Introduction of Lazy Luna an automatic software-driven multilevel comparison of ventricular function quantification in cardiovascular magnetic resonance imaging

Abstract

Cardiovascular magnetic resonance imaging is the gold standard for cardiac function assessment. Quantification of clinical results (CR) requires precise segmentation. Clinicians statistically compare CRs to ensure reproducibility. Convolutional Neural Network developers compare their results via metrics. Aim: Introducing software capable of automatic multilevel comparison. A multilevel analysis covering segmentations and CRs builds on a generic software backend. Metrics and CRs are calculated with geometric accuracy. Segmentations and CRs are connected to track errors and their effects. An interactive GUI makes the software accessible to different users. The software’s multilevel comparison was tested on a use case based on cardiac function assessment. The software shows good reader agreement in CRs and segmentation metrics (Dice > 90%). Decomposing differences by cardiac position revealed excellent agreement in midventricular slices: > 90% but poorer segmentations in apical (> 71%) and basal slices (> 74%). Further decomposition by contour type locates the largest millilitre differences in the basal right cavity (> 3 ml). Visual inspection shows these differences being caused by different basal slice choices. The software illuminated reader differences on several levels. Producing spreadsheets and figures concerning metric values and CR differences was automated. A multilevel reader comparison is feasible and extendable to other cardiac structures in the future.

Introduction of a cascaded segmentation pipeline for parametric T1 mapping in cardiovascular magnetic resonance to improve segmentation performance

“…With respect to an intra-observer performance of 72% DSC and 15.61 mm HD in native data and 83% DSC and 9.03 mm HD in CE data the CASEG pipelines showed a robust geometric outcome compared to a human reader 13 . However, errors made by the automated segmentation are prone to be atypical as compared to a human reader 30 such that the human segmentation is not necessarily substitutable by a completely unsupervised CASEG pipeline at the current stage.…”

“…In CMR, the development of automated segmentation methods based on CNNs aims to substitute the necessity of an expert segmentation 30 . More complex network structures showed an improvement in segmentation quality 8,9,14 while it is also known that the segmentation quality highly depends on the input data quality [31][32][33] .…”

Multilevel comparison of deep learning models for function quantification in cardiovascular magnetic resonance: On the redundancy of architectural variations

“…Ethical approval for this retrospective study was obtained from the ethics committee of Charité — Universitätsmedizin Berlin (approval number EA1/367/20). Part of this work has been presented at the scientific sessions of the 2022 Joint Annual Meeting ISMRM-ESMRMB & ISMRT 31st Annual Meeting ( 16 ).…”

“…Network predictions for the 29 test cases were evaluated against the expert segmentations using the recently published dedicated software Lazy Luna ( 29 ). Analyses were performed on image (segmentation metrics) and patient (clinical parameters) levels using contours and not pixel-masks for all calculations.…”

Kardiovaskuläre MRT: akute Myokarditis und myokardiale Mitbeteiligung bei Systemerkrankungen

“…In der Zukunft wird es notwendig sein, standortübergreifende Qualitätskriterien in der Bildgebung einzuführen, die ähnlich den Kriterien in Labor und Industrie eine bessere Vergleichbarkeit von Ergebnissen ermöglichen. Entsprechende erste Anstrengungen werden bereits unternommen, sowohl im Einzelnen [44] als auch am Anfang der Bildgebungskette im Sinne der Etablierung von Ausbildungsstandards [45].…”