Lung Tumor Detection by Using Image Segmentation and Neural Network

Nowadays cancer has become huge threat in human life. There are many types of cancer, Lung cancer is one of the common types causing very high mortality rate. The best way of protection from lung cancer is its early detection and diagnoses. With the fast development of the technology of computed tomography (CT) technology, medical test images become one of the most efficient examination methods to detect clinically the lung disease. The use of digital processing techniques like automatic segmentation for those images can help the Radiologist and the Surgeons to detect and remove lung tumors easily and efficiently. Early detection of cancer is the most promising way to enhance a patient’s chance for survival. The back propagation is a systematic method of training multilayer neural networks in a supervised manner. The back propagation method, also known as the error back propagation algorithm, is based on the error-correction learning rule. The objective of our work is to take CT scan images and perform segmentation of their using OTSU’s thresholding method and various shape parameters include optimal thresholding, area, energy, entropy etc. On the basis of their parameter back propagation network is trained to classify the tumor as per its extent. If the calculated parameter value above the threshold value then extent of cancer is high otherwise low. This work has been done on few CT images and results are analyzed graphically as well as numerically.