven that an early diagnosis of breast cancer can increase the survival rate. Radiologists have to examine a large number of images. Use of normal and abnormal mammograms from a standard database like Mammographic Image Analysis Society (MIAS) database is proposed to develop and test the algorithm. Computer image processing techniques will be applied to enhance images (Pre-processing). The texture features will be used to classify the region of interest ROIs as either normal or abnormal. The extraction of the textural features of ROIs is done by using gray level co-occurrence matrices (GLCM) which is constructed at four different directions for each ROI. It is expected that the GLCM at 0, 45, 90 and 135 with a block size of 8x8 will give significant texture information to classify tissues as normal and abnormal tissues. The overall accuracy can be improved by most relevant GLCM features, which are selected by feature selection algorithm. In the proposed algorithm a large image is divided into many small size blocks and similar processing steps are applied to all blocks. This leads to large number of computations and greater time. For improving time efficiency, computations can be done using CPU and GPU both. The less Computationally complex part is to be done on CPU and GPU is to be used where parallel programming is possible. It provides efficient and fast response. Thus, this approach is suitable for automated real-time breast cancer diagnosis system.