Applications of artificial intelligence for pre-implantation kidney biopsy pathology practice: a systematic review

Background and aims: Kidney biopsy is a crucial diagnostic tool used in the management ofkidney disease. It involves the procedure of removing a small piece of kidney tissue to be examinedfor any abnormalities. The success of pre-implantation kidney biopsy depends on theaccuracy of the pathologist’s diagnosis. Pathological evaluation of kidney biopsy is complex, andat times the interpretation can be subjective, leading to inter-observer variability. The use of artificialintelligence (AI) applications presents an opportunity to improve the diagnostic accuracy ofpre-implantation kidney biopsy pathology practice. The aim of this systematic review is to evaluatethe state of the art in AI applications for pre-implantation kidney biopsy pathology practice.Method: We conducted a systematic literature review of articles published in three electronicdatabases: PubMed, Scopus, and IEEE Xplore. We looked for articles published in English from۲۰۱۰ to ۲۰۲۲ and carried out the search using keywords such as “kidney”, “biopsy”, “transplantation”and “artificial intelligence” and their aliases. We reviewed the studies and extracted therelevant data on AI applications used in pre-implantation kidney biopsy pathology practice.Results: The systematic review included ۳۳ studies that used AI applications to augment or replacethe traditional pathological evaluation of kidney biopsy specimens. Machine learning (ML)algorithms were themost commonly used AI technique. Several studies employed ML techniques to develop predictivemodels that could differentiate between different kidney pathologies, such as glomerulonephritis,tubulointerstitial nephritis, or acute tubular necrosis with high precision. Other studiesused AI-based systems to classify the severity ofkidney damage in biopsy samples. Moreover, AI applications have been shown to reduce inter-observervariability through training pathologists with a standardized system.Conclusion: The use of AI in pre-implantation kidney biopsy pathology practice has shown promisingresults in improving diagnostic accuracy, reducing inter-observer variability, and streamliningthe review process. However, applying AI-based systems to a clinical setting is a challengingtask, and concerns should be raised regarding patient safety, accuracy, and data protection. Futureresearch will require extensive study to validate and refine the utility of these AI applications ina practical clinical setting. Further development, validation, and integration will be crucial inassisting physicians in making a more accurate and precise diagnosis in pre-implantation kidneybiopsy pathology practice.

کلیدواژه ها:

Artificial intelligence ، Digital pathology ، Kidney biopsy ، Pre-implantation biopsy ، Review ، Transplantation