A Scoring System Using Visual and Artificial Intelligence Analysis Quantifies HEp-2 Cell Slide Quality for Optimal Antinuclear Antibody Testing 2253905

Abstract Introduction High-quality HEp-2 cell slides are essential for antinuclear autoantibodies (ANA) testing. However, there is currently no established method to quantify the substrate quality. This leads to subjective evaluation and affects the selection of high-quality HEp-2 cell slides. Our goal was to establish a method to thoroughly quantify HEp-2 cell slide quality and select the best HEp-2 cell slides for accurate and reliable ANA testing. Methods This study used both visual and Artificial Intelligence (AI) assisted image analysis methods to quantify the quality of HEp-2 slides based on a scoring system. The score system included cell distribution scores (density, aggregation, and separation); cell morphology scores (interphase/mitotic cell appearance, and nuclear/cytoplasmic ratio); mitotic cell scores; fluorescence scores; and pattern scores. Higher scores represent higher quality. dIFine M1® manual microscope, dIFine® automated microscope, Sebia ANA HEp-2 slides and another brand of HEp-2 slides from 2 different batches respectively, were used in the validation. Results Results from quantifying HEp-2 slide quality across different ANA patterns illustrate high agreement and accuracy between visual and AI-assisted analysis. Comparison of the total scores and the scores for each category can clearly differentiate the quality of HEp-2 cell slides and identify the substrate defects. Sebia ANA HEp-2 slides show significantly higher scores in all categories and present optimal quality for ANA testing. Conclusion This quantification scoring system demonstrates for the first time that a quantitative approach can be used to assess the quality of HEp-2 cells on ANA test slides. Thus, it is now possible to reliably select the best quality HEp-2 cell slides for ANA testing. Funding Source n/a Topic Categories Technological Innovations in Immunology (TECH)