Automated and Explainable Kidney Abnormality Detection from CT Images Using CNN-LSTM Architecture

Abstract

Kidney diseases such as cysts, stones, and tumors are among the most prevalent health issues worldwide. Early and accurate diagnosis is critical to preventing severe complications, and computed tomography (CT) imaging plays a key role in this process. However, manual interpretation of CT scans is time-consuming and prone to subjectivity. To address this, we propose a hybrid deep learning model combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) units for the automated multiclass classification of kidney diseases using CT images. The CNN component extracts spatial features from the input scans, while the LSTM layers model spatial dependencies and enhance the learning of complex patterns. The model was trained and evaluated on a curated dataset consisting of four kidney conditions: Cyst, Normal, Stone, and Tumor. Extensive experimentation demonstrates that the proposed CNN- LSTM model achieves a classification accuracy of 99.6%, with precision, recall, and F1-score values exceeding 99% across all classes. Additionally, Gradient-weighted Class Activation Mapping (Grad-CAM) is employed for model interpretability, enabling visualization of discriminative regions in the images responsible for predictions. The results indicate the potential of the model to serve as a reliable decision-support tool for radiologists and clinicians. This framework paves the way for enhanced diagnostic accuracy and faster clinical workflows in nephrological imaging.