Model Fitting Using RANSAC for Surgical Tool Localization in 3-D Ultrasound Images

Abstract

Abstract-Ultrasound guidance is used for many surgical interventions such as biopsy and electrode insertion. We present a method to localize a thin surgical tool such as a biopsy needle or a microelectrode in a 3-D ultrasound image. The proposed method starts with thresholding and model fitting using random sample consensus for robust localization of the axis. Subsequent local optimization refines its position. Two different tool image models are presented: one is simple and fast and the second uses learned a priori information about the tool's voxel intensities and the background. Finally, the tip of the tool is localized by finding an intensity drop along the axis. The simulation study shows that our algorithm can localize the tool at nearly real-time speed, even using a MATLAB implementation, with accuracy better than 1 mm. In an experimental comparison with several alternative localization methods, our method appears to be the fastest and the most robust one. We also show the results on real 3-D ultrasound data from a PVA cryogel phantom, turkey breast, and breast biopsy.

Line filtering for surgical tool localization in 3D ultrasound images

“…RANSAC is used to find an approximate solution, which is subsequently refined in a local optimization step. We have shown experimentally [20] that the model fitting method achieves the lowest failure rate from all tested method. Nevertheless, the obtained failure rate is still too high to allow clinical applicability; it can approach 100% on our most challenging datasets (e.g.…”

“…In our previous work [20] we have introduced another type of tool localization technique, which first finds candidate voxels which are likely to belong to the tool, and then fits the selected voxels by a parametric model of the tool shape. The model fitting approach is much faster than projection-based methods because only a small fraction of the voxels is considered.…”

3D ultrasound-guided robotic steering of a flexible needle via visual servoing

“…Thus, the shape of the needle is usually modeled as a polynomial curve [19], [21], as represented in Fig. 1.…”

“…However the computation of the generalized HT on large 3D volumes remains computationally expensive, even with an implementation on a graphics processing unit (GPU). Alternatively, the random sample consensus (RANSAC) algorithm has been used to detect polynomial curves in 3D ultrasound volumes [19]. Zhao et al [20] improve the stability of this detection using a Kalman filter.…”

3D Steering of a Flexible Needle by Visual Servoing

“…It is therefore possible to extract the shape of the needle by performing some image processing. We will not detail any method to extract the needle but the reader can refer for example to the RANSAC-based approach proposed in [12]. By the use of such extraction algorithm it is therefore possible to obtain the current location of the needle tip point that we denoted N of coordinates w T n = (t nx , t ny , t nz ) and its tip direction orientation w R n with respect to the world frame F w .…”