NeuroForest: Combining CNN and Random Forest for Superior Lung Cancer Detection

Using a hybrid model combining Convolutional Neural Networks (CNNs) and Random Forest classifiers, using histopathology images, this study offers a fresh technique for lung cancer detection. Using a collection of 4,257 labelled lung tissue pictures, the work painstakingly preprocessed to guarantee uniformity and quality. Extensive features from the photos were extracted using the CNN, which had three convolutional layers followed by max-pooling and another set of three convolutional layers. Then, utilising 15 decision trees in a Random Forest, these characteristics were categorised. The accuracy, precision, recall, F1-score, as well as specificity showed the model’s performance is $\mathbf{76.}\mathbf{69\%}$ all-around. In specific, the dimension changes in many forms of lung cancer cells particularly and the results indicated the strong detecting capability of the method. As for the evaluation of the model, it was observed the need to improve, with important results in reduce false positives and false negatives, identified in confusion matrix. High performance for automated lung cancer diagnosis when it comes to this hybrid paradigm: CNN’s feature extraction option and Random Forest’s classification option. Summarising the promotion of deep learning in enhancing the accuracy of cancer diagnosis, the paper concludes with better improvement and future research direction.