Reconstruction of 3-D volumetric data from C-arm CT projections is a computationally demanding task. For
interventional image reconstruction, hardware optimization is mandatory. Manufacturers of medical equipment
use a variety of high-performance computing (HPC) platforms, like FPGAs, graphics cards, or multi-core CPUs.
A problem of this diversity is that many different frameworks and (vendor-specific) programming languages are
used. Furthermore, it is costly to switch the platform, since the code has to be re-written, verified, and optimized.
OpenCL, a relatively new industry standard for HPC, promises to enable portable code. Its key idea is to
abstract hardware in a way that allows an efficient mapping onto real CPUs, GPUs, and other hardware. The
code is compiled for the actual target by the device driver.
In this work we investigated the suitability of OpenCL as a tool to write portable code that runs efficiently
across different hardware. The problems chosen are back- and forward-projection, the most time-consuming
parts of (iterative) reconstruction. We present results on three platforms, a multi-core CPU system and two
GPUs, and compare them against manually optimized native implementations.
We found that OpenCL allows to share a common framework in one language across platforms. However,
considering differences in the underlying architecture, a hardware-oblivious implementation cannot be expected
to deliver maximal performance. By optimizing the OpenCL code for the specific hardware we reached over 90%
of native performance for both problems, back- and forward-projection, on all platforms.
Phase-contrast imaging approaches suffer from a severe problem which is known in Magnetic Resonance Imaging
(MRI) and Synthetic Aperture Radar (SAR) as phase-wrapping. This work focuses on an unwrapping solution for
the grating based phase-contrast interferometer with X-rays. The approach delivers three types of information
about the x-rayed object - the absorption, differential phase-contrast and dark-field information whereas the
observed differential phase values are physically limited to the interval (-π, π]; values higher or lower than the
interval borders are mapped (wrapped) back into it. In contrast to existing phase-unwrapping algorithms for MRI
and SAR the presented algorithm uses the absorption image as additional information to identify and correct
phase-wrapped values. The idea of the unwrapping algorithm is based on the observation that at locations with
phase-wrapped values the contrast in the absorption image is high and the behavior of the gradient is similar
to the real (unwrapped) phase values. This can be expressed as a cost function which has to be minimized by
an integer optimizer. Applied on simulated and real datasets showed that 95.6% of phase-wraps were correctly
unwrapped. Based on the results we conclude that it is possible to use the absorption information in order to
identify and correct phase-wrapped values.
Time-resolved 3-D imaging of the heart is a major research topic in the medical imaging community. Recent advances in the interventional cardiac 3-D imaging from rotational angiography (C-arm CT) are now also making 4-D imaging feasible during procedures in the catheter laboratory. State-of-the-art reconstruction algorithms try to estimate the cardiac motion and utilize the motion field to enhance the reconstruction of a stable cardiac phase (diastole). The available data offers a handful of opportunities during interventional procedures, e.g. the ECG-synchronized dynamic roadmapping or the computation and analysis of functional parameters. In this paper we will demonstrate that the motion vector field (MVF) that is output by motion compensated image reconstruction algorithms is in general not directly usable for animation and motion analysis. Dependent on the algorithm different defects are investigated. A primary issue is that the MVF needs to be inverted, i.e. the wrong direction of motion is provided. A second major issue is the non-periodicity of cardiac motion. In algorithms which compute a non-periodic motion field from a single rotation the in depth motion information along viewing direction is missing, since this cannot be measured in the projections. As a result, while the MVF improves reconstruction quality, it is insufficient for motion animation and analysis. We propose an algorithm to solve both problems, i.e. inversion and missing in-depth information in a unified framework. A periodic version of the MVF is approximated. The task is formulated as a linear optimization problem where a parametric smooth motion model based on B-splines is estimated from the MVF. It is shown that the problem can be solved using a sparse QR factorization within a clinical feasible time of less than one minute. In a phantom experiment using the publicly available CAVAREV platform, the average quality of a non-periodic animation could be increased by 39% by applying the proposed periodization and inversion method.
