Virtual palpation of skeletal landmarks with multimodal display interfaces.

Abstract

The 3D location of skeletal landmarks on CT datasets is an important procedure, used in many research and clinical contexts. The standard procedure involves the segmentation of the CT images, the creation of a 3D surface bone model, and the location of the landmarks on this surface. However, the segmentation is time-consuming and requires skilled operators and sophisticated software. The aim of the present study is to evaluate the efficacy of a multimodal display interface to direct volumetric interactive visualization in performing a virtual palpation task. An expert operator used the CT dataset of a patient's thigh region to locate 14 femoral skeletal landmarks. This operation was repeatedly performed using different CT data representation; the accuracy and repeatability were compared to those achievable with the conventional procedure based on the segmented 3D surface. When a multimodal display interface (formed by an orthogonal slice, RXCT and interactive isosurface views) was used to perform the virtual palpation directly on the CT data, the average coordinates of the landmarks did not differ significantly from those located on the 3D surface, and the measurement repeatability was actually better with the multimodal display of the volumetric data than with the 3D surface. Thus, we can conclude that skeletal virtual palpation can be performed directly on the CT dataset, as far as the virtual palpation is performed with a multimodal display interface.