Renal cell carcinoma (RCC) is considered the most aggressive and common form of renal cancer. Therefore, early detection is crucial to ensure appropriate and effective treatment planning. In our study, we propose a novel computer-aided diagnostic (CAD) approach which incorporates a deep learning ensemble to differentiate between five renal tumor subtypes, utilising the modality of contrast-enhanced computed tomography (CE-CT). The addressed renal lesions are malignant tumors (chromophobe RCC (chRCC), papillary RCC (pRCC), and clear cell RCC (ccRCC)) and benign tumors (renal oncocytoma (RO) and angiomyolipoma (AML)). Our study includes 280 patients who underwent renal biopsy, 112 patients were diagnosed with benign tumors and 168 patients were diagnosed with malignant tumors. Specifically, we propose a multi-stage classification pipeline to categorize the five types of renal tumors, and at each stage, we use a novel ensemble system that is composed of three components: (i) taking the average probability of the output of a convolutional neural network (CNN) resulting from multiple images per patient; (ii) applying a long-short term memory (LSTM) followed by a feed-forward neural network on the last dense layer of the CNN; (iii) applying a 1D convolutional encoder succeeded by a feed-forward neural network on the last dense layer of the CNN. Our proposed approach outperformed previous studies in discriminating between benign and malignant tumors, and to the best of our knowledge, our study is the first paper to account for the five mentioned renal tumor subtypes. Our classification accuracies were 96.4%, 100%, 91.2%, 93.8% for the discrimination between benign vs. malignant, AML vs. RO, ccRCC vs. non-ccRCC, and pRCC vs. chRCC, respectively. Our proposed CAD and the obtained results establish the potential of a reliable non-invasive diagnostic utility for renal tumors.
Abdeltawab et al. (2026) studied this question.