The phantom studies indicate that the accuracy of intra-user alignment for the method is less than 0.5° for the rotation and approximately 1mm for shifts. The inter-user variation is greater, with an accuracy of 1° for the rotation and approximately 1.2mm for the shifts.
How much of this variation relates to the unfamiliarity of the users with the program and the alignment procedure is unknown. The majority of the users in the inter-user study were using the program for the first time. Given that the results for the inter-user study are still acceptable however, suggests that the algorithms are robust.
The greater variation in the inter-user alignment indicates that before image registration software could be use in a clinical environment users would require proper training to ensure proficiency. This could take the form of a series of test images of various anatomical sights with known offsets. The user could then attempt alignment of the images. When a suitable level of accuracy on each image and across the whole image set they could be deemed competent.