Digital image analysis shows high reproducibility and agreement with human interpretation on hep-2 cells

Abstract

<h3>Objectives</h3> Evaluate the precision of the NOVA View1 automated digital image analysis system for anti-nuclear antibody (ANA) testing on HEp-2 cells, and compare the NOVA View interpretation of positive/negative to human interpretation. <h3>Methods</h3> For assessing intra-assay precision, four specimens were assayed in 12 replicates. For inter-assay precision, five serum samples were tested in 25 runs. Endpoint titres of four sera were established in 25 separate runs. To evaluate positive/negative agreement, 204 clinically defined sera were processed on HEp-2 cells and analysed manually and on the NOVA View. <h3>Results</h3> The intra-assay and inter-assay variability were <20%. The same endpoint was found by titration on average 15 out of 25 runs, and was one dilution lower or higher in the rest of the cases. The total agreement for positive and negative interpretation between the automated system and the human was 92%. <h3>Conclusion</h3> The high precision of the NOVA View, and the good agreement of results on 204 clinically defined sera demonstrate the capability of the NOVA View to reliably discriminate between positive and negative specimens. In addition, the archived images can be stored, which provides a basis for objective ANA interpretation, and may facilitate the establishment of ANA testing standardisation.