Abstract
As part of the ITK v4 project efforts, we have developed ITK filters for physics-based non-rigid registration (PBNRR), which satisfies the following requirements: account for tissue properties in the registration, improve accuracy compared to rigid registration, and reduce execution time using GPU and multi-core accelerators. The implementation has three main components: (1) Feature Point Selection, (2) Block Matching (mapped to both multi-core and GPU processors), and (3) a Robust Finite Element Solver. The use of multi-core and GPU accelerators in ITK v4 provides substantial performance improvements. For example, for the non-rigid registration of brain MRIs, the performance of the block matching filter on average is about 10 times faster when 12 hyperthreaded multi-cores are used and about 83 times faster when the NVIDIA Tesla GPU is used in Dell Workstation.
Highlights
Image-guided Neurosurgery (IGNS) is a system that can track in real-time the movement of the surgical tools in the patient space and report the movement to surgeons via the trajectory in the image space based on the established transform between the patient space and the image space
PHYSICS-BASED NON-RIGID REGISTRATION METHOD Given preoperative MRI and intraoperative MRI, we aim to find the deformation between them and deform the preoperative MRI according to the deformation
This mesh generation includes two steps: first produce a coarse Body-Centered Cubic (BCC) mesh based on the segmented mask image, and compress the surface of the coarse BCC mesh to the boundary of the mask image
Summary
Image-guided Neurosurgery (IGNS) is a system that can track in real-time the movement of the surgical tools in the patient space and report the movement to surgeons via the trajectory in the image space based on the established transform between the patient space and the image space. The preoperatively acquired navigation image must be deformed To achieve this end, a nonrigid registration can be used to align the preoperative image with the intra-operative modalities, such as Laser Range Scanning image, intra-operative ultrasound (iUS), or Magnetic Resonance Imaging (iMRI). Wang et al (2010) presented a block matching based non-rigid registration method, in which the block matching was adapted and implemented on GPU. The resulting displacement vector field was smoothed by Gaussian and served to regularize the matching using normalized cross correlation. This method was applied to 4D lung CT images registration and planning CT and Daily Cone Bean CT registration. The landmark-based evaluation for both experiments showed the proposed GPU-based implementation achieved comparable registration accuracy, and compared to the CPU-based AtamaiWarp program, the GPU-based implementation is about 25 times faster
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
