MobileNet Architecture for Deep Learning-Based Automated Detection and Classification of Medullary Sponge Kidney in Computed Tomography Imaging

The paper introduces a deep learning framework that identifies and classifies Medullary Sponge Kidney (MSK) automatically in computer tomography (CT) imagery with the MobileNet framework. The curated dataset was used to train the proposed model and test it on, using data augmentation and transfer learning to improve generalization. Confusion matrix, ROC curve, and classification statistics showed an ideal classification, with 100 % accuracy, precision, recall, and F1-score. ROC curve area under the curve (AUC) was 1.0, which established the discriminative power of the model. The training and validation accuracy curves showed quick convergence with very little loss, which means there was excellent learning without overfitting. These outcomes show that MobileNet can be a computationally effective yet very precise MSK detection method in CT scanning, which can be used in faster, automatic clinical decision-making. Such a method has the potential to be implemented in radiology workflows to enhance the effectiveness of diagnosis and help early identify disorders of the kidneys.