Interactive segmentation based on the live wire for 3D CT chest image analysis

Abstract

Object The definition of regions of interest (ROIs) such as suspect cancer nodules or lymph nodes in 3D MDCT chest images is often difficult because of the complexity of the phenomena that give rise to them. Manual slice tracing has been used commonly for such problems, but it is extremely time consuming, subject to operator biases, and does not enable reproducible results. Proposed automated 3D imagesegmentation methods are generally application dependent, and even the most robust methods have difficulty in defining complex ROIs. Materials and methods The semi-automatic interactive paradigm known as live wire has been proposed by researchers, whereby the human operator interactively defines an ROI's boundary, guided by an active automated method. We propose 2D and 3D live-wire methods that improve upon previously proposed techniques. The 2D method gives improved robustness and incorporates a search region to improve computational efficiency. The 3D method requires the operator to only consider a few 2D slices, with an automated procedure performing the bulk of the analysis. Results For tests run with five human operators on both 2D and 3D ROIs in 3D MDCT chest images, the reproducibility was >97% and the ground-truth correspondence was at least 97%. The 2D live-wire approach was ≥14 times faster than manual slice tracing, while the 3D method was ≥28 times faster than manual slice tracing. Finally, we describe a computer-based tool and its application to 3D MDCT-based planning and follow-on live guidance of bronchoscopy. Conclusion The live-wire methods are efficient, reliable, easy to use, and applicable to a wide range of circumstances.

Automatic segmentation and centroid detection of skin sensors for lung interventions

“…10,11 The distance between the centroids, from the automatic and semi-automatic approaches, is calculated as an error measurement. Table 1 summarizes the performance of the automatic fiducial detection approach using the error measurement.…”

Live-wire-based segmentation of (3D) anatomical structures for image-guided lung interventions

“…6,7 In the SSLW, the termination criteria of this approach rely on: 1) number of voxels of the boundary candidate in the previous processed sectional image; 2) variation of the voxel intensity feature of enclosed object areas between two adjacent slices; 3) boundary conditions, such as shape; 4) difference of incremental total costs, calculated based on live wire paradigm, of a defined boundary candidate compared with that derived from the previous sectional image. Due to the complicated phenomena that might be involved in the processed ROI, the stopping criteria might not be able to terminate the process properly.…”

“…[3][4][5][6][7] After initial boundary information is given by a user, the activecontour-based method iteratively minimizes a corresponding cost function so as to define a desired contour upon converge of the algorithm. The performance not only relies on the design of the iterative optimization algorithm, but also depends on the initial boundary data, which is usually defined by a user or from a priori knowledge.…”

“…[5][6][7] It has been well recognized as an efficient tool that arguably provides the best combination of manual intervention and automated analysis for general medical image segmentation. Though the live wire method is able to deal with almost all kinds of 2D anatomical structures, its extension to 3D segmentation still requires considerable efforts.…”

3D CT-Video Fusion for Image-Guided Bronchoscopy

“…For central-axes extraction, we have drawn upon techniques based on 3D thinning, B-spline analysis, and differential geometry [37,39,42]. For ROI definition, either direct manual slice tracing or more robust semi-automatic live-wire analysis have been used to identify region borders [32,43,44]. Surface data is generated using the wellknown Marching Cubes algorithm built into the VTK package [30].…”