Philip A Corrado1, Andrew L Wentland2, Jitka Starekova2, Archana Dhyani2, Kara N Goss3, Oliver Wieben2. 1. University of Wisconsin-Madison, 1111 Highland Ave, Madison, WI, 53705, USA. pcorrado2@wisc.edu. 2. University of Wisconsin-Madison, 1111 Highland Ave, Madison, WI, 53705, USA. 3. UT Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX, 75390, USA.
Abstract
OBJECTIVES: 4D flow MRI allows for a comprehensive assessment of intracardiac blood flow, useful for assessing cardiovascular diseases, but post-processing requires time-consuming ventricular segmentation throughout the cardiac cycle and is prone to subjective errors. Here, we evaluate the use of automatic left and right ventricular (LV and RV) segmentation based on deep learning (DL) network that operates on short-axis cine bSSFP images. METHODS: A previously published DL network was fine-tuned via retraining on a local database of 106 subjects scanned at our institution. In 26 test subjects, the ventricles were segmented automatically by the network and manually by 3 human observers on bSSFP MRI. The bSSFP images were then registered to the corresponding 4D flow images to apply the segmentation to 4D flow velocity data. Dice coefficients and the relative deviation between measurements (automatic vs. manual and interobserver manual) of various hemodynamic parameters were assessed. RESULTS: The automated segmentation resulted in similar Dice scores (LV: 0.92, RV: 0.86) and lower relative deviations from manual segmentation in left ventricular (LV) average kinetic energy (KE) (8%) and RV KE (15%) than the Dice scores (LV: 0.91, RV: 0.87) and relative deviations between manual segmentation observers (LV KE: 11%, p = 0.01; RV KE: 19%, p = 0.03). CONCLUSIONS: The automated post-processing method using deep learning resulted in hemodynamic measurements that differ from a manual observer's measurements equally or less than the variation between manual observers. This approach can be used to decrease post-processing time on intraventricular 4D flow data and mitigate interobserver variability. KEY POINTS: • Our proposed method allows for fully automated post-processing of intraventricular 4D flow MRI data. • Our method resulted in hemodynamic measurements that matched those derived from manual segmentation equally as well as interobserver variability. • Our method can be used to greatly accelerate intraventricular 4D flow post-processing and improve interobserver repeatability.
OBJECTIVES: 4D flow MRI allows for a comprehensive assessment of intracardiac blood flow, useful for assessing cardiovascular diseases, but post-processing requires time-consuming ventricular segmentation throughout the cardiac cycle and is prone to subjective errors. Here, we evaluate the use of automatic left and right ventricular (LV and RV) segmentation based on deep learning (DL) network that operates on short-axis cine bSSFP images. METHODS: A previously published DL network was fine-tuned via retraining on a local database of 106 subjects scanned at our institution. In 26 test subjects, the ventricles were segmented automatically by the network and manually by 3 human observers on bSSFP MRI. The bSSFP images were then registered to the corresponding 4D flow images to apply the segmentation to 4D flow velocity data. Dice coefficients and the relative deviation between measurements (automatic vs. manual and interobserver manual) of various hemodynamic parameters were assessed. RESULTS: The automated segmentation resulted in similar Dice scores (LV: 0.92, RV: 0.86) and lower relative deviations from manual segmentation in left ventricular (LV) average kinetic energy (KE) (8%) and RV KE (15%) than the Dice scores (LV: 0.91, RV: 0.87) and relative deviations between manual segmentation observers (LV KE: 11%, p = 0.01; RV KE: 19%, p = 0.03). CONCLUSIONS: The automated post-processing method using deep learning resulted in hemodynamic measurements that differ from a manual observer's measurements equally or less than the variation between manual observers. This approach can be used to decrease post-processing time on intraventricular 4D flow data and mitigate interobserver variability. KEY POINTS: • Our proposed method allows for fully automated post-processing of intraventricular 4D flow MRI data. • Our method resulted in hemodynamic measurements that matched those derived from manual segmentation equally as well as interobserver variability. • Our method can be used to greatly accelerate intraventricular 4D flow post-processing and improve interobserver repeatability.
Authors: Kelly Jarvis; Susanne Schnell; Alex J Barker; Julio Garcia; Ramona Lorenz; Michael Rose; Varun Chowdhary; James Carr; Joshua D Robinson; Cynthia K Rigsby; Michael Markl Journal: Pediatr Radiol Date: 2016-06-27
Authors: Anne Brandts; Matteo Bertini; Evert-Jan van Dijk; Victoria Delgado; Nina Ajmone Marsan; Rob J van der Geest; Hans-Marc J Siebelink; Albert de Roos; Jeroen J Bax; Jos J M Westenberg Journal: J Magn Reson Imaging Date: 2011-02 Impact factor: 4.813
Authors: Kate Hanneman; Milani Sivagnanam; Elsie T Nguyen; Rachel Wald; Andreas Greiser; Andrew M Crean; Sebastian Ley; Bernd J Wintersperger Journal: Acad Radiol Date: 2014-08 Impact factor: 3.173
Authors: Jonatan Eriksson; Carl Johan Carlhäll; Petter Dyverfeldt; Jan Engvall; Ann F Bolger; Tino Ebbers Journal: J Cardiovasc Magn Reson Date: 2010-02-12 Impact factor: 5.364
Authors: Vivian P Kamphuis; Roel L F van der Palen; Patrick J H de Koning; Mohammed S M Elbaz; Rob J van der Geest; Albert de Roos; Arno A W Roest; Jos J M Westenberg Journal: J Magn Reson Imaging Date: 2017-06-22 Impact factor: 4.813
Authors: Petter Dyverfeldt; Malenka Bissell; Alex J Barker; Ann F Bolger; Carl-Johan Carlhäll; Tino Ebbers; Christopher J Francios; Alex Frydrychowicz; Julia Geiger; Daniel Giese; Michael D Hope; Philip J Kilner; Sebastian Kozerke; Saul Myerson; Stefan Neubauer; Oliver Wieben; Michael Markl Journal: J Cardiovasc Magn Reson Date: 2015-08-10 Impact factor: 5.364
Authors: Philip A Corrado; Jacob A Macdonald; Christopher J François; Niti R Aggarwal; Jonathan W Weinsaft; Oliver Wieben Journal: BMC Med Imaging Date: 2019-12-30 Impact factor: 1.930