Dice similarity coefficient calculation

We calculated the dice similarity coefficient (DSC) using the following formula to verify the accuracy of 3D nonlinear registration to the head and neck template images³⁶.

A expresses the mask region of the image which is transformed from the mask image created by manually tracing the masticatory muscle form in the TE20 image by the transformation parameter for spatial normalization, while B expresses the mask region of the mask image created by manually tracing the masticatory muscle form in the head and neck template image. N(A) is the number of voxels included in A, N(B) is the number of voxels included in B, and N(A ∩ B) is the number of voxels common to A and B. The manufacturing of mask images, image calculation to detect common areas, and voxel number count was executed using Fiji. DSC(A,B) was calculated for all 40 images, and statistical analysis was performed using the Friedman test on whether there was a significant difference among the total of four conditions (before and after clenching under two conditions of “intact dentition” and “simulated left molar loss”).