Kidney stone detection is important in urology to detect kidney stones, as the presence of stones which are not previously diagnosed can have serious health implications. Conventional approaches are mostly including the heavy burden of having to interpret images manually, which is tedious, and does not guarantee uniformity. This study investigates the application of machine learning algorithms for detecting kidney stones using CT scan pictures. In this study, feature extraction methods Local Binary Patterns (LBP) and Histogram of Orientated Gradients (HOG) are used to represent images, as well as multiple hybrid ensemble classifiers, to accurately detect kidney stone. The dataset contains images labeled as normal, cyst, tumor, and kidney stone for training, validation, and test sets. Following ensemble models have been used: combinations of Random Forest with XGBoost, SVM, Naive Bayes, and Decision Trees. The performance was evaluated using accuracy, classification report, and confusion matrix. It demonstrated that the Random Forest + XGB ensemble model reached up to 99.6% accuracy and proved how efficiently we can detect kidney stones using machine learning methods. Such an approach might serve as an alternative to traditional manual image analysis and would lead to higher diagnostic efficacy in medical imaging.