2020 Volume 69 Issue 3 Pages 403-407
We installed a computer-assisted microscopy system, EUROPattern Cosmic (EPA), in 2016 to solve the problems of complicated steps and visual judgement of the results of conventional indirect immunofluorescence assay for anti-nuclear antibody detection. However, there was a deviation of the judgement matching rate on the predicted antibody titer and chromosome type between visual judgements; thus, full automation was not achieved. The matching rates of the predicted antibody titer using a photo image obtained by EPA and visual judgement were κ factor = 0.762 in 2016 and κ factor = 0.891 in 2019. The κ factor difference improved significantly (p < 0.01). The matching rates of chromosome-type judgment of the speckled pattern, nucleolar pattern, and centromere pattern in terms of the κ factor difference improved significantly (p < 0.05, p < 0.01, and p < 0.01, respectively). The weak positive predicted judgement of the homogeneous pattern and speckled pattern should be improved to “negative”. The development of an updated version of the Auto Predicted Judgement system with improved judgement accuracy is desired.
