DIAGNOSTIC PERFORMANCE OF DIGITAL PATHOLOGY COMPARED WITH CONVENTIONAL MICROSCOPY IN CANCER DIAGNOSIS: A SYSTEMATIC REVIEW AND META-ANALYSIS

Background: Whole-slide imaging (WSI) and digital pathology are increasingly used for primary histopathologicaldiagnosis, second opinions, multidisciplinary review, remote reporting, education, and computational pathology. Beforeroutine use in cancer diagnosis, digital systems must demonstrate performance comparable to conventional light microscopy,especially for clinically significant findings such as malignancy, grade, invasion, margins, lymphovascular invasion, smallmetastatic deposits, and dysplasia. The objective is to systematically compare the diagnostic performance of digitalpathology using WSI with conventional light microscopy for cancer diagnosis and to quantify overall diagnosticconcordance, complete concordance, clinically significant discordance, sensitivity, specificity, and intraobserver agreement.Materials and Methods: A systematic review and meta-analysis was structured according to PRISMA 2020.MEDLINE/PubMed, Embase, Web of Science, Scopus, and the Cochrane Library were considered for comparativevalidation studies evaluating WSI against conventional microscopy in human cancer or cancer-related surgical pathology.Studies were eligible when the same or comparable specimens were interpreted using both modalities and diagnosticconcordance or diagnostic-accuracy outcomes were reported. Random-effects models were used because of expecteddifferences in organ systems, scanners, pathologists, and study design. Risk of bias was assessed using QUADAS-2.Results: Twenty-five comparative validation studies comprising approximately 10,410 histology samples and more than19,000 paired digital-microscopy comparisons were included. Pooled overall clinical concordance was 98.3% (95% CI 97.4-98.9%), while complete concordance was approximately 92.0% (95% CI 87.2-95.1%). Heterogeneity for completeconcordance was substantial (I² approximately 90%). In a large randomized non-inferiority study of 1,992 cases, majordiscordancerates were approximately 4.9% for WSI and 4.6% for conventional microscopy. Across diagnostic-accuracystudies, sensitivity generally ranged from 86% to 100% and specificity from 75% to 100%. Major discordances wereconcentrated in nuclear atypia, dysplasia or grading, intrinsically difficult diagnoses, and small or focal findings.Conclusion: Digital pathology using WSI demonstrates diagnostic performance highly comparable to conventional lightmicroscopy for routine cancer diagnosis. The evidence supports WSI as a safe diagnostic platform when appropriatelyvalidated, although particular caution remains warranted for nuclear grading, dysplasia, mitotic counting, small metastaticdeposits, focal invasion, lymphovascular invasion, and technically suboptimal scans.