We are proud to announce the first public version of GMAC.

GMAC is a user-level library that implements an Asymmetric Distributed Shared Memory model to be used by CUDA programs. An ADSM model builds a global memory space that allows CPU code to transparently access data hosted in accelerators' (GPUs) memories. Moreover, the coherency of the data is automatically handled by the library. This removes the necessity for manual memory transfers (cudaMemcpy) between the host and GPU memories.

GMAC is being developed by the Operating System Group at the Universitat Politecnica de Catalunya and the IMPACT