Detection And Classification Of Bone Tumor Using Machine Learning Classifiers

Bone tumors can be associated with severe health risks. Early detection is one of the most critical aspects of treatment. This project focuses on developing a machine learning approach that would help in predicting bone tumors at a very early stage. This will consequently lead to improved medicine, saving lives. A system employing Keras and OpenCV for bone tumor image detection has been designed. The system employed the use of certain preprocessing techniques, such as the OTSU method, the image thresholding technique, wherein every photograph's clarity is enhanced, focusing only on regions of interest. The model was trained to label tumor regions with the use of CNN to enable full learning and proper classification of tumors. The model was successful and achieved an accuracy of 85% in the prediction of bone tumors. The system not only provides accurate predictions but also points out the region of the tumor after processing the image, which helps the treating doctor interpret the results visually. Such a tool may be used as an ancillary aid in clinical diagnosis to produce more accurate and quicker diagnoses.