BONE TUMOR DETECTION USING MACHINE LEARNING

Author Name: 1Deepak Mane, 2 Kiran Shibe

Volume: 01 &  Issue:

Country: India

DOI NO.: 08.2020-25662434 DOI Link: http://www.doi-ds.org/doilink/01.2021-86625985/UIJIR

Affiliation:

¹Senior Data Scientist, Tata Research Development & Design Center, Pune, Maharashtra, India

²Analytics and Insights, Tata Research Development & Design Center, Pune, Maharashtra, India

ABSTRACT

Cancer is a deadly illness, caused by unchecked growth of the cells. Nearly 100 different forms of cancer have been found in the human body, following much study. One of the most commonly spread out of these is bone cancer, which leads to death. Bone cancer diagnosis is very important, and has little expectation. Part of the work is currently conducted using data mining tools and the image processing technologies for the process of medical image analysis. Many academic scholars have become reliable with the data and information obtained from broad datasets and associated websites. The most widely used approaches for identifying and classifying bone cancer are association rule mining, support vector machines, fuzzy theory and probabilistic neural networks, and learning vector quantizations. This paper used superpixel algorithm for segmentation of the bone image. For identification of bone cancer the segmented image is further processed by determining the mean strength of the detected region. Threshold values for classifying medical images for presence or absence of bone cancer are proposed. With less computational time, the results using this method give 97.5 per cent accuracy.

Key words: Bone Cancer, SVM, superpixel, segmentation.