If the results from fMRI experiments are to be reliable, comparable and efficiently obtained, it is necessary to standardise the way that such experiments are carried out. To this end, a standard fMRI protocol was written. The aim of this was to provide an experimental outline that contained the basic procedures that should be carried out, but could be adapted to fit the study in question. The standard fMRI protocol is given in Appendix A.
Correct configuration of the scanner is given purely as a reference, as it is assumed that the operator will be familiar with the operation of the 3.0 Tesla scanner [10]. The standard cylindrical phantom used on the 3.0 T scanner contains structure that can enable its orientation to be determined from the images. A standard orientation of the phantom is given to avoid confusion when looking at the images at a later date, and for this reason a phantom reference set of images is always stored with the data from that particular experiment. The stability of the scanner is tested each day that fMRI is carried out, by imaging a single slice of a phantom for about 5 minutes. A repetition rate of 10 seconds ensures that the effect of stimulated echoes does not interfere with the result. The stability can be calculated on a pixel by pixel basis by dividing the standard deviation of the pixel time course by its mean. The stability is regarded as satisfactory if this ratio is less than 0.5% throughout the phantom.
Anatomical reference scans are acquired with each data set, so that the precise location of the activation can be determined. If thick slices (> 1 cm) of the brain are to be imaged in the fMRI experiment, then it is better to acquire thinner slices for the reference scans, since the reduction in through slice susceptibility dephasing that this brings will improve the image quality considerably. A method for acquiring these thin slices such that they directly correspond to the thick slices is described in the protocol.
No details of how the stimulus paradigm should be designed are included in the standard protocol. This is because each paradigm may need to be developed depending on many factors such as the size of signal change observed on activation, how the stimulus is presented, how the subject is to respond, and the length of time a subject is able to stay in the scanner. For an initial experiment, a 16 cycle paradigm, with each cycle consisting of 16 seconds of rest and 16 seconds of activity has been shown to be reliable in a number of situations. The results presented later in this thesis, however, show that functional imaging of a single cognitive event is possible, and paradigms that exploit this may yield extra results over those obtained in an epoch based paradigm.