HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Master Thesis

De l’immunofluorescence indirecte manuelle sur cellules HEp-2 au machine learning : état des lieux au CHU de Bordeaux

Abstract : Antinuclear autoantibodies are essential biological markers for the diagnosis of most autoimmune diseases. Their screening using the indirect immunofluorescence (IIF) assay on Human epithelial cells (HEp-2) is the first line analysis when an autoimmune assessment is required. The main issue of this analysis lies in the recognition of the many patterns found in day lab routine. Indeed, in addition to the subjective nature of any assay using a microscope, expertise is acquired only after many years of practice. With the desire for harmonization, the International Consensus on Antinuclear Antibody (ANA) Pattern" (ICAP) was established. In order to improve the efficiency of medical biology laboratories, increasingly automated systems are being developed. Our objective was to evaluate the performance of the NOVA®View (Werfen) system at Bordeaux University Hospital from a collection of 6 months of activity. This work is an important step in understanding the critical points of this assay for its improvement. The study of each sample with a positive fluorescence allowed us to quantify the level of agreement in the rendering of the fluorescence aspects, between the initial result given by the automated system and the final result after the technician’s/medical pathologist’s review. The concordance assessed by Cohen’s kappa coefficient is poor (0.077). The overall concordance is 24.3% for single patterns and 37.6% for mixed patterns. NOVA®View detected 89.4% of the cytoplasmic patterns that are clinically relevant. The development of softwares to better recognize the patterns while being in agreement with international recommendations is a major issue in the field. The arrival of artificial intelligence through machine learning, or even deep learning, offers new perspectives that could turn the market upside down.
Document type :
Master Thesis
Complete list of metadata

Contributor : Odonto Ub - Connect in order to contact the contributor
Submitted on : Thursday, January 27, 2022 - 9:28:47 AM
Last modification on : Saturday, January 29, 2022 - 3:38:52 AM


 Restricted access
To satisfy the distribution rights of the publisher, the document is embargoed until : 2022-12-06

Please log in to resquest access to the document


  • HAL Id : dumas-03545062, version 1



Maxime Dubois. De l’immunofluorescence indirecte manuelle sur cellules HEp-2 au machine learning : état des lieux au CHU de Bordeaux. Sciences du Vivant [q-bio]. 2021. ⟨dumas-03545062⟩



Record views