Automated mitral inflow Doppler peak velocity measurement using deep learning.
Jevgeni Jevsikov, Tiffany Ng, Elisabeth S Lane, Eman Alajrami, Preshen Naidoo, Patricia Fernandes, Joban S Sehmi, Maysaa Alzetani, Camelia D Demetrescu, Neda Azarmehr, Nasim Dadashi Serej, Catherine C Stowell, Matthew J Shun-Shin, Darrel P Francis, Massoud Zolgharni
Author Information
Jevgeni Jevsikov: School of Computing and Engineering, University of West London, United Kingdom; National Heart and Lung Institute, Imperial College London, United Kingdom. Electronic address: Jevgeni.Jevsikov@uwl.ac.uk.
Tiffany Ng: National Heart and Lung Institute, Imperial College London, United Kingdom.
Elisabeth S Lane: School of Computing and Engineering, University of West London, United Kingdom.
Eman Alajrami: School of Computing and Engineering, University of West London, United Kingdom.
Preshen Naidoo: School of Computing and Engineering, University of West London, United Kingdom.
Patricia Fernandes: School of Computing and Engineering, University of West London, United Kingdom.
Joban S Sehmi: West Hertfordshire Hospitals NHS Trust, Wafford, United Kingdom.
Maysaa Alzetani: Luton & Dunstable University Hospital, Bedfordshire, United Kingdom.
Camelia D Demetrescu: Luton & Guy's and St Thomas' NHS Foundation Trust, London, United Kingdom.
Neda Azarmehr: School of Computing and Engineering, University of West London, United Kingdom.
Nasim Dadashi Serej: School of Computing and Engineering, University of West London, United Kingdom.
Catherine C Stowell: National Heart and Lung Institute, Imperial College London, United Kingdom.
Matthew J Shun-Shin: National Heart and Lung Institute, Imperial College London, United Kingdom.
Darrel P Francis: National Heart and Lung Institute, Imperial College London, United Kingdom.
Massoud Zolgharni: School of Computing and Engineering, University of West London, United Kingdom; National Heart and Lung Institute, Imperial College London, United Kingdom.
Doppler echocardiography is a widely utilised non-invasive imaging modality for assessing the functionality of heart valves, including the mitral valve. Manual assessments of Doppler traces by clinicians introduce variability, prompting the need for automated solutions. This study introduces an innovative deep learning model for automated detection of peak velocity measurements from mitral inflow Doppler images, independent from Electrocardiogram information. A dataset of Doppler images annotated by multiple expert cardiologists was established, serving as a robust benchmark. The model leverages heatmap regression networks, achieving 96% detection accuracy. The model discrepancy with the expert consensus falls comfortably within the range of inter- and intra-observer variability in measuring Doppler peak velocities. The dataset and models are open-source, fostering further research and clinical application.
