Bone Cancer Detection from MRI Scan Imagery Using Mean Pixel Intensity

Abstract

Cancer is a dangerous disease, which is caused because of unregulated cell growth. After many researches, almost 100 different types of cancer has been detected in human body. Out of these, one of the most widely spread is bone cancer, which leads to death. The detection of bone cancer is very critical and which has no anticipation. Presently, most of the study is done by using data mining methods and the image processing techniques for medical image analysis process. The data and the knowledge collecting from large databases and related websites have been predictable by many scientific researchers. Association rule mining, supports vector machines, fuzzy theory and probabilistic neural networks and learning vector quantization are the mostly used methods for detection and classification of bone cancer. This paper used k means clustering algorithm for bone image segmentation. The segmented image is further processed for bone cancer detection by evaluating the mean intensity the identified area. Threshold values are proposed for the classification of medical images for the presence or absence of bone cancer. This method uses jpeg images, but also applicable for original format of DICOM (digital imaging communication of medicine) medical images if any modifications are done. The results using this method gives 95% accuracy with less computational time.