A method to automatically segment the internal ear structures in CT images

Abstract

Hereby it is presents a novel methodology to segment the internal structures of the human ear based on image registration. The first step of the methodology concerns the registration of the volumetric image to be segmented with a reference volumetric image previously defined and manually segmented. Then, the geometric transformation found in the registration step is applied to the image to be segmented and consequently, the segmentation of this image is achieved. The methodology was applied on a set of images obtained by X-ray computed tomography revealing very promising results.