Automated Identification of Breast Cancer from Low Resolution Microscopic Videos

Health informatics has been qualified as prominent domain in the progress of information technology. Due to this evolution in the health care informatics, it is viable to diagnose several diseases like cancer in a short span of time. In the conventional method, cancer diagnosis requires manual analysis of numerous slides, which is a labour intensive work for pathologists and the procedure is tedious and time taking which causes late diagnosis. The existing works try to capture numerous microscopic high resolution images of the slide for diagnosis which are difficult to take and require high RAM for analysis. Therefore, automatic image handling framework is required that can overcome related limitations in visual investigation. The proposed strategy uses a low resolution video of the slides which is captured by moving a regular camera over the slides. Initially, it decomposes the video to frames then, utilizes image enhancement strategies to improve the quality in terms of contrast and standardize the pixel values in the picture. After enhancement, k-mean segmentation is done to know the precise size, shape and area of the cancerous cell. Finally, k-NN classifier helps to classify and determine whether the cell is cancerous or not.