Inas A Yassine1, Ahmed M Ghanem2, Nader S Metwalli2, Ahmed Hamimi2, Ronald Ouwerkerk2, Jatin R Matta2, Michael A Solomon3, Jason M Elinoff4, Ahmed M Gharib2, Khaled Z Abd-Elmoniem5. 1. Systems and Biomedical Engineering Department, Faculty of Engineering, Cairo University, Egypt. 2. Biomedical and Metabolic Imaging Branch, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institutes of Health, Bethesda, MD, USA. 3. Cardiovascular Branch of the National Heart, Lung, and Blood Institute (NHLBI), NIH, Bethesda, MD, USA; Critical Care Medicine Department, NIH Clinical Center, Bethesda, MD, USA. 4. Critical Care Medicine Department, NIH Clinical Center, Bethesda, MD, USA. 5. Biomedical and Metabolic Imaging Branch, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institutes of Health, Bethesda, MD, USA. Electronic address: abdelmoniemkz@mail.nih.gov.
Abstract
BACKGROUND: Assessment of regional myocardial function at native pixel-level resolution can play a crucial role in recognizing the early signs of the decline in regional myocardial function. Extensive data processing in existing techniques limits the effective resolution and accuracy of the generated strain maps. The purpose of this study is to compute myocardial principal strain maps εp1 and εp2 from tagged MRI (tMRI) at the native image resolution using deep-learning local patch convolutional neural network (CNN) models (DeepStrain). METHODS: For network training, validation, and testing, realistic tMRI datasets were generated and consisted of 53,606 cine images simulating the heart, the liver, blood pool, and backgrounds, including ranges of shapes, positions, motion patterns, noise, and strain. In addition, 102 in-vivo image datasets from three healthy subjects, and three Pulmonary Arterial Hypertension patients, were acquired and used to assess the network's in-vivo performance. Four convolutional neural networks were trained for mapping input tagging patterns to corresponding ground-truth principal strains using different cost functions. Strain maps using harmonic phase analysis (HARP) were obtained with various spectral filtering settings for comparison. CNN and HARP strain maps were compared at the pixel level versus the ground-truth and versus the least-loss in-vivo maps using Pearson correlation coefficients (R) and the median error and Inter-Quartile Range (IQR) histograms. RESULTS: CNN-based local patch DeepStrain maps at a phantom resolution of 1.1mm × 1.1 mm and in-vivo resolution of 2.1mm × 1.6 mm were artifact-free with multiple fold improvement with εp1 ground-truth median error of 0.009(0.007) vs. 0.32(0.385) using HARP and εp2 ground-truth error of 0.016(0.021) vs. 0.181(0.08) using HARP. CNN-based strain maps showed substantially higher agreement with the ground-truth maps with correlation coefficients R > 0.91 for εp1 and εp2 compared to R < 0.21 and R < 0.82 for HARP-generated maps, respectively. CONCLUSION: CNN-generated Eulerian strain mapping permits artifact-free visualization of myocardial function at the native image resolution.
BACKGROUND: Assessment of regional myocardial function at native pixel-level resolution can play a crucial role in recognizing the early signs of the decline in regional myocardial function. Extensive data processing in existing techniques limits the effective resolution and accuracy of the generated strain maps. The purpose of this study is to compute myocardial principal strain maps εp1 and εp2 from tagged MRI (tMRI) at the native image resolution using deep-learning local patch convolutional neural network (CNN) models (DeepStrain). METHODS: For network training, validation, and testing, realistic tMRI datasets were generated and consisted of 53,606 cine images simulating the heart, the liver, blood pool, and backgrounds, including ranges of shapes, positions, motion patterns, noise, and strain. In addition, 102 in-vivo image datasets from three healthy subjects, and three Pulmonary Arterial Hypertension patients, were acquired and used to assess the network's in-vivo performance. Four convolutional neural networks were trained for mapping input tagging patterns to corresponding ground-truth principal strains using different cost functions. Strain maps using harmonic phase analysis (HARP) were obtained with various spectral filtering settings for comparison. CNN and HARP strain maps were compared at the pixel level versus the ground-truth and versus the least-loss in-vivo maps using Pearson correlation coefficients (R) and the median error and Inter-Quartile Range (IQR) histograms. RESULTS: CNN-based local patch DeepStrain maps at a phantom resolution of 1.1mm × 1.1 mm and in-vivo resolution of 2.1mm × 1.6 mm were artifact-free with multiple fold improvement with εp1 ground-truth median error of 0.009(0.007) vs. 0.32(0.385) using HARP and εp2 ground-truth error of 0.016(0.021) vs. 0.181(0.08) using HARP. CNN-based strain maps showed substantially higher agreement with the ground-truth maps with correlation coefficients R > 0.91 for εp1 and εp2 compared to R < 0.21 and R < 0.82 for HARP-generated maps, respectively. CONCLUSION: CNN-generated Eulerian strain mapping permits artifact-free visualization of myocardial function at the native image resolution.
Authors: Lorenzo Mannelli; Gregory J Wilson; Theodore J Dubinsky; Christopher A Potter; Puneet Bhargava; Carlos Cuevas; Ken F Linnau; Orpheus Kolokythas; Martin L Gunn; Jeffrey H Maki Journal: J Magn Reson Imaging Date: 2012-07-06 Impact factor: 4.813
Authors: Hans-Christian Thorsen-Meyer; Annelaura B Nielsen; Anna P Nielsen; Benjamin Skov Kaas-Hansen; Palle Toft; Jens Schierbeck; Thomas Strøm; Piotr J Chmura; Marc Heimann; Lars Dybdahl; Lasse Spangsege; Patrick Hulsen; Kirstine Belling; Søren Brunak; Anders Perner Journal: Lancet Digit Health Date: 2020-03-12
Authors: Luigi E Perotti; Ilya A Verzhbinsky; Kévin Moulin; Tyler E Cork; Michael Loecher; Daniel Balzani; Daniel B Ennis Journal: Med Image Anal Date: 2020-12-05 Impact factor: 8.545