Mean Geometric Error Research Articles

Canopy detection over roads using mobile lidar data

ABSTRACT A high percentage of forest fires take place around roads. These infrastructures provide an escape route for the population in case of fire. Optimization of forest management in the surroundings of roads is a necessary task in terms of wildfires prevention and mitigation of their effects. Therefore, it is essential to avoid the horizontal continuity of vegetation across roads. A methodology for the measurement of canopy area over roads is developed and based on mobile LiDAR (Light Detection and Ranging) point clouds. The acquisition of LiDAR data is done by Lynx Mobile Mapper System from the University of Vigo. The methodology is automated using LiDAR data processing (M-estimator Sample Consensus and near neighbour algorithms) and image processing techniques (rasterization and binarization). The developed algorithms are tested on a study area, the DP-3606 road (Spain). Results are compared with ground truth data of the canopy projected on the road. The best obtained results present a mean geometric error of 2.82% for 0.25 m resolution and 104.02% for 2 m resolution. Furthermore, the higher the pixel size, the greater the error was obtained with a linear correlation value of 0.99.

Development and validation of a semi-automatic landmark extraction method for mesh morphing

Currently, landmark-based mesh morphing technology is widely used to rapidly obtain meshes with specific geometry, which is suitable to develop parametric human finite element (FE) models. However it takes too much time for landmark extraction to obtain high geometric accuracy. The purpose of this study is to develop and validate a semi-automatic landmark extraction method to reduce the time of manual selection of landmarks without sacrificing the accuracy of identifying landmarks in the process of mesh morphing. A few contour edge landmarks were extracted manually. Mathematical landmarks and pseudo-landmarks were extracted automatically by user-defined algorithm. The radial basis function (RBF) was used to morph the baseline FE model into the target geometry based on these landmarks. The cervical vertebra (C5), rib (R7) and femur were selected as the target geometries to verify the effectiveness of the method. The maximum mean geometric error of the three types of target geometries was less than 1 mm. The mesh quality of the morphed FE model was similar to that of the baseline FE model. Compared to the traditional manual method, 2/3 to 3/4 of the time for landmark extraction was saved by the semi-automatic method.

A 4D biomechanical lung phantom for joint segmentation/registration evaluation

At present, there exists few openly available methods for evaluation of simultaneous segmentation and registration algorithms. These methods allow for a combination of both techniques to track the tumor in complex settings such as adaptive radiotherapy. We have produced a quality assurance platform for evaluating this specific subset of algorithms using a preserved porcine lung in such that it is multi-modality compatible: positron emission tomography (PET), computer tomography (CT) and magnetic resonance imaging (MRI). A computer controlled respirator was constructed to pneumatically manipulate the lungs in order to replicate human breathing traces. A registration ground truth was provided using an in-house bifurcation tracking pipeline. Segmentation ground truth was provided by synthetic multi-compartment lesions to simulate biologically active tumor, background tissue and a necrotic core. The bifurcation tracking pipeline results were compared to digital deformations and used to evaluate three registration algorithms, Diffeomorphic demons, fast-symmetric forces demons and MiMVista’s deformable registration tool. Three segmentation algorithms the Chan Vese level sets method, a Hybrid technique and the multi-valued level sets algorithm. The respirator was able to replicate three seperate breathing traces with a mean accuracy of 2–2.2%. Bifurcation tracking error was found to be sub-voxel when using human CT data for displacements up to 6.5 cm and approximately 1.5 voxel widths for displacements up to 3.5 cm for the porcine lungs. For the fast-symmetric, diffeomorphic and MiMvista registration algorithms, mean geometric errors were found to be , and voxels widths respectively using the vector field differences and , and voxel widths using the bifurcation tracking pipeline. The proposed phantom was found sufficient for accurate evaluation of registration and segmentation algorithms. The use of automatically generated anatomical landmarks proposed can eliminate the time and potential innacuracy of manual landmark selection using expert observers.

SU-E-T-140: Dynamic Wave Arc Trajectory Verification Using KV X-Ray Fluoroscopy

Purpose: Purpose: This study investigates the geometric accuracy of simultaneous Gantry/Ring rotation during Dynamic Wave Arc (DWA) delivery. Methods: The Vero SBRT system consists of a 6MV LINAC mounted on an O-ring gantry that can rotate around the vertical axis (±60°), similar to couch rotation on C-arm gantries. To provide CBCT and fluoroscopy imaging functionalities, two orthogonal kV imaging units are attached to the O-ring at −45°/+45° from the beam axis.Dynamic Wave Arc maximizes Vero's motion capabilities by employing synchronized gantry and ring motion on a complex non-coplanar trajectory in combination with aperture based optimized MLC segments.Four wave arc trajectories (T1-4) were delivered using a cubic phantom with a configuration of five lead beads. O-ring gantry position information was retrieved through continuous dual-source kV X-ray image acquisition during DWA. An in-house algorithm read in the image set, extracted the projected marker positions and determined the angulation through reconstruction of the beam source position. The geometric error was quantified as the distance between the independently detected positions from kV-images and reference trajectory derived from the treatment plan in the Ring-Gantry space. Results: The average displacement between the 3D gantry/ring positions reconstructed from the fluoroscopy images and the reference trajectory was 0.346 mm (SD 0,171) for T1. A mean offset of 0.348 mm (SD 0,182) and 0.357 mm (SD 0.194) was observed for trajectory T2(2segmens) and T3(4segments), respectively. The saw shape T4 presented a mean geometric error of 0.363 (SD 0.156). The overall systematic error of 0.350 was caused by the difference between planned reference trajectory created by linear interpolation between CP, and the machine delivery following a spline curve. Conclusion: An independent geometric QA approach has been developed for DWA delivery verification, successfully applied on diverse trajectories and disclosed that the Vero system is capable of accurately follow complex trajectories.

