Superior Lung Cancer Detection through CNN and Random Forest: The NeuroForest Approach

Cancer of the lung remains one of the leading causes of cancer death around the world, and early and accurate diagnosis enables improvement of the survival of patients with the ailment. This research brings forth NeuroForest, a model combining a CNN for feature extraction and an RF classifier for stable classification. Our model alleviates the problems associated with independent CNNs and conventional classifiers by taking the strengths of deep learning and ensembling. The performance of the NeuroForest model was evaluated with the LIDC-IDRI lung cancer database. The model attained an accuracy of 96%, markedly surpassing the solo CNN (accuracy: Experimental results demonstrate that compared with the other methods, SVM (accuracy: 92. 3%) and RF (accuracy: 89. 8%) are the most efficient. NeuroForest delivered an accuracy of 94.5 %, an average of five advertising exposures, effectively producing an overall recall of 97. 2% and 95.8%, it is also ideal for discriminating malignant nodules. In the validation mode, the model reached the Area Under the ROC Curve (AUC) 0. of 98 per cent, further establishing high efficiency in tissue differentiation of malignant from benign pulmonary lesions. The results from the hyperparameters’ optimization suggested that the CNN learning rate was 0. RF Decision probes at best performance for 001 and 100 decision trees in the RF. Here, our findings indicate that the combination of CNN & RF enhances the classification performance, resists overfitting, and enhances universally across patient populations.