Limits...
Multi-parametric representation of voxel-based quantitative magnetic resonance imaging.

Engström M, Warntjes JB, Tisell A, Landtblom AM, Lundberg P - PLoS ONE (2014)

Bottom Line: The resulting parameter images were normalized to a standard template.The results were visualized by conventional geometric representations and also by multi-parametric representations.In conclusion, the proposed method has the potential to visualize both high-probability focal anomalies and diffuse tissue changes.

View Article: PubMed Central - PubMed

Affiliation: Division of Radiology, Department of Medical and Health Sciences, Linköping University, Linköping, Sweden; Center for Medical Image Science and Visualization (CMIV), Linköping University, Linköping, Sweden.

ABSTRACT
The aim of the study was to explore the possibilities of multi-parametric representations of voxel-wise quantitative MRI data to objectively discriminate pathological cerebral tissue in patients with brain disorders. For this purpose, we recruited 19 patients with Multiple Sclerosis (MS) as benchmark samples and 19 age and gender matched healthy subjects as a reference group. The subjects were examined using quantitative Magnetic Resonance Imaging (MRI) measuring the tissue structure parameters: relaxation rates, R(1) and R(2), and proton density. The resulting parameter images were normalized to a standard template. Tissue structure in MS patients was assessed by voxel-wise comparisons with the reference group and with correlation to a clinical measure, the Expanded Disability Status Scale (EDSS). The results were visualized by conventional geometric representations and also by multi-parametric representations. Data showed that MS patients had lower R(1) and R(2), and higher proton density in periventricular white matter and in wide-spread areas encompassing central and sub-cortical white matter structures. MS-related tissue abnormality was highlighted in posterior white matter whereas EDSS correlation appeared especially in the frontal cortex. The multi-parameter representation highlighted disease-specific features. In conclusion, the proposed method has the potential to visualize both high-probability focal anomalies and diffuse tissue changes. Results from voxel-based statistical analysis, as exemplified in the present work, may guide radiologists where in the image to inspect for signs of disease. Future clinical studies must validate the usability of the method in clinical practice.

No MeSH data available.


Related in: MedlinePlus

Geometrical representation of tissue parameter differences between groups.The figure shows differences in tissue parameters between the group of Multiple Sclerosis (MS) patients and the reference group. Shown as the color overlay is the T-statistics of the voxel-based differences between groups. A) longitudinal relaxation rate (R); B) transversal relaxation rate (R); C) proton density (PD) in one coronal, three axial and one sagittal slice. The color scale represents T  = 0.0–5.0. The figure shows uncorrected statistics, thresholded at T = 2. In the background a synthetic T2-weighted image of the same slice is displayed for visual guidance.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4230947&req=5

pone-0111688-g002: Geometrical representation of tissue parameter differences between groups.The figure shows differences in tissue parameters between the group of Multiple Sclerosis (MS) patients and the reference group. Shown as the color overlay is the T-statistics of the voxel-based differences between groups. A) longitudinal relaxation rate (R); B) transversal relaxation rate (R); C) proton density (PD) in one coronal, three axial and one sagittal slice. The color scale represents T  = 0.0–5.0. The figure shows uncorrected statistics, thresholded at T = 2. In the background a synthetic T2-weighted image of the same slice is displayed for visual guidance.

Mentions: In Figure 1 the averaged, normalized R, R, and PD maps are shown for a single slice in the reference brain (top row) and the MS benchmark brain (bottom row). It is clearly seen that cerebrospinal fluid (CSF), WM, and GM have different characteristics and the maps can discriminate between the different tissue types using any of the three parameters. By inspecting the images it is, however, not completely evident to discriminate differences in tissue characteristics between the two groups. By voxel-based statistical analysis, on the other hand, distinct tissue differences between MS patients and the reference group were detected very clearly. In general, MS patients had lower R and R, and higher PD as compared to the reference group. Figure 2 shows differences in R, R, and PD between MS patients and the healthy reference group using a statistical threshold of T = 2. By inspecting Figure 2, marked differences between MS patients and healthy subjects are shown in periventricular WM and in wide-spread areas encompassing especially central and sub-cortical WM structures. In peripheral brain, only a few structures with differences between the groups appeared, but a number of sulci, mainly in the frontal and parietal lobes were highlighted. Correcting for multiple comparisons, we observed significant differences, p0.05, when testing for R and R lower in MS, and PD higher in MS in all regions, except occipital white matter when testing for PD (p = 0.07). No significant results were observed for the reverse comparisons. The analysis of the three different tissue structure parameters, R, R, and PD, yielded similar, but not identical results, which will be discussed further below.


