Abstract:

Objective To explore the method and influencing factors of establishing a statistical shape model of the bilateral inner ears, and to investigate the use of the mean model as a standard bilateral inner ear model with spatial information. Methods A total of 37 bilateral inner ear models were segmented through Otsu method and manual segmentation. Those models were registered to generate a statistical shape model. The mean model as a standard model was observed for the spatial information. Results High quality inner ear model was obtained through Otsu method segmentation. After those models were registered, the Statismo software was used to generate inner ear statistical shape model stably. The spatial direction information of the model was representative. Conclusion The statistical shape model of the bilateral inner ear can derive the mean model as the standard model with spatial orientation information.

Key words: Inner ear, Statistical shape model, Standard model, Image registration, Segmentation, Magnetic resonance imaging, Otsu method, Human