Affine-invariant anisotropic detector for soft tissue tracking in minimally invasive surgery

Reliable feature tracking is important for accurate tissue deformation recovery, 3D anatomical registration and navigation in computer assisted minimally invasive surgical procedures. Despite a wide range of feature detectors developed in the computer vision community, direct application of these approaches to surgical navigation has shown significant difficulties due to the paucity of reliable feature landmarks coupled with free-form tissue deformation and contrastingly different visual appearances of changing surgical scenes. The purpose of this paper is to introduce an affine-invariant feature detector based on anisotropic features to ensure reliable and persistent feature tracking. A novel scale-space representation is proposed for scale adaptation based on the strength of the anisotropic pattern whereas affine adaptation relies on its intrinsic Fourier properties with an efficient spatial implementation based on the second moment matrix. The proposed detector is compared against the current state-of-the-art feature detectors and their respective performance is evaluated with in vivo video sequences recorded from robotic assisted minimally invasive surgical procedures.