Linear feature extraction using adaptive least‐squares template matching and a scalable slope edge model

This paper presents a linear feature extraction method. Least squares template matching (LSTM) is adopted as the computational tool to fit the linear features with a scalable slope edge (SSE) model, which is based on an explicit function to define the blurred edge profile. In the SSE model, the magnitude of the grey gradient and the edge scale can be described by three parameters; additionally, the edge position can be obtained strictly by the ‘zero crossing’ location of the profile model. In our method the edge templates are locally and adaptively generated by estimating the three parameters via fitting the image patches with the model, accordingly the linear feature can be positioned with high accuracy by using LSTM. We derived the computational models to rectify straight line and spline curve features and tested those algorithms using the synthetic and real remotely sensed images. The experiments using synthetic images show that the method can position the linear features with the mean geometric error of pixel location of less than one pixel in certain noise levels. Examples of semiautomatic extraction of buildings and linear objects from real imagery are also given and demonstrate the potential of the method.

European Society of Biomechanics S.M. Perren Award 2008: Using temporal trends of 3D bone micro-architecture to predict bone quality

In longitudinal studies, three-dimensional (3D) bone images are acquired at sequential time points essentially resulting in four-dimensional (4D) data for an individual. Based on the 4D data, we propose to calculate temporal trends and project these trends to estimate future bone architecture. Multiple consecutive deformation fields, calculated with Demons deformable image registration algorithm, were extrapolated on a voxel-by-voxel basis. Test data were from in vivo micro-computed tomography ( μ CT ) scans of the proximal tibia of Wistar rats that were either ovariectomized (OVX; N = 5 ) or sham operated (SHAM; N = 6 ). Measurements performed at baseline, 4 and 8 weeks were the basis to predict the 12 week data. Predicted and actual 12 week data were compared using qualitative (3D rendering) and quantitative (geometry, morphology and micro-finite element, μ FE ) methods. The results indicated a voxel-based linear extrapolation scheme yielded mean geometric errors that were smaller than the voxel size of 15 μ m . Key morphological parameters that were estimated included bone volume ratio (BV/TV; mean error 0.4%, maximum error 9%), trabecular thickness (Tb.Th; - 1.1 % , 11%), connectivity density (Conn.D; 9.0%, 18.5%) and the apparent Young's modulus ( E 1 ; 6.0%, 32%). These data demonstrated a promising and novel approach for quantitatively capturing in vivo bone dynamics at the local trabecular level. The method does not require an a priori understanding of the diseases state, and can provide information about the trends of the bone remodeling process that may be used for better monitoring and treatment of diseases such as osteoporosis.

USING TEMPORAL TRENDS OF 3D BONE MICRO-ARCHITECTURE TO PREDICT BONE QUALITY

In vivo measurements of trabecular bone architecture provide insight into the process of bone remodeling for quantification of important components of bone quality, including morphology and mechanical integrity. In longitudinal studies, three-dimensional (3D) bone images are acquired at sequential time points essentially resulting in four-dimensional (4D) data for an individual. Based on the 4D data, we propose to calculate temporal trends and project these trends to estimate future bone architecture. The method consists of calculating deformation fields between subsequent time points utilizing Demons deformable image registration. Multiple consecutive deformation fields were extrapolated on a voxel-by-voxel basis, and five different extrapolation strategies were tested. Test data were from in vivo μCT scans of the proximal tibia of Wistar rats that were either ovariectomized (OVX; N=5) or sham operated (SHAM; N=6), and measurements were performed at baseline, 4, 8 and 12 weeks. The first three measurements for each animal were the basis to predict the 12 week data, and validation included a qualitative (3D rendering) and quantitative (geometry, morphology and micro-FE) comparison with actual 12 week data (see Fig. 1). The results demonstrated that the temporal trends could be used to predict 3D bone structure, and indicated a voxel-based linear extrapolation scheme yielded mean geometric errors that were smaller than the voxel size of 15μm. Key morphological parameters that were estimated included bone volume ratio (BV/TV; mean error 0.4%, maximum error 9%), trabecular thickness (Tb.Th; -1.1%, 11%), connectivity density (Conn.D; 9.0%, 18.5%) and the apparent Young's modulus (E1; 6.0%, 32%). These data demonstrated a promising and novel approach for quantitatively capturing in vivo bone dynamics at the local trabecular level. The method does not require an a priori understanding of the diseases state, and can provide information about the trends of the bone remodeling process that may be used for better monitoring and treatment of diseases such as osteoporosis.

Fast, Simple, Feature Preserving and Memory Efficient Simplification of Triangle Meshes

We propose a new iterative algorithm for the automatic geometric simplification of a triangle mesh based on edge collapse transformations. The way how geometric deviation resulted from an edge collapse transformation is measured plays a basic role in inducing a priority ordering on the set of edge collapse transformations to control the simplification process. We introduce a new technique to measure this deviation based only on current simplified mesh and the new concept of accumulating the cost of collapse, which are easy to implement and involve simple computations making the algorithm computationally efficient. No geometric history is needed to be retained, so the resulting algorithm is also memory efficient. Results and numerical comparisons show that our algorithm generates simplified meshes of good visual fidelity, which compares well with those by other methods in terms of maximum and mean geometric error and it preserves the visually important features of the original mesh.