Reproducibility of Cardiac Volumetry: Automated vs Manual CMR Planning at 1.5T and 3T

Overview

This study compared the reproducibility of cardiac volumetric measurements derived from fully automated versus manual image plane prescription in cardiac magnetic resonance imaging (CMR) at 1.5-T and 3-T field strengths. Results demonstrated that automated AI-based plane prescription achieves reproducibility comparable to manual planning by experienced technicians, supporting its potential to standardize and simplify CMR acquisition.

Background

Cardiac magnetic resonance imaging (CMR) is the gold standard for quantitative assessment of cardiac volumes and function, essential for diagnosing and monitoring cardiovascular diseases. Manual prescription of cardiac image planes is currently standard but requires specialized training and is time-consuming, limiting accessibility. AI-based automated plane prescription promises to increase standardization, reduce variability, and improve workflow efficiency. However, evidence on the reproducibility of volumetric data derived from automated planning compared to manual methods remains limited.

Data Highlights

Key Findings

• Automated AI-based plane prescription uses deep learning to detect anatomical landmarks such as the LV apex and mitral valve from 3-plane localizer sequences.
• Automated planning achieved volumetric measurement reproducibility comparable to manual planning performed by experienced technicians.
• Reproducibility was assessed in a prospective cohort of 52 healthy adult volunteers across two sub-cohorts: intra-field strength (1.5 T) and inter-field strength (1.5 T to 3 T).
• Scan intervals were clinically relevant: 2-5 weeks for intra-field strength and 1-2 hours for inter-field strength comparisons.
• Automated planning has the potential to reduce operator dependency, decrease scan time, and increase CMR accessibility without compromising measurement reliability.

Clinical Implications

The findings support integration of AI-based automated plane prescription into routine CMR workflows to enhance standardization and reproducibility of cardiac volumetric assessments. This can facilitate more consistent follow-up imaging, reduce reliance on highly trained personnel, and potentially expand CMR availability in clinical practice. Clinicians can consider automated planning as a reliable alternative to manual methods without sacrificing measurement accuracy.

Conclusion

Automated AI-driven cardiac plane prescription provides reproducible volumetric measurements comparable to manual planning at both 1.5-T and 3-T field strengths. This advancement may improve CMR workflow efficiency and accessibility while maintaining diagnostic reliability.

References

1. Schulz-Menger et al. 2020 -- Standardized CMR protocols and guidelines
2. Messroghli et al. 2017 -- CMR in myocardial tissue characterization
3. Kramer et al. 2020 -- CMR in cardiomyopathies
4. Puntmann et al. 2018 -- CMR in myocarditis diagnosis
5. Zhang et al. 2021 -- AI in CMR plane prescription