Minho Kim

Evaluation of similarity measures for use in the intensity-based rigid 2D-3D registration for patient positioning in radiotherapy.

PURPOSE Rigid 2D-3D registration is an alternative to 3D-3D registration for cases where largely bony anatomy can be used for patient positioning in external beam radiation therapy. In this article, the authors evaluated seven similarity measures for use in the intensity-based rigid 2D-3D registration using a variation in Skerls similarity measure evaluation protocol. METHODS The seven similarity measures are partitioned intensity uniformity, normalized mutual information (NMI), normalized cross correlation (NCC), entropy of the difference image, pattern intensity (PI), gradient correlation (GC), and gradient difference (GD). In contrast to traditional evaluation methods that rely on visual inspection or registration outcomes, the similarity measure evaluation protocol probes the transform parameter space and computes a number of similarity measure properties, which is objective and optimization method independent. The variation in protocol offers an improved property in the quantification of the capture range. The authors used this protocol to investigate the effects of the downsampling ratio, the region of interest, and the method of the digitally reconstructed radiograph (DRR) calculation [i.e., the incremental ray-tracing method implemented on a central processing unit (CPU) or the 3D texture rendering method implemented on a graphics processing unit (GPU)] on the performance of the similarity measures. The studies were carried out using both the kilovoltage (kV) and the megavoltage (MV) images of an anthropomorphic cranial phantom and the MV images of a head-and-neck cancer patient. RESULTS Both the phantom and the patient studies showed the 2D-3D registration using the GPU-based DRR calculation yielded better robustness, while providing similar accuracy compared to the CPU-based calculation. The phantom study using kV imaging suggested that NCC has the best accuracy and robustness, but its slow function value change near the global maximum requires a stricter termination condition for an optimization method. The phantom study using MV imaging indicated that PI, GD, and GC have the best accuracy, while NCC and NMI have the best robustness. The clinical study using MV imaging showed that NCC and NMI have the best robustness. CONCLUSIONS The authors evaluated the performance of seven similarity measures for use in 2D-3D image registration using the variation in Skerls similarity measure evaluation protocol. The generalized methodology can be used to select the best similarity measures, determine the optimal or near optimal choice of parameter, and choose the appropriate registration strategy for the end user in his specific registration applications in medical imaging.

Box Spline Reconstruction On The Face-Centered Cubic Lattice

We introduce and analyze an efficient reconstruction algorithm for FCC-sampled data. The reconstruction is based on the 6-direction box spline that is naturally associated with the FCC lattice and shares the continuity and approximation order of the triquadratic B-spline. We observe less aliasing for generic level sets and derive special techniques to attain the higher evaluation efficiency promised by the lower degree and smaller stencil-size of the C1 6-direction box spline over the triquadratic B-spline.

Fast and stable evaluation of box-splines via the BB-form

To repeatedly evaluate linear combinations of box-splines in a fast and stable way, in particular along knot planes, the box-spline is converted to and tabulated as piecewise polynomial in BB-form (Bernstein–Bézier-form). We show that the BB-coefficients can be derived and stored as integers plus a rational scale factor and derive a hash table for efficiently accessing the polynomial pieces. This pre-processing, the resulting evaluation algorithm and use in a widely-used ray-tracing package are illustrated for splines based on two trivariate box-splines: the seven-directional box-spline on the Cartesian lattice and the six-directional box-spline on the face-centered cubic lattice.

Simulation for Training With the Autosuture TM Endo StitchTM Device

The rapid development and deployment of novel laparoscopic instruments present the surgical educator and trainee with a significant challenge. Several useful instruments have been particularly difficult to teach the novice. We have developed a platform that allows the combination of the actual instrument handle with a virtual re-creation of the instrument tip. We chose the AutosutureTM Endo StitchTM device as the prototypical instrument because it satisfies our subjective experience of “useful, but hard to teach.” A software package was developed to support the re-creation of the needle and suture that accompany the device. The apparatus has haptic capabilities and collision detection so that the needle driver is “aware” of suture and instrument contact. The developed virtual environment allows re-creation of the necessary motion to simulate the instrument, the trainee can use the actual instrument handle, and the system can be altered to accommodate other instruments.

Developing a multimedia environment for customized teaching of an adrenalectomy

We have developed a computer based simulation process which allows a surgical expert to create a customized operative environment. This virtual environment, the Toolkit for Illustration of Procedures in Surgery (3D TIPS), is deployed on a low-cost computer system and requires minimal training for the programmer. The learner can be engaged in training immediately and the educator can modify the system and annotate the procedure to highlight specific points using video clips, operative images, and the like. A laparoscopic adrenalectomy is presented as a proof of concept in the accompanying article.

Realtime loop subdivision on the GPU

Recently, [1] showed that, in principle, all major features of subdivision algorithms can be realized in the framework of highly parallel stream processing. The authors tested the approach by implementing Catmull-Clark subdivision, with semi-smooth creases and global boundaries, in programmable graphics hardware, at near-realtime speed. Here, we report on adapting the approach to Loop subdivision, discussing challenges we encountered.