Objective To evaluate the agreement of four common HEp-2 indirect immunofluorescence assay(IFA)kits in the patients with antinuclear antibody(ANA)-associated rheumatic immune diseases(AARD)and the patients with non-autoimmune diseases(NAD).Methods The experiment in this study included two stages.In stage 1,the serum samples were randomly selected from 134 patients,and ANAs were detected by IFA at Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine from Janu-ary to June 2023.All of the samples were tested using four kinds of HEp-2 IFA kits,and the consistency of qualitative results was eval-uated by statistical analysis.The kit exhibited highest positive rate was defined as Kit X.In the stageⅡ,a total of 554 serum samples(from 218 AARD and 336 NAD patients)with positive results detected by initial screening of reagent X were selected during the same period,and then the samples were tested by the other three HEp-2 IFA kits.The patterns and titers of ANA were recorded,and a semi-quantitative evaluation system was established.The reproducibility of different patterns of ANA and the consistency of the results among varying clinical characteristics,fluorescence reaction intensities and positive reaction sites in nucleus was statistically analyzed.Results There were no significant differences of qualitative results among the results from four kits(P＞0.05).The highest positive rate ap-peared in the kit m(45.86%)which was deemed as the initial screening kit X.Significant differences in the consistency of ANA pat-terns were observed.The reproducibility scores of centromeric pattern and granular pattern were higher than those of homogeneous pat-tern,dense fine speckled pattern,nuclear cytoplasmic mixed pattern and other mixed pattern with significant difference(P＜0.05).The reproducibility score of simple pattern was higher than that of mixed patterns(P＜0.05).In the nucleoplasmic region,the consistency score of the AARD group was higher than that of NAD group(P＜0.01).The consistency scores of each reaction site increased with the rise of the intensity of reaction.In the three reaction parts(nucleoplasm,nucleolus and equatorial plate),the scores between the weak and strong fluorescence reaction intensity groups showed significant differences(P＜0.001).The lowest consistency score occurred in cytoplasmic region.Conclusion The clinical interpretation for IFA ANA reports should be more cautious for the results showing weak fluorescence intensity,mixed patterns,and staining positive cytoplasmic sites.For the choice for reagents,the clinical laboratories should be also mindful of the impacts of fluorescent secondary antibodies of anti-human immunoglobulin on the test results.The develop-ment of standardized official guidelines for the manufacture of HEp-2 IFA kits should be crucial initiative for enhancing the consistency of ANA detection and promoting mutual recognition for the results between laboratories.

Objective:To develop a prototype artificial intelligence immunofluorescence image recognition system for classification of antinuclear antibodies in order to meet the growing clinical requirements for an automatic readout and classification of immunof luorescence patterns for antinuclear antibody (ANA) images.Methods:Immunofluorescence images with positive results of ANA in Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine from April 2020 to December 2021 were collected. Three senior technicians independently and in parallel interpreted the Immunofluorescence images to determine the ANA results. Then the images were labeled according to the ANA International Consensus on Fluorescence Patterns (ICAP) classification criteria. There were 7 labeled groups: Fine speckled, Coarse speckled, Homogeneous, nucleolar, Centromere, Nuclear dots and Nuclear envelope. Each group was randomly divided into training dataset and validation dataset at a ratio of 9∶1 by using random number table. On the deep learning framework PyTORCH 1.7, the convolutional neural network (CNN) training platform was constructed based on ResNet-34 image classification network, and the automatic ANA recognition system was established. After the model was established, the test set was set up separately, the judgment results of the model were output by ranking the prediction probability, with the results of the 2 senior technicians was taken as "golden standard". Parameters such as accuracy, precision, recall and F1-score were used as indicators to evaluate the performance of the model.Results:A total of 23138 immunofluorescence images were obtained after segmentation and annotation. A total of 7 models were trained, and the effects of different algorithms, image processing and enhancement methods on the model were compared. The ResNet-34 model with the highest accuracy andswas selected as the final model, with the classification accuracy of 93.31%, precision rate of 91%, and recall rate of 90% and F1-score of 91% in the test set. The overall coincidence rate between the model and manual interpretation was 90.05%, and the accuracy of recognition of nucleolus was the highest, with the coincidence rate reaching 100% in the test set.Conclusion:The current AI system developed based on deep learning of the ANA immunofluorescence images in the present study showed the ability to recognize ANA pattern, especially in the common, typical, simple pattern.