International Journal of Pharma and Bio Sciences
ijpbs.net
editorijpbs@rediffmail.com (or) editorofijpbs@yahoo.com (or) prasmol@rediffmail.com
10.22376/ijpbs.2019.10.1.p1-12
Volume 8 Issue 2
2017 (April - June)
Prediction of cancer using histopathology by applying machine learning algorithms
Predicting cancer at its early stages has always been a challenging task. Different ways are already in practice to detect cancer, which of most are very costly methods. Our objective is to propose a cancer diagnosing application that uses the simple digital images of human tissues. An extreme impression of digital images on changing society is proving it as an important component in science and technology. Histopathology plays an important role in automated diagnosis of cancer using tissue images. Our work aims to develop an infrastructure to design an automated cancer diagnosing system by processing and extracting data and information from the biopsy tissue images. The tissue images are segmented using the component graph of watershed transform to extract the pixels of the image. The curvilinear structures and features of the tissue image are extracted and compared with the features of the trained database images. A robust matching algorithm SIFT, with repeated structures is used for image comparison. The images stored to database in an array are iteratively used for feature comparison. The algorithm results can be used to conclude whether the given image is cancerous or not. The early detection of cancer disease increases the probability of curing the disease. Our discussion was helpful in building this application at a very low cost whose usage is easy and effortless. The implementation of this application includes techniques of simple image processing, image transformation and image comparison. The machine learning algorithms are opted to make it easy for the application to train itself with the changes that might be made to application or data-base for future use.
E.NAGARAJAN, CH.UTHPALA AND CH.DEEKSHITHA
Digital image, watershed transform, curvilinear structure, image processing, SIFT algorithm.
279-284