Multi-parametric representation of voxel-based quantitative magnetic resonance imaging.

Engström M, Warntjes JB, Tisell A, Landtblom AM, Lundberg P - PLoS ONE (2014)

Geometrical representation of tissue parameter differences between groups.The figure shows differences in tissue parameters between the group of Multiple Sclerosis (MS) patients and the reference group. Shown as the color overlay is the T-statistics of the voxel-based differences between groups. A) longitudinal relaxation rate (R); B) transversal relaxation rate (R); C) proton density (PD) in one coronal, three axial and one sagittal slice. The color scale represents T  = 0.0–5.0. The figure shows uncorrected statistics, thresholded at T = 2. In the background a synthetic T2-weighted image of the same slice is displayed for visual guidance.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4230947&req=5

pone-0111688-g002: Geometrical representation of tissue parameter differences between groups.The figure shows differences in tissue parameters between the group of Multiple Sclerosis (MS) patients and the reference group. Shown as the color overlay is the T-statistics of the voxel-based differences between groups. A) longitudinal relaxation rate (R); B) transversal relaxation rate (R); C) proton density (PD) in one coronal, three axial and one sagittal slice. The color scale represents T  = 0.0–5.0. The figure shows uncorrected statistics, thresholded at T = 2. In the background a synthetic T2-weighted image of the same slice is displayed for visual guidance.
Mentions: In Figure 1 the averaged, normalized R, R, and PD maps are shown for a single slice in the reference brain (top row) and the MS benchmark brain (bottom row). It is clearly seen that cerebrospinal fluid (CSF), WM, and GM have different characteristics and the maps can discriminate between the different tissue types using any of the three parameters. By inspecting the images it is, however, not completely evident to discriminate differences in tissue characteristics between the two groups. By voxel-based statistical analysis, on the other hand, distinct tissue differences between MS patients and the reference group were detected very clearly. In general, MS patients had lower R and R, and higher PD as compared to the reference group. Figure 2 shows differences in R, R, and PD between MS patients and the healthy reference group using a statistical threshold of T = 2. By inspecting Figure 2, marked differences between MS patients and healthy subjects are shown in periventricular WM and in wide-spread areas encompassing especially central and sub-cortical WM structures. In peripheral brain, only a few structures with differences between the groups appeared, but a number of sulci, mainly in the frontal and parietal lobes were highlighted. Correcting for multiple comparisons, we observed significant differences, p0.05, when testing for R and R lower in MS, and PD higher in MS in all regions, except occipital white matter when testing for PD (p = 0.07). No significant results were observed for the reverse comparisons. The analysis of the three different tissue structure parameters, R, R, and PD, yielded similar, but not identical results, which will be discussed further below.

Bottom Line: The resulting parameter images were normalized to a standard template.The results were visualized by conventional geometric representations and also by multi-parametric representations.In conclusion, the proposed method has the potential to visualize both high-probability focal anomalies and diffuse tissue changes.

View Article: PubMed Central - PubMed

Affiliation: Division of Radiology, Department of Medical and Health Sciences, Linköping University, Linköping, Sweden; Center for Medical Image Science and Visualization (CMIV), Linköping University, Linköping, Sweden.

ABSTRACT
The aim of the study was to explore the possibilities of multi-parametric representations of voxel-wise quantitative MRI data to objectively discriminate pathological cerebral tissue in patients with brain disorders. For this purpose, we recruited 19 patients with Multiple Sclerosis (MS) as benchmark samples and 19 age and gender matched healthy subjects as a reference group. The subjects were examined using quantitative Magnetic Resonance Imaging (MRI) measuring the tissue structure parameters: relaxation rates, R(1) and R(2), and proton density. The resulting parameter images were normalized to a standard template. Tissue structure in MS patients was assessed by voxel-wise comparisons with the reference group and with correlation to a clinical measure, the Expanded Disability Status Scale (EDSS). The results were visualized by conventional geometric representations and also by multi-parametric representations. Data showed that MS patients had lower R(1) and R(2), and higher proton density in periventricular white matter and in wide-spread areas encompassing central and sub-cortical white matter structures. MS-related tissue abnormality was highlighted in posterior white matter whereas EDSS correlation appeared especially in the frontal cortex. The multi-parameter representation highlighted disease-specific features. In conclusion, the proposed method has the potential to visualize both high-probability focal anomalies and diffuse tissue changes. Results from voxel-based statistical analysis, as exemplified in the present work, may guide radiologists where in the image to inspect for signs of disease. Future clinical studies must validate the usability of the method in clinical practice.

No MeSH data available.


Related in: MedlinePlus