Kidney Stone Detection from CT Images Using ALEXNET and Hybrid ALEXNET-RF Models

Nowadays, kidney stone disease is one of the most common health issue which needs more attention for early diagnosis. Several imaging modalities are used for the detection of kidney stone. The gold standard CT scans are valuable for kidney stone detection. For kidney stone detection, machine and deep learning-based algorithms are widely used. In order to enhance the performance of earlier techniques, two techniques are developed. Initially, an AlexNet-based model is developed in this work. By using the enhanced recognition capability of Random Forest (RF), we developed a hybrid AlexNet-RF model. Both the models are tested against Kidney Stone Detection dataset. The performance of the proposed model proved that in terms of accuracy and loss the hybrid AlexNet-RF model secured reliable higher detection rate of approximately 97.1% to 97.5%. This showed that embedding RF in the Softmax layer of AlexNet significantly improves the prediction rate of kidney stone.