Histopathological Image Analysis Using Improved Gorilla Troops Optimizer for Lung Cancer Detection

Among many different modalities, histopathological imaging is one of the important and popular diagnostic technique for lung cancer. Computer aided diagnostic tools that are based on Artificial Intelligence will be very useful in analyzing those images and helps the clinician to take the decision about the disease occurrence and treatment options. In this research work, deep features are extracted from the histopathological images using three popular deep learning techniques namely VGG, DenseNet, and AlexNet. Then Improved Gorilla Troops Optimizer is suggested as feature transformation technique on the deep features extracted from the histopathological images for lung cancer identification. Random Forest algorithm is used for classifying the transformed features. Significance of proposed optimization as feature transform is proved through experimentation on the publicly available LC25000 dataset that contains 3 classes where one class belongs to benign category and another two classes belong to malignant lung lesions namely adenocarcinoma and squamous cell carcinoma. The proposed technique attains $98.2 \%$ accuracy in classifying the histopathological images into above mentioned three classes.