Table 1 Factors that have an effect on the ability to detect epileptic lesions on MRI

Protocol

Slice thickness

Thinner slices give more chance of lesion detection. A routine scan with 5 mm thick slices and 0.5 mm interslice gaps with T1W and T2W transverse image acquisitions and gadolinium contrast enhancement may be adequate to evaluate gross cerebral abnormalities such as large tumours or malformations but may not detect subtle epileptic lesions. Slice thickness of 3 mm or less in at least 2 orientations is recommended for examination of the epileptic brain and larger slice size risks missing lesions less than 5 mm [38]. However MRI machines of 1 T or less cannot provide thin slices with sufficient SNR within reasonable time. For this reason machines under 1.5 T are considered insufficient for the imaging of human epilepsy patients unless there is no alternative [38].

Sequence

Failure or inability to select the appropriate sequences to detect lesions. For example in humans, high resolution, volumetric and 3D MRI acquisition is recommended to obtained detailed information on hippocampal anatomy, cortical gyral patterns, improve grey and white matter contrast and to enable co-registration with other modalities or sequential MRI examinations [13, 38]. This requires a good quality machine (1.5 T or more) and careful orientation of slice plane relative to patient position. FLAIR sequence is regarded as the most useful image for detecting epileptic lesions in humans [38] however many low field machines produce FLAIR with low resolution.

Magnetic field strength

Low field versus high field

Imaging with higher magnetic field-strength provides improved signal-to-noise ratio and spatial resolution which allows shorter imaging times for a given resolution and/or higher resolution for a given imaging time. Higher signal-to-noise ratio allows better resolution with smaller voxel size and thinner slice thickness [7].

1.5 T versus 3 T

Coil

Type of coil used (for example Knee vs Head coil)

Coils with minimum distance between receiving coil and brain surface and minimal diameter increase SNR and therefore image quality. Some coils (for example brain coils) may limit the field of view that can be imaged before significant signal drop-off occurs. The lack of availability of dog-specific coils and variation in dog head size makes coil selection challenging in some cases.

Available channels

An 8 channel brain coil is usual in veterinary MRI but a 32 channel brain coil will provide much better SNR and contrast resolution.

Operator factors

Inexperience / lack of training

A fully trained radiography technician understands the physics of MRI and anatomy allowing them to create images with excellent contrast and clarity and target the brain structures to be studied. Typically, a trained MRI technician has undertaken a 3-year radiography degree plus an additional 2–3 years of post-graduate MRI training. A poorly trained or unqualified operator may not be able to achieve optimal results from the machine that they have. In veterinary medicine it is possible to operate a MRI service without a specialist qualification.

Diligence

There are ways of improving image quality, for example increasing the number of averages (NEX) however these tend to increase the acquisition time. Out with other reasons for decreasing imaging time (economic / duration of anaesthesia), operator motivation is a factor. Bearing this in mind any recommended epilepsy-specific MRI protocol should not be overly onerous in order to improve compliance. A basic protocol of 6 sequences is recommended [38].

Interpreter factors

Inexperience / lack of training

Failure to recognise significant lesions or over-interpretation of other features. A study in humans found that 61 % of epileptogenic lesions remained undetected following “non-expert” reports of “standard” MRI scans. The failure rate dropped to 9 % using an epilepsy tailored MRI protocol with interpretation by experienced neuro-radiologists [39].

Patient factors

Skull and air interface

In some machines may cause susceptibility artefacts on gradient echo and T1W 3D imaging

Small brain

Slice thickness should be proportional to the brain volume to achieve images with diagnostic quality i.e. animals with smaller brain volume require thinner slices.

Brain conformation

Changes in skull shape, in particular brachycephaly have resulted in changes in brain conformation [40].

General anaesthetic

Increased time under general anaesthesia may increase risk to patient.

Economic factors

Time

Increased time of scanning increases cost and risks of anaesthesia. It is important to consider the balance between time of acquisition and image quality in an animal under general anaesthesia.

Machine costs (purchase of hardware, software, housing and maintenance)

Imaging with higher magnetic field-strength allows for superior images in a shorter imaging time but at a greater cost.

Relevance

Identification and localisation of epileptic lesion is vital in humans with drug-resistant epilepsy, who may be candidates for potentially curative resective epilepsy surgery. Whether this is applicable for dogs with idiopathic epilepsy remains to be seen. Technology that is only capable of detecting large structural pathology such as tumours may be sufficient if it does not alter the management. However acquisition of high quality scans may enable future identification of resectable lesions that are currently hypothesised.