Quantification of Late Gadolinium Enhancement in Acute Myocarditis: A Comparative and Longitudinal Study of Automatic and Manual Thresholding Methods

Presenting Author(s)

• KB

  Khaoula Bouazizi, PhD

  Rsearch engineer
  Laboratoire d'Imagerie Biomédicale
  Paris, Ile-de-France, France

Primary Author(s)

• KB

  Khaoula Bouazizi, PhD

  Rsearch engineer
  Laboratoire d'Imagerie Biomédicale
  Paris, Ile-de-France, France

Co-Author(s)

• MZ

  Mohamed ZARAI, MSc

  MSc
  ICAN Imaging, Institute of Cardiometabolism and Nutrition (ICAN), Paris, France, France
• AT

  Anne Tavian, BSc

  Intern
  ICAN Imaging, Institute of Cardiometabolism and Nutrition (ICAN), France
• PG

  Paul Garrigoux, BSc

  Intern
  ICAN Imaging, Institute of Cardiometabolism and Nutrition (ICAN), France
• ND

  Naoual Djebra, MSc

  Engineer
  University Hospital Pitie-Salpetriere, France
• LN

  Lan-anh Nguyen, MD

  MD
  Assistance Publique – Hôpitaux de Paris, France
• TW

  Tatiana Wong, MD

  Attending physician
  Sorbonne Université, Department of Cardiovascular Imaging, Sorbonne Université, Pitié-Salpêtrière University Hospital AP-HP,  INSERM 1146, CNRS 7371, Laboratoire d'Imagerie Biomédicale, Paris, France, France
• CC

  Cécile Chung, MD

  Dr
  University Hospital Pitie-Salpetriere, France
• SS

  Salim Si-Mohamed, MD

  Medical Doctor
  Louis Pradel Hospital, Hospices Civils de Lyon; University of Lyon, France
• TG

  Thibaud Genet, MD

  Dr
  5. Cardiology Department, University Hospital of Tours,, France
• OL

  Olivier Lairez, MD

  Dr
  CHU de Toulouse, France
• JD

  Jean-François Deux, MD, PhD

  Radiologist
  Geneva University Hospitals, Switzerland
• EM

  Elie Mousseaux, MD, PhD

  Professor
  Université Paris Cité, INSERM, PARCC, APHP, Hôpital Européen Georges Pompidou, France
• BA

  Benjamin Alos, MD

  Attending physician
  University of Poitiers, Clinical Investigation Center (Inserm 1402), University Hospital of Poitiers, Cardiology Department, Poitiers, FACT network, France, France
• 

  Alexis Jacquier, MD, PhD

  Prof
  Aix Marseille Univ, CNRS, CRMBM, Marseille, France; APHM, Hôpital Universitaire Timone, Radiology Dept, Marseille, France
  Marseille, Provence-Alpes-Cote d'Azur, France
• PO

  Phalla Ou, MD, PhD

  Professor in radiology
  Radiology Department, Bichat University Hospital, Assistance Publique–Hôpitaux de Paris, Paris, France, France
• CB

  Claire Bouleti, MD

  Dr
  University of Poitiers, Clinical Investigation Center (INSERM 1402), Cardiology Department, Poitiers Hospital, Poitiers, France
  Massy, France
• 

  Nadjia Kachenoura, PhD

  PhD
  Laboratoire d'Imagerie Biomédicale, Sorbonne Université
  Paris, Ile-de-France, France
• 

  Alban Redheuil, MD PhD

  MD, PhD
  Assistance Publique – Hôpitaux de Paris, France

Background: Late gadolinium enhancement (LGE) imaging has become a critical tool for evaluating myocardial injury, particularly in acute myocarditis¹. The aim of this study is to investigate LGE quantification methods in the largest French population cohort² with acute myocarditis.

Methods: We analyzed 192 patients (mean age: 33±13, sex: 104 men, median follow-up: 6 months) with both baseline (acute-phase) and follow-up CMR who were enrolled in the MyocarditIRM³ Study, and for whom LGE images were suitable for analysis. LGE imaging was acquired 10–15 minutes after intravenous gadolinium administration on either 1.5T or 3T systems. Endocardial and epicardial contours were manually segmented using a dedicated software package (MRFibrosis⁴, Sorbonne Université, France). LGE burden was quantified using multiple approaches with both automated and manual threshold-based techniques and reported as a percentage of total LV mass in % (LGE fraction). Specifically, the following methods were evaluated: (1) automated fuzzy c-means clustering; (2) automated thresholding at 2, 3, 4, 5, and 6 standard deviations (kSD) above reference myocardium; and (3) manual thresholding at 2, 3, 4, 5, and 6SD, with classification based on a user-defined reference region (standard approach). In the automated mode, the healthy myocardial reference was selected automatically, whereas in the manual mode, this region was operator-defined. An LGE “grey zone” was delineated for both automated and manual methods as voxels within the [2SD–5SD] range above reference myocardium.

Results: Quantitative analysis of LGE revealed a constant decrease in LGE fraction during follow-up (Figure 1). The amount of LGE decreased from 14% to 7% using the fuzzy c-means method between the baseline and follow-up visits, while the gray zone showed a 9% to 5% reduction over the same period. Automatic threshold techniques, particularly those using kSD showed significant correlation with manual measurements (r >0.15, p< 0.0001) and generated systematically lower values than manual methods. LGE as well as the gray zone quantification correlated significantly between manual segmentation method and fuzzy means clustering method both for baseline and follow-up exams (p< 0.0001).

Conclusion: This study represents a detailed exploration of LGE imaging in the context of acute myocarditis within the French population, using a wide range of LGE measurement techniques. The findings underscore the importance of method selection in quantifying myocardial injury and highlight the variability in LGE quantification across different approaches. Future studies should aim to compare the different methods in order to assess their capacity to predict clinical events in acute myocarditis.

Evolution of LGE and gray zone amounts between baseline and follow-up, assessed using different methodologies.