It is of great importance to accurately locate the optic radiation when performing a neurosurgical procedure. Neuronavigation tools offer an imaging source and a good foundation or road map of the area that is being operated. However, it can be difficult to precisely identify the location of structures due to anatomical variations from normality or changes in the pressure inside the cranial cavity once the craniotomy is performed. In delicate procedures such as those required for epilepsy surgery, a minimally-invasive few millimetres might substantially change the patient’s outcome by damaging the optic radiations.
T1-weighted scans do not identify the location of the optic radiations and tractography analysis is still being developed for the clinical environment. The primary visual area and the optic radiations cannot be clearly visualised by MRI alone because of inadequate image contrast concerning those structures. The aim of this study was to combine functional activity with DTI images to try to further improve tractography analysis, with the long-term aim of improving clinical outcomes. The functional data from the VEP recordings supplies additional information that can be used to identify the locations of the structures of interest. This section describes how these VEP recordings can be used to improve tractography analysis. The whole procedure was performed independently for each subject.
5.3.1
Segmentation process
The NeuroScan Curry 6.0 software was used for this analysis. The T1-weighted MRI was added to Curry from the original DICOM files. The first step was segmentation in order to visualise the different thresholds for skin, brain, soft tissue and CSF. The parameters for each of these was changed in Curry until the segmentation was performed satisfactorily.
The next step was to add anatomical landmarks to the imaging data. Curry produced a three-dimensional reconstruction of the imaging data with the approximate locations of the nasion, inion, right tragus and left tragus, which were fixed by hand as necessary. The intracranial structures were also identified and registered; their coordinate points are known as the Talairach parameters. Figure 5.7 shows the brain extraction of the different sources and the anatomical layers into which the images can be divided.
The final step in the segmentation process was to create a three-dimensional model by blending the MRI data with the previously-identified anatomical reference structures. Once segmentation had been completed successfully, a three-dimensional anatomical reconstruction of the brain was created that incorporated the MRI scan and the anatomical structures (Figure 5.8).
5.3.2
Visualising functional recordings
The three-dimensional image is used to show the approximate location of the electrical activity recorded and localised in the cortex by the dipole when the functional data is added and merged with the anatomical data. Fixed sources of electrical activity can be represented in Curry by a dipole or by a colour area. The dipole is created using a source reconstruction model in which the electrode positions are loaded into a virtual coordinate system. Once the model is fitted to the anatomical information that was defined in the segmentation, the fitted dipole is transformed onto the same coordinate system as the anatomical model (Fuchs et al., 2002).
The choice of locations assumes the neural model of orthogonal to cortical flow of electrical neuronal activity. Areas that are the most likely source of electrical activity are identified. Also identified are nearby cortical areas that might be additional sources of neuronal activity, in order to indicate the uncertainty in the location. Figures 5.9 and 5.10 show examples of neuronal activity from transversal and coronal views, respectively.
The combined MRI–VEP images can be exported in ANALYZE format, for use in the clinical environment (e.g. in BrainLab as a navigation tool) or in other neuroscience software.
Figure 5.7 Brain extraction in the NeuroScan Curry software. The yellow lines differentiate the skin, skull and brain. The green crosshairs are used to identify the same anatomical areas in different scans in order to achieve high- quality segmentation in three dimensions.
Figure 5.8 Anatomical reconstruction of a subject brain in Curry. The image on the left shows the head shape, including skin, looking towards the left. The image on the right shows a similarly-orientated plain reconstruction of the brain area.
Figure 5.9 Reconstruction of neuronal electrical activity in the occipital lobe from a transversal view. The figure on the right is a magnification of the rear of the brain. The colour code shows the magnitude of the recorded neurophysiological activity, with values ranging from 0.38 μA mm-2 (black/dark
Figure 5.10 Reconstruction of neuronal electrical activity in the occipital lobe from a coronal view. The colour code shows the magnitude of the recorded neurophysiological activity, with values ranging from 0.38 μA mm-2 (black/dark
red) to 0.75 μA mm-2 (yellow).
5.3.3
Combining DTI images and VEP functional recordings
DTI functionality has only recently been added to the NeuroScan Curry software. A similar processing route was followed to that described above for T1-weighted images, except that it was necessary to import the DTI data in NIfTI format. The segmentation was adapted for use with DTI data.
5.3.4
Adding VEP information to the tractography
The VEP recording data, with the dipole source locations, were added to the T1- weighted data using NeuroScan. The source location files were saved in HTM format so they could be exported to an alternative source. The FSL software was then used to co-register the files and to visualise the area corresponding to V1 in each subject. The digitised V1 area was used as a secondary seed ROI in the tractography analysis, in addition to the primary ROI located within the LGN vicinity.
The ROI was limited to areas that the VEP identified as having high probability of being the source of electrical activity (i.e. those coloured yellow in Figures 5.9 and 5.10).
Some patients, such as those with hydrocephalus, have highly distorted anatomies and it is not possible to determine the location of the LGN using clinical imaging alone and hence a seed point for tractography analysis. Tractography is a particularly valuable tool for such patients as the anatomy of the optic radiations is otherwise very
uncertain. For these patients, measuring cortical activity using VEPs and combining this with DTI images enables an alternative seed ROI to be identified so that tractography can be performed. While tractography is normally performed separately 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.