Similar studies exist in the literature which measure registration error as the difference between transforms [WMC93][NCH+92]. The use of the ILM technique has been validated for PET/MR [EMT+91][NCH+92] and MR/CT [HHH+92] registrations, but a lack of standardisation exists in the quotation of the registration errors. Due to the random nature in which the same homologous points pairs may be selected, the homology errors introduced to the point sets when registering the same two functional and anatomical volumes many times will effect a random distribution in the translation and rotation errors. The mean of the errors is a measure of the systematic error - how well the true registration was at first specified - and therefore does not yield any interesting information. The evaluation of the error in terms of the standard deviation is more valuable because it is independent of the accuracy of the true registration which may be subject to systematic errors in marker location or from distortions.