Complexity of ballooned hepatocyte feature recognition: Defining a training atlas for artificial intelligence-based imaging in NAFLD - Sorbonne Université
Article Dans Une Revue Journal of Hepatology Année : 2022

Complexity of ballooned hepatocyte feature recognition: Defining a training atlas for artificial intelligence-based imaging in NAFLD

Elizabeth M Brunt
Andrew D Clouston
  • Fonction : Auteur
David E Kleiner
Dina G Tiniakos
Aileen Wee
Matthew Yeh
  • Fonction : Auteur
Wei Qiang Leow
  • Fonction : Auteur
Elaine Chng
  • Fonction : Auteur
Yayun Ren
  • Fonction : Auteur
George Goh Boon Bee
Elizabeth E Powell
  • Fonction : Auteur
Michael Charlton
  • Fonction : Auteur
Vlad Ratziu
Stephen A Harrison
Dean Tai
  • Fonction : Auteur
Quentin M Anstee
Wei Qiang Leow
  • Fonction : Auteur
George Goh Boon Bee

Résumé

Background Histologically assessed hepatocyte ballooning is a key feature discriminating nonalcoholic steatohepatitis (NASH) from steatosis (NAFL). Reliable identification underpins patient inclusion in clinical trials and serves as a key regulatory-approved surrogate endpoint for drug efficacy. High inter/intra-observer variation in ballooning measured using the NASH-CRN semi-quantitative score has been reported yet no actionable solutions have been proposed. Methods A focussed evaluation of hepatocyte ballooning recognition was conducted. Digitised slides were evaluated by 9 internationally recognized expert liver pathologists on two separate occasions: each pathologist independently marked every ballooned hepatocyte and later provided an overall non-NASH NAFL/NASH assessment. Interobserver variation was assessed and a ‘concordance atlas’ of ballooned hepatocytes generated to train second harmonic generation/two-photon excitation fluorescence imaging-based artificial intelligence (AI). Results Fleiss kappa statistic for overall interobserver agreement for presence/absence of ballooning was 0.197 (95%CI 0.094-0.300), rising to 0.362 (0.258-0.465) with a ≥5-cell threshold. However, intraclass correlation coefficient for consistency was higher (0.718 [0.511-0.900]), indicating ‘moderate’ agreement on ballooning burden. 133 ballooned cells were identified using a ≥5/9 majority to train AI ballooning detection (AI-pathologist pairwise concordance 19–42%, comparable to inter-pathologist pairwise concordance of between 8–75%). AI quantified change in ballooned cell burden in response to therapy in a separate slide set. Conclusions The substantial divergence in hepatocyte ballooning identified amongst expert hepato-pathologists suggests that ballooning is a spectrum, too subjective for its presence or complete absence to be unequivocally determined as a trial endpoint. A concordance atlas may be used to train AI assistive technologies to reproducibly quantify ballooned hepatocytes that standardise assessment of therapeutic efficacy. This atlas serves as a reference-standard for ongoing work to refine how ballooning is classified by both pathologists and AI.
Fichier principal
Vignette du fichier
1-s2.0-S0168827822000241-main(1).pdf (9.62 Mo) Télécharger le fichier
Origine Publication financée par une institution

Dates et versions

hal-03549768 , version 1 (31-01-2022)

Identifiants

Citer

Elizabeth M Brunt, Andrew D Clouston, Zachary Goodman, Cynthia Guy, David E Kleiner, et al.. Complexity of ballooned hepatocyte feature recognition: Defining a training atlas for artificial intelligence-based imaging in NAFLD. Journal of Hepatology, 2022, ⟨10.1016/j.jhep.2022.01.011⟩. ⟨hal-03549768⟩
62 Consultations
48 Téléchargements

Altmetric

Partager

More