Balanced Histology Dataset Analysis for Improved CNN Performance in Lung Cancer Diagnosis

Convolutional neural network (CNN) performance using a confusion matrix and classification report in order to detect lung cancer and colon cancer from medical photos. Especially in differentiating between malignant and non-cancerous tissues, the CNN shows strong classification capacity. With one false positive, the algorithm shows remarkable accuracy in spotting 499 adenocarcinoma patients for colon cancer detection. But it points to a minor misclassification issue by misclassifying six benign events .CNN detects 487 adenocarcinoma cases in lung cancer with little false positives and negatives. With lung squamous cell cancer, it performs less successfully showing 14 false negatives but no false positives, suggesting space for development in this subtype. With great accuracy measures across all assessed classes, the model shows remarkable dependability and precision generally. Precision for colon adenocarcinoma is 0.99, recall 1.00, and an F1-score of 0.99, thereby implying rather few misclassifications. Likewise presenting good performance with a precision and recall of 0.97 is lung adenocarcinoma. The CNN's probable value as a diagnostic tool is highlighted by its capacity to precisely identify benign lung tissues and differentiate between malignant types. Greater datasets and iterative changes seem to help to increase dependability even further. This paper provides important information for clinical use since it shows how successfully CNN detects lung and colon tumors.