Binary Count Ratio for Lung Cancer Classification in Computerized Tomography Scan Images

Accurate lung cancer classification is important for patient treatment. However, existing methods inefficiently classify lung cancer. Therefore, the binary count ratio (BCR) is proposed to enhance the accuracy of lung cancer classification. This method utilizes adaptive thresholding based on column mean to binarize CT images. The features of the BCR method are computed by using the ratio of black and white pixels to identify the lung cancer areas. The proposed BCR method captures the white pixels which are cancer areas in CT scan images. After that, the Euclidean distance method is used to classify a normal or benign or malignant class. For 1,097 images of the IQ-OTH/NCCD dataset, the proposed BCR method achieves 0.9810, 0.9618, 0.9705, and 0.9832 of precision, recall, F1-score, and classification accuracy values which are higher than the conventional methods. The proposed method is able to efficiently extract pixel areas of lung cancer, thus improving the effectiveness of lung cancer classification in CT scan images.