Explainable AI in kidney stone detection and segmentation: a mini review.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Renal Disorders

Background:

• Kidney stones are common renal disorders requiring early diagnosis and treatment.
• Advances in AI, particularly deep learning (DL) and explainable AI (XAI), are enhancing diagnostic capabilities.
• Automatic segmentation and detection of kidney stones from medical imaging improve efficiency and accuracy.

Purpose of the Study:

• To review recent studies (2020-2025) on machine learning, deep learning, and hybrid models for kidney stone segmentation.
• To identify commonly used XAI techniques in kidney stone detection.
• To assess the potential of integrated DL and XAI in clinical decision-making.

Main Methods:

• Review of eighteen representative studies published between 2020 and 2025.
• Analysis of machine learning, deep learning, and hybrid models for kidney stone segmentation.
• Identification and categorization of XAI techniques (SHAP, LIME, Grad-CAM, LRP, EigenCAM).

Main Results:

• DL and XAI models demonstrate significant improvements in kidney stone segmentation and detection accuracy.
• XAI techniques enhance clinician trust and support clinical decision-making, especially in resource-limited settings.
• Commonly utilized XAI methods include SHAP, LIME, Grad-CAM, Layer-wise Relevance Propagation, and EigenCAM.

Conclusions:

• Integrating DL with XAI provides a transparent, reliable, and clinically acceptable method for kidney stone detection and segmentation.
• Despite advancements, limitations like dataset diversity, multimodal integration, and real-world validation need addressing.
• This approach holds promise for early diagnosis and improved patient outcomes in managing renal disorders.