for the left and right optic radiations, this approach is not appropriate for subjects with highly distorted anatomies and no exclusion ROIs are defined. This method was used to analyse two hydrocephalus patients and the results are examined in Chapter 8.
5.4 Adult control cohort tractography using VEP
The impact of combining VEP functional data with imaging data was tested using a cohort of adult control subjects, as these were more cooperative in concentrating throughout the VEP recording so were expected to produce functional data with lower noise than children. The same multichannel scanning procedure was performed for all subjects. Tractography was performed both with and without the VEP data and the difference in the mean FA was assessed statistically.
The cohort consisted of 6 males and 6 females aged between 26 and 40, with one further male aged 65. None of the subjects wore glasses at the time of enrolment. They were healthy with no diagnosis of psychiatric or other organic disease that could bias the results. None of subjects had regular anti-epileptic medication or antidepressants at the time of scanning.
5.4.1
Qualitative impact of using functional VEP data on
tractography
An example of the impact of including a second ROI seeding region based on VEP data is shown for one subject in Figure 5.11. Each optic radiation was analysed separately but both have been combined in these images. The functional data enables the tractography to identify the visual regions at the rear of the brain that are often omitted by the standard method, and were generally found to improve the tracking of the optic radiations across the cohort.
(a) MRI (b) MRI + VEP
Figure 5.11 Impact of including a second ROI seeding region based on VEP functional data. The tractography maps are for the same subject for: (a) defining seed points only in the vicinity of the LGN; and, (b) defining additional seed points in the V1 area based on functional VEP recordings.
5.4.2
Quantitative impact of using functional VEP data on
tractography
The numbers of voxels in each optic radiation are summarised in Table 5.1 for the 13 adult controls. Using the functional ROI seed method leads to a much greater number of voxels at the rear on both optic radiations. Although the variability, represented by the standard deviation, is higher for the functional ROI seed method, the relative error9 is lower than for the standard seed method.
The impact of including a second ROI seeding region, based on VEP data, on the left optic radiation mean FA is shown in Figure 5.12. For the standard method, the mean FA varies between 0.4 and 0.5 across the cohort. A similar trend is observed across the cohort for the new method with a second ROI seeding region, but most of the values are around 0.05 lower. A similar trend is observed for the right optic radiation in Figure 5.13, but with an average difference of around 0.1.
Left Right
MRI MRI + VEP MRI MRI + VEP
Full 466 ± 185 812 ± 276 336 ± 149 640 ± 246
Front 173 ± 88 173 ± 77 126 ± 68 114 ± 80
Rear 293 ± 116 639 ± 228 209 ± 98 526± 215
Table 5.1 Number of voxels in each optic radiation from the standard and functional ROI seed methods for adult controls. Figures are presented in the form mean ± standard deviation.
Figure 5.12 Impact of using a functional-derived ROI on the left optic radiation mean FA for adult controls. The mean FA for the standard tractography method is coloured blue and the mean FA for the same subjects with a secondary seed ROI defined using functional VEP data is coloured red.
0.0 0.1 0.2 0.3 0.4 0.5 0 10 20 30 40 50 60 70 F A Age Standard ROI
Figure 5.13 Impact of using a functional-derived ROI on the right optic radiation mean FA for adult controls. The mean FA for the standard tractography method is coloured blue and the mean FA for the same subjects with a secondary seed ROI defined using functional VEP data is coloured red.
The cause of the difference in mean FA can be investigated by comparing changes in the front and rear mean FA. The mean FA of the front of the left optic radiation does not show any major differences between the two methods (Figure 5.14). For the rear, however, the mean FA when the extra ROI seeding region is included is lower for all subjects (Figure 5.15). When using an extra seeding region, the optic tracts extend to the rear of the brain and include a greater number of voxels, as illustrated in Figure 5.11. These voxels tend to have lower FA than the other parts of the tract, which causes the mean FA across the tract to reduce. The changes in front of the parieto- occipital sulci tend to be much more minor as the tracts are mainly derived from the primary seed point in the vicinity of the LGN, and this is defined in the same way in both analyses for each patient.
Statistical differences between the two cohorts are examined in Table 5.2. Adding a secondary seeding region causes statistically-significant reductions in the full and rear mean FA on both optic radiations, but has no statistically-significant impact on the front of the optic radiations mean FA.
0.00 0.10 0.20 0.30 0.40 0.50 0 20 40 60 80 F A Age Standard ROI
Figure 5.14 Impact of using a functional-derived ROI on the mean FA for the front of the left optic radiation for adult controls. The mean FA for the standard tractography method is coloured blue and the mean FA for the same subjects with a secondary seed ROI defined using functional VEP data is coloured red.
Figure 5.15 Impact of using a functional-derived ROI on the mean FA for the rear of the left optic radiation for adult controls. The mean FA for the standard tractography method is coloured blue and the mean FA for the same subjects with a secondary seed ROI defined using functional VEP data is coloured red.
0.0 0.1 0.2 0.3 0.4 0.5 0 10 20 30 40 50 60 70 F A Age Standard ROI 0.0 0.1 0.2 0.3 0.4 0.5 0 10 20 30 40 50 60 70 F A Age Standard ROI
Change in mean FA T-test
Left Right Left Right
Full –0.05 –0.10 <0.00001 <0.00001
Front +0.01 –0.02 0.463 0.108
Rear –0.07 –0.12 <0.00001 <0.00001
Table 5.2 Paired t-test comparing the mean FA from the standard and functional ROI seed methods for adult controls.
5.5 Discussion
The tractography is qualitatively improved by including an additional ROI seeding region based on VEP recordings. It is important to consider why this causes a substantial reduction in the mean FA of the rear of the brain.
Mapping white matter tracts using DTI tractography in areas with high amounts of grey matter is difficult. The rear of the optic radiation is at the very end of a transition area between the visual primary cortex and the axonal white matter pathway confined in the optic radiation. It is a very rich area that is characterised by the pruning and division of the terminal ends of the optic radiation, which do not follow a straight line. The tortuous nature of the fibres means that they could spread out in multiple directions, which makes them difficult to follow and acts as a confounding factor for tractography. The photographs of the macroscopic dissection presented in Section 2.2 illustrate the anatomy of these fibres.
In the tractography analysis, the mathematical tracking system excludes voxels in which the FA is lower than the threshold value of 0.1. In contrast with the standard PiCo algorithm (Parker et al., 2003), the tractography algorithm in this study did not include angle limitations, so the rear area, which has many transition voxels between grey and white matter, was not excluded. Yet the probabilistic model has a voxel size of 2.5 mm, which is large in comparison to the size of axonal fibres of the optic radiations (Ebeling and Reulen, 1988). This voxel size is more suitable to examine larger fibres with constant diameters, but it was not possible to produce DTI scans at higher resolutions with the available MRI machine. This resolution is suitable for the
optic radiations in the vicinity of the LGN, where the fibres are more homogenous and larger.
The volumes of the tracked optic radiations are larger when VEP data is used to set ROIs, leading to the inclusion of lower-FA areas at the rear of the optic radiations. A higher-resolution scan might improve the FA estimate in transitional areas with more complex fibres.