Στοιχεία επιβλεπόντων καθηγητών:
Ευσταθόπουλος Ευστάθιος, Αναπληρωτής Καθηγητής, Ιατρική, ΕΚΠΑ
Κελέκης Νικόλαος, Καθηγητής, ΕΚΠΑ
Καραβασίλης Ευστράτιος, Διδάκτορας, ΕΚΠΑ
Diffusion weighted imaging (DWI) is a non-invasive technique that combines qualitative and quantitative information about the mobility of the molecules of water in tissues by assuming a monoexponential attenuation for the magnetic resonance (MR) signal. B factor is a parameter that encloses the intensity of the diffusion sensitizing gradients and in clinical practice b factor fluctuates approximately from 500 to 1000 sec/〖mm〗^2. The limitation for DWI is its dependence on the intravascular motion with the outcome of providing no reliable results, for example in normal appearing white matter (NAWM). The development on DWI is diffusion tensor imaging (DTI) which quantifies the spatial information on the behavior of diffusion by assuming free Gaussian displacement distribution. However, DTI is limited to isotropic tissues as well as to tissues with complex microstructure where free diffusion collapses [1,2].
In the recent years, the interest of research has been focused on the behavior of diffusion at ultralow and ultrahigh b values where the deviation from the Gaussian displacement distribution and the monoexponential model is very intense. In order to reconcile the conflicting findings, scientists have proposed three different models; the biexponential or IVIM model, the stretched exponential model and the kurtosis (DK) model [3-6]. Each model quantifies the behavior of diffusion to parametric maps purporting to show more accurate measurements. Simultaneously several papers investigate the probable correlation of diffusion parameters to pathological conditions [2,7-10].
This study is focused on the IVIM model. Intravoxel Incoherent Motion (IVIM) imaging is a method that provides quantitative assessment of all the microscopic translational motions that could contribute to the signal acquired with diffusion weighted imaging (DWI) in MRI. The fundamental idea was that the molecular motion of water is randomly oriented in the capillary network at ultralow b values of diffusion imaging, mimicking a random walk. Diffusion parameters derived from IVIM model - perfusion fraction f, true-ADC (D_slow), pseudo-ADC (D_fast) – have roused the researchers' interest for their biological basis.
The aim of this study has two goals, first the development of an algorithm to quantify the IVIM model diffusion parameters - fraction f, D_fast and D_slow - with their parametric maps and second a clinical practice to patients with brain lesions with the correlation of the IVIM model diffusion parameters in contra lateral regions of interest.
The diffusion parameters〖 D〗_slow, D_fast and f were calculated using a biexponential curve fitting model based on Levenberg - Marquardt algorithm applied in an in-house platform in Matlab. The development of the algorithm was based on three methods in order to choose the more efficient technique [57-61], first by fitting simultaneously the 3 parameters, second by fitting D_slow(high b) and then fitting f,D_fast , third by linearly fitting f,D_slow (high b) and then fitting〖 D〗_fast .
For the MRI experiment seven patients with diagnosed brain lesions (ischemia, inflammation, gliosis etc.) underwent Diffusion Weighted Imaging (DWI) with a single-shot spin echo planar imaging (EPI) sequence with 10 b values (0,10,15,50,80,100,200,400,700,1000 s/mm2).
The three methods were applied on a female woman with inflammatory brain lesions. Comparing the R square values, method 3 describes a better fitting process for the majority of pixels. Histogram for the counts of R2 near unit was 49,72% for method 3, 46,99% for method 2 and 43,44% for method 1. In clinical practice we found in one patient (inflammation) increased values for f (0.12±0.07 versus 0.07±0.03) and in two patients (gliosis, ischemia) decreased values for f (0.08±0.05 versus 0.10±0.06 and 0.05±0.03 versus 0.09±0.06, respectively) revealing a hyper and a hypo perfusion region respectively. The parameters D_fast and D_slow do not seem to expose any significant differentiations.
The conclusion of this study is that parameter f may constitute a new sensitive biomarker and must be examined further. The f maps may provide additional information on conventional diffusion parameters and may predict preliminary brain damage.
This study was conducted as part of a “thesis” for the interdepartmental program of postgraduate studies in medical physics of the faculty of medicine of the University of Athens.