الفهرس 
Introduction and aims of the work
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 181 |  |  |
| | | from | 181 | | |
Abstract
Glioma is the most common intra-axial brain tumor characterized by invasion into the surrounding white matter (WM) tracts.
Recently, advanced diffusion tensor MRI enables the non-invasive assessment of glioma morphology and functionality providing a point of likeness into histopathological grading of the tumor and serving in a more successful gliomas’ diagnosis, prognosis, therapy evaluation, monitoring of disease progression and planning of neurosurgical strategies. Using DTI sequences, it is possible to explore the integrity of white matter structures and detect abnormalities not visible in the conventional MRI acquisitions, as well as help neurosurgeons in preoperative assessment of the gliomas neighbouring white matter tracts
In our study, we retrospectively recruited 46 patients who were provisionally diagnosed to have untreated gliomas based on conventional MR imaging, referred to have a pre-surgical MRI evaluation at the National Hospital for Neurology and Neurosurgery, London, UK. Conventional 3D T1WI, T2WI and FLAIR MRI sequences in addition to diffusion tensor images with 60 diffusion weighted directions at b value 1400 sec/mm2.
The aims of the work were subdivided into two studies; first one aimed at assessing the role of diffusion tensor metrics (FA, MD, AD, RD) in differentiating gliomas grades and their molecular subtypes according to 2016 WHO classification for brain tumors. Second study aimed at comparing and assessing the level of agreement between two different fiber tracking software probabilistic methods in delineating corticospinal tract (hand, foot and lips fibers), arcuate fasciculus, and optic radiation to obtain reliable fiber tractography for robust presurgical planning.
In the first study, we defined three volumes of interest (tumor, perilesional white matter PLWM, normal appearing white matter NAWM) and identified 23 low grade (LGG) and 23 high grade gliomas (HGG). We found that the only DTI metrics which could differentiate LGG from HGG were FA mean at the tumor region [AUC 0.716, cutoff point >0.14 identified via the Youden Index, with specificity of 65.23% and a sensitivity of 73.9%] as well as MD, AD, and RD means at the PLWM [AUC were 0.708, 0.693, 0.677 respectively with specificity/sensitivity 91.3%/56.52%, 91.3%/47.83%, 52.17/78.26% respectively].
Comparison between volumes of interests, there was a highly statistically significant difference for all DTI metrics between both tumor and PLWM regions. Also, between tumor and NAWM volumes, all of DTI metrics were statistically significant except for the AD minimum value. While comparing PLWM with NAWM showed a highly statistically significant difference for all DTI metrics in between except for FA maximum value and MD, AD, and RD minimum values.
For differentiating between IDH-mutant glioma/IDH-wild type (non-mutant) glioma, only two parameters showed statistically significant difference; FA mean within the tumor area [AUC was 0. 761, cutoff point ≤0.158 identified via the Youden Index, specificity of 72.73% and a sensitivity of 74.29%] and AD minimum within the tumor area [AUC was 0.703, cutoff point was >0.00094 according to the Youden Index with a specificity of 81.82% and a sensitivity of 60%].
AD mean and AD minimum are the only diffusion metrics which could differentiate between MGMT methylated versus MGMT non-methylated gliomas within the tumor area, AUC value 0.684, 0.668 respectively with cutoff point ≤0.0014 and ≤0.00098 respectively identified via the Youden Index. These thresholds resulted in a specificity of 72.73%, 50% and a sensitivity of 70.83%, 79.17% respectively.
FA mean in the tumor area is the only diffusion metric who differentiated between 1p19q co-deleted and 1p19q non co-deleted gliomas, AUC value was 0. 669, with cutoff point ≤0.159 identified via the Youden Index. This threshold resulted in a low specificity of 46.15% and a high sensitivity of 85%.
In the second study, the qualitative analysis showed that the highest degree of overlap percentage between both methods was complete overlap in AF and OR (rest of OR) being 65% and 55.56% respectively, while near_complete overlap was the highest degree percentage in the generated CST foot, hand, lips fibers as well as Meyer’s loop of OR by 55%, 55%, 40%, and 31.25% respectively.
There was a strong, positive correlation between both methods’ generated tracts’ volume and also between most of the generated DTI parameters (FA, MD, AD, RD) for both methods’ tracts. Only the CST-FOOT fibers, AF, and Rest of OR showed strong, positive correlation between quantitative Dice score and qualitative degree of overlap with no positive correlation could be detected in CST-hand and lips fibers as well as Meyer’s loop of OR.