Image guidance during cardiac interventional procedures (IP) using cardiac C-arm CT systems is desirable for
many procedures. Applying the concept of retrospective electrocardiogram gating (ECG) to the acquisition of
multiple, ECG-triggered rotational acquisitions using a C-arm system allows the 3D+t reconstruction of the
heart. The process of retrospective gating is a crucial component of 3-D reconstruction. The gold-standard
in gating is still ECG based. However, the ECG signal does not directly reflect the mechanical situation of
the heart. Therefore an alternative gating method, based on the acquired projection data is required. Our
goal is to provide an image-based gating (IBG) method without ECG such that already acquired projection
data from a multi-sweep acquisition can still be used for reconstruction. We formulate the gating problem as a
shortest-path optimization problem. All acquired projection images build a directed graph and the path costs are
defined by projection image similarities that are based on image metrics to measure the heart phase similarity.
The optimization is additionally regularized to prefer solutions where the path segment of consecutive selected
projections acquired along a particular forward or backward C-arm sweep is short. This regularization depends
on an estimated average heart rate that is also estimated using an image-based method. First promising results
using in-vivo data are presented and compared to standard ECG gating. We conclude that the presented IBG
method provides a reliable gating.
Proc. SPIE. 6510, Medical Imaging 2007: Physics of Medical Imaging
KEYWORDS: Signal to noise ratio, 3D image reconstruction, Heart, Image registration, Data acquisition, Temporal resolution, Electrocardiography, Motion measurement, Motion estimation, 3D image processing
The combination of real-time fluoroscopy and 3D cardiac imaging on the same C-arm system is a promising technique
that might improve therapy planning, guiding, and monitoring in the interventional suite. In principal, to reconstruct a 3D
image of the beating heart at a particular cardiac phase, a complete set of X-ray projection data representing that phase is
required. One approximate approach is the retrospectively ECG-gated FDK reconstruction (RG-FDK). From the acquired
data set of Ns multiple C-arm sweeps, those projection images which are acquired closest in time to the desired cardiac
phase are retrospectively selected. However, this approach uses only 1/
of the obtained data. Our goal is to utilize data from
other cardiac phases as well. In order to minimize blurring and motion artifacts, cardiac motion has to be compensated for,
which can be achieved using a temporally dependent spatial 3D warping of the filtered-backprojections. In this work we
investigate the computation of the 4D heart motion based on prior reconstructions of several cardiac phases using RG-FDK.
A 4D motion estimation framework is presented using standard fast non-rigid registration. A smooth 4D motion vector
field (MVF) represents the relative deformation compared to a reference cardiac phase. A 4D deformation regridding by
adaptive supersampling allows selecting any reference phase independently of the set of phases used in the RG-FDK for
a motion corrected reconstruction. Initial promising results from in vivo experiments are shown. The subjects individual
4D cardiac MVF could be computed from only three RG-FDK image volumes. In addition, all acquired projection data
were motion corrected and subsequently used for image reconstruction to improve the signal-to-noise ratio compared to
In this paper, we propose a multi-modal non-rigid 2D-3D registration technique. This method allows
a non-rigid alignment of a patient pre-operatively computed tomography (CT)
to few intra operatively acquired fluoroscopic X-ray images obtained with a C-arm
system. This multi-modal approach is especially focused on the 3D
alignment of high contrast reconstructed volumes with intra-interventional low
contrast X-ray images in order to make use of up-to-date information for surgical guidance and other interventions. The key issue of non-rigid 2D-3D registration is how to define the distance
measure between high contrast 3D data and low contrast 2D projections.
In this work, we use algebraic reconstruction theory to handle
this problem. We modify the Euler-Lagrange equation by
introducing a new 3D force. This external force term is computed
from the residual of the algebraic reconstruction procedures. In the
multi-modal case we replace the residual between the digitally reconstructed
radiographs (DRR) and observed X-ray images with a statistical based distance measure. We integrate the algebraic reconstruction technique
into a variational registration framework, so that the 3D
displacement field is driven to minimize the reconstruction
distance between the volumetric data and its 2D projections using mutual
information (MI). The benefits of this 2D-3D registration approach are its
scalability in the number of used X-ray reference images and the proposed distance that can handle low contrast fluoroscopies as well.
Experimental results are presented on both artificial phantom and
3D C-arm CT images.