A COMPARATIVE DEEP LEARNING FRAMEWORK FOR AUTOMATED LUNG CANCER DETECTION AND CLASSIFICATION USING CNN AND RESIDUAL NETWORKS

Lung cancer is one of the major causes of death that arise from various kinds of cancers. For this reason, early diagnosis and classification of the disease is essential for treatment purposes.A model for the detection and classification of lung cancer was developed in this study using CT scans of lungs. This involved the use of the CNN and ResNet 50 algorithms to detect and classify features in the model. Various image pre-processing methods, such as resizing, normalizing, filtering, and augmenting the images were performed before the process of classification began using the binary and multiclass methods. The accuracy achieved in the experiment was fairly accurate in the classification of lung cancer. Accuracy of about 98.08% was attained using the two methods. However, higher accuracy was recorded using the ResNet50 algorithm than using the CNN algorithm for multiclass classification. Accuracy attained was 97.10% and 96.54%, respectively.