A Novel Technique to Predict and Detect Lung Cancer in the Computerized Tomography Images

The number of hospital-generated digital images is increasing rapidly, as effective medical images can play an important role in assisting diagnosis and treatment. These images can also be useful in the educational field, especially for medical students. Many new techniques have been developed in the field of image retrieval using automatic classification technology over the last few years. For this purpose, many imaging methods have been manufactured. Numerous classification algorithms have been developed for medical images, both grayscale and color images. In this paper we present a new algorithm capable to separate anomalies in the lung region ideally, which is scanned by CT-scan. The algorithm works on grayscale images and consists of four steps, fuzzy clustering using C-mean, median filter, edge detection and selection, features extraction and classification using a neural network. The algorithm was trained and tested using 400 images from the NH / NCI lung Consortium. The results were evaluated using a statistical model where the accuracy, sensitivity and specificity were very high compared to the existing algorithms, considering the simplicity of the algorithm developed by us.