Surgical tools localization in 3D ultrasound images

摘要: This thesis deals with automatic localization of thin surgical tools such as needles or electrodes in 3D ultrasound images. The precise and reliable is important for medical interventions needle biopsy electrode insertion into tissue. reader introduced to basics (US) imaging. state the art methods are reviewed work. Many Hough transform (HT) Parallel Integral Projection (PIP) based on projections. As existing PIP implementations relatively slow, we suggest an acceleration by using a multiresolution approach. We propose use model fitting approach which uses randomized sample consensus (RANSAC) local optimization. It fast method suitable real-time it robust respect presence other high-intensity structures background. two new shape appearance models tool US Tool can be improved exploiting its tubularity. line filtering incorporated scheme. robustness algorithm at expense additional time pre-processing. demonstrated implementation scanner Ultrasonix RP. All proposed were tested simulated data, phantom data (a replacement tissue) real tissue breast needle. had comparable accuracy lower number failures than projection methods.