Georgios S Ioannidis1, Kostas Marias2, Nikolaos Galanakis3, Kostas Perisinakis4, Adam Hatzidakis5, Dimitrios Tsetis5, Apostolos Karantanas6, Thomas G Maris4. 1. Foundation for Research and Technology - Hellas (FORTH), Institute of Computer Science (ICS), Computational Bio-Medicine Laboratory (CBML), Heraklion, Greece; Medical School, University of Crete, Heraklion, Greece. Electronic address: geo3721@ics.forth.gr. 2. Foundation for Research and Technology - Hellas (FORTH), Institute of Computer Science (ICS), Computational Bio-Medicine Laboratory (CBML), Heraklion, Greece; Technological Educational Institute of Crete, Department of Informatics Engineering, Heraklion, Greece. 3. Department of Medical Imaging, University of Crete, Heraklion, Greece; Medical School, University of Crete, Heraklion, Greece. 4. Foundation for Research and Technology - Hellas (FORTH), Institute of Computer Science (ICS), Computational Bio-Medicine Laboratory (CBML), Heraklion, Greece; Department of Medical Physics, University of Crete, Heraklion, Greece. 5. Department of Medical Imaging, University of Crete, Heraklion, Greece. 6. Foundation for Research and Technology - Hellas (FORTH), Institute of Computer Science (ICS), Computational Bio-Medicine Laboratory (CBML), Heraklion, Greece; Department of Medical Imaging, University of Crete, Heraklion, Greece.
Abstract
PURPOSE: The purpose of this study was to correlate diffusion and perfusion quantitative and semi-quantitative MR parameters, on patients with peripheral arterial disease. In addition, we investigated which perfusion model better describes the behavior of the dynamic contrast-enhanced (DCE) MR data signal on ischemic regions of the lower limb. METHODS: Linear and nonlinear least squares algorithms, were incorporated for the quantification of the parameters through a variety of widely used models, able to extract physiological information from each imaging technique. All numerical calculations were implemented in Python 3.5 and include the: Intra voxel incoherent motion for diffusion and Patlak's, Extended Toft's and Gamma Capillary Transit time (GCTT) models for perfusion MRI. RESULTS: Our initial voxel by voxel correlation analysis didn't show any significant correlation based on the Pearson's Correlation metric between diffusion and perfusion parameters. To account for the inherited noise from the raw data, a Gaussian filter was applied to the parametric maps in order for the data to be comparable. By repeating our analysis in the filtered image maps, a good correlation (>0.5) of diffusion and perfusion parameters was achieved. CONCLUSIONS: Perfusion and diffusion MRI quantitative and semi-quantitative parameters can be obtained through a variety of physiological-pharmacokinetic models. This paper compares most of the widely-known models and parameters in both techniques with data from patients with peripheral arterial disease. Initial analysis showed no correlation in the perfusion parametric maps of DWI and DCE MRI data but a good correlation was obtained after smoothing the parametric maps indicating that perfusion information could be obtained from diffusion MRI images in patients with peripheral arterial disease.
PURPOSE: The purpose of this study was to correlate diffusion and perfusion quantitative and semi-quantitative MR parameters, on patients with peripheral arterial disease. In addition, we investigated which perfusion model better describes the behavior of the dynamic contrast-enhanced (DCE) MR data signal on ischemic regions of the lower limb. METHODS: Linear and nonlinear least squares algorithms, were incorporated for the quantification of the parameters through a variety of widely used models, able to extract physiological information from each imaging technique. All numerical calculations were implemented in Python 3.5 and include the: Intra voxel incoherent motion for diffusion and Patlak's, Extended Toft's and Gamma Capillary Transit time (GCTT) models for perfusion MRI. RESULTS: Our initial voxel by voxel correlation analysis didn't show any significant correlation based on the Pearson's Correlation metric between diffusion and perfusion parameters. To account for the inherited noise from the raw data, a Gaussian filter was applied to the parametric maps in order for the data to be comparable. By repeating our analysis in the filtered image maps, a good correlation (>0.5) of diffusion and perfusion parameters was achieved. CONCLUSIONS: Perfusion and diffusion MRI quantitative and semi-quantitative parameters can be obtained through a variety of physiological-pharmacokinetic models. This paper compares most of the widely-known models and parameters in both techniques with data from patients with peripheral arterial disease. Initial analysis showed no correlation in the perfusion parametric maps of DWI and DCE MRI data but a good correlation was obtained after smoothing the parametric maps indicating that perfusion information could be obtained from diffusion MRI images in patients with peripheral arterial disease.
Authors: Evangelia G Chryssou; Georgios C Manikis; Georgios S Ioannidis; Vrettos Chaniotis; Thomas Vrekoussis; Thomas G Maris; Kostas Marias; Apostolos H Karantanas Journal: Diagnostics (Basel) Date: 2022-03-12