Real-Time CT-Based Renal Calculus Segmentation and Analysis
Abstract
Renal calculi is a significant urological disease that needs to be detected and characterized precisely for appropriate clinical diagnosis and treatment planning. This paper proposes a framework for real-time renal calculus detection, segmentation and quantitative analysis based on CT imaging, which is automated. The proposed methodology is integrated into a unified biomedical imaging framework, which includes adaptive preprocessing, entropy-guided contrast enhancement, HU-guided hybrid segmentation, morphological refinement and morphometric characterization. About 120 heterogeneous renal CT volumes were used in experimental evaluation under different imaging conditions. The Dice Similarity Coefficient was 0.918, the Jaccard Index was 0.854, the sensitivity was 94.7%, the specificity was 98.2%, and the overall segmentation accuracy was 96.9%. Clinical descriptors such as stone size, stone perimeter and attenuation density were successfully extracted by quantitative analysis. The entire pipeline took around 257 milliseconds to process one CT slice, with the throughput rate near real time at 3.8-3.9 slices per second. The segmentation fidelity, robustness and computational efficiency were compared and found to be superior to conventional image-processing methods.