Background: The Fourier transform (FT) is an image-processing tool that can be used to analyze space or linear systems. We used the FT to analyze the shape of the subpubic line (SL) and the right greater sciatic notch area line (GSNAL) of the pelvis to identify sexing in computed tomographic (CT) images. Methods: 120 bodies (60 male, 60 female) were scanned by multislice CT before autopsy. SL and GSNAL images were reconstructed using an image workstation. The range of the SL is between the bottom ends of the bilateral acetabulum, and the range of the right GSNAL is from the posterior superior iliac spine to the ischial tuberosity. Coordinates (x, y) of the bone outline were extracted from CT images and calculated by fast FT. The Mann-Whitney U test comparing the amplitude of each frequency was used to determine significant differences between the two populations. Multiple regression analyses were performed to find statistically significant amplitudes of each frequency. We verified the accuracy of prediction of the regression equation, and a Cohen's kappa coefficient was measured. Results: Statistically significant differences of amplitude were found in the 2nd, 3rd, and 4th frequencies of the SL and the 1st, 3rd, and 4th frequencies of GSNAL. Adjusted R2 using all amplitudes having a significant difference was 0.772. The accuracy was 0.983. Kappa coefficient was 0.983. Conclusion: We could determine sexing from SL and GSNAL of the pelvis using FT with high accuracy. FT is a useful tool to analyze the curved shape of human bone.
ASJC Scopus subject areas