Implementation and Validation of an Adaptive Template Registration Method for ¹⁸F-Flutemetamol Imaging Data

Intercept image (top) and slope image (bottom) from linear regression of input images on SUVRs for a neocortical composite region. Images are overlaid on MR T1 template image.

Synthetic images showing typical ¹⁸F-flutemetamol patterns going from most negative (upper left) to most positive case (lower right). Value of x is increased in steps by 0.2 going from left to right, top to bottom.

Flowchart illustrating spatial normalization using the adaptive template approach. Optimization method will change both transformation parameters and adaptive template parameter until maximal similarity between study image and template is achieved. Refined registration of reference region area is performed as a second step.

Average gray matter probabilistic maps for Aβ− and Aβ+ groups superimposed on MNI template obtained using adaptive template registration (A) and SPM (B). Visually, there was no obvious difference between Aβ− and Aβ+, and adaptive template registration and SPM showed similar performance in cerebral cortex whereas adaptive template registration gave sharper cerebellar cortex.

Box plots showing amount of gray matter contained in reference region for Aβ− and Aβ+ groups. Adaptive template registration (A) shows higher amounts of gray matter in cerebellar reference region and less variability than does SPM (B).

Box plots showing amount of brain tissue contained in pons reference region for Aβ− and Aβ+ groups. Adaptive template registration (A) shows higher amounts of brain tissue in pons reference region and less variability than does SPM (B).