La taxonomía de chile

Geometric image distortion results from any artefacts that violate the assumption of linear encoding of position in space with the frequency and phase encoding gradients in MR imaging (Table 5.3). Soft tissue distortion (brain shift) discussed above (Section 5.4.2) occurs in addition to this.

Equipment-induced Patient-induced

Static field inhomogeneity Gradient field non-linearity

Eddy currents from gradient switching

Magnetic susceptibility artefacts Chemical shift artefact

Flow effects

5.4.3.2

Gradient Field Non-Linearity

Gradient field non-linearity represents a technical imperfection in the gradient that leads to both geometric and intensity distortions. The distortion is specific to the scanner and independent of the patient’s position within the scanner. It is particularly important when different scanners are used, for example when combining preoperative imaging from one scanner with intraoperative imaging from another scanner.

Manufacturer-supplied software may only correct in-plane (in two-dimensions) so distortions remain through plane. For example, in a coronally acquired T1-weighted volumetric scan distortions remain in the antero-posterior direction (Figure 5.2) whilst in an axially acquired scan distortions remain in the supero-inferior direction (Figure 5.3).

The displacements can be calculated and applied from the non-linear terms of the magnetic field generated by each of the gradient coils. The fields can be supplied by the vendor as a truncated series of spherical harmonic coefficients (Janke et al. 2004). Corrections are applied to pre- and intraoperative data used for neurosurgical guidance in iMRI using custom written software in this thesis (Chapters 6 and 10).

5.4.3.3

Magnetic Susceptibility Artefacts

A patient in a scanner induces microscopic variations in the magnetic field strength. At interfaces with large magnetic susceptibility differences, such as air/tissue or bone/tissue boundaries, this results in significant geometric (and intensity) distortions. It thus particularly affects the anterior temporal lobe and in the context of neurosurgery, the resection cavity introduces air and thus susceptibility differences and distortions around the area of resection.

Echo-planar imaging (EPI) is used for functional MRI and DTI sequences due to its high temporal resolution. However since it samples the entire frequency space of a single slice with fast gradient blipping following a single excitation pulse, this results in a very low bandwidth in the phase encoding direction. Thus EPI sequences are particularly susceptible to magnetic susceptibility artefacts. As the phase encoding direction is typically anterior-posterior, this is critical when considering the anterior extent of a structure such as the optic radiation.

Correction of susceptibility induced distortion may be performed either by field map estimation or non- linear image registration. Field maps are acquired at the same time as the EPI data to estimate the magnetic field inhomogeneity from two phase images acquired at different echo times (Jezzard & Balaban 1995). Non-linear image registration is performed between the distorted EPI image and a high resolution undistorted T1-weighted anatomical image (Merhof et al. 2007). EPI images acquired intraoperatively have low signal-to-noise ratio and artefacts which make this approach challenging. The former approach with field maps and custom written software is used in this thesis (Figure 5.4).

Figure 5.2 - Gradient non-linearity correction in a human subject

T1-weighted volumetric image from scanner with on scanner in acquisition correction in third (anterior

respectively).

Figure 5.3 - Gradient non-linearity correction in a phantom

Axial acquisition with top row showing coronal plane, and bottom row showing axial plane. Uncorrected scan (A), scan with on-scanner in-plane (axial) correction (B), and image after post acquisition correction in third (superior inferior) dimension (C).

linearity correction in a human subject

weighted volumetric image from scanner with on scanner in-plane (coronal) correction (A), image after post acquisition correction in third (anterior-posterior) dimension (B), and comparison of imag

linearity correction in a phantom

Axial acquisition with top row showing coronal plane, and bottom row showing axial plane. Uncorrected scan (A), plane (axial) correction (B), and image after post acquisition correction in third (superior

plane (coronal) correction (A), image after post posterior) dimension (B), and comparison of images (red and green

Axial acquisition with top row showing coronal plane, and bottom row showing axial plane. Uncorrected scan (A), plane (axial) correction (B), and image after post acquisition correction in third (superior-

Figure 5.4 - Magnetic susceptibility artefact correction in in

Intraoperative T1-weighted scan (A), non uncorrected (B) and field-map corrected (C). Image

5.5 Conclusions

There are several approaches to intraoperative

temporal and spatial resolution, good soft tissue contrast and full multiplanar imaging without any radiation exposure. It is safe and can be use to maximise the degree of resection whilst minimisin associated morbidity in both tumour and epilepsy surgery. Technical challenges including correction for brain shift and geometric image distortion must be addressed in order to use the data optimally.

Magnetic susceptibility artefact correction in intraoperative data

weighted scan (A), non-diffusion weighted image from DTI sequence acquired with EPI map corrected (C). Image courtesy of Pankaj Daga.

There are several approaches to intraoperative imaging but intraoperative MRI has the advantages of high temporal and spatial resolution, good soft tissue contrast and full multiplanar imaging without any radiation exposure. It is safe and can be use to maximise the degree of resection whilst minimisin associated morbidity in both tumour and epilepsy surgery. Technical challenges including correction for brain shift and geometric image distortion must be addressed in order to use the data optimally.

diffusion weighted image from DTI sequence acquired with EPI

imaging but intraoperative MRI has the advantages of high temporal and spatial resolution, good soft tissue contrast and full multiplanar imaging without any radiation exposure. It is safe and can be use to maximise the degree of resection whilst minimising the associated morbidity in both tumour and epilepsy surgery. Technical challenges including correction for brain shift and geometric image distortion must be addressed in order to use the data optimally.

6 STUDYDESIGNANDGENERICMETHODS

6.1 Introduction

The studies described in this thesis predominantly involve a cohort of patients with refractory TLE undergoing surgery who were prospectively recruited and followed up longitudinally. This chapter introduces the cohort, their clinical and demographic features and the main assessments undertaken including MRI, neuropsychology and visual fields. Specific features of the methods for each study are included in the relevant chapters.