Advancements in Cholelithiasis Diagnosis: A Systematic Review of Machine Learning Applications in Imaging Analysis

Area of Science:

• Medical Imaging and Artificial Intelligence
• Gastroenterology and Hepatology

Background:

• Gallstone disease affects many globally, with risk factors including obesity, diabetes, and genetics.
• Complications range from cholecystitis to gallbladder cancer, necessitating accurate diagnosis.
• Current imaging methods like ultrasound (US), computed tomography (CT), and magnetic resonance cholangiopancreatography (MRCP) have varying sensitivities and specificities.

Purpose of the Study:

• To systematically review and assess the diagnostic accuracy of machine learning (ML) technologies for detecting gallstones.
• To evaluate the performance of AI/ML algorithms in analyzing medical images for gallstone identification.

Main Methods:

• A systematic review adhering to PRISMA guidelines was conducted.
• Searches were performed in PubMed, Cochrane Library, Scopus, and Embase up to April 2024.
• Included studies used AI/ML algorithms with imaging modalities (US, CT, MRCP) for gallstone detection; excluded were non-imaging studies and reviews.

Main Results:

• Eight studies met the inclusion criteria, primarily using retrospective designs with ultrasonography, CT, or MRCP.
• All studies employed neural networks, predominantly convolutional neural networks (CNNs), for gallstone detection.
• Models were trained and validated on datasets ranging from 60 to 2,386 patients and 60 to 17,560 images.

Conclusions:

• Reviewed ML models demonstrate promising performance in gallstone diagnosis, comparable to expert radiologists.
• Significant work is required to ensure the reliability and generalizability of these AI diagnostic tools in diverse clinical settings.
• Rigorous validation and consideration of limitations are crucial for successful clinical integration of ML in gallstone detection.