Neuro-fuzzy Stereo Camera Calibration Architectures Used in Interactions with Medical Volumes

Computer assisted diagnosis/surgery systems need improved solutions for visualization and interactions, as the virtual probe tools. This work presents such an interaction tool and its applications in medical imaging. The novelty is represented by the use of neuro-fuzzy systems in 3D virtual probe positioning; two such configurations are presented and compared to the crisp positioning system. We prove that the neuro-fuzzy solutions outperform the crisp solutions in terms of computational complexity and positioning accuracy. The proposed framework integrates several functionalities: positioning of arbitrary cutting planes; 3D editing, measurements and annotations; stereo visualization. Being a flexible architecture, other functionalities could be integrated in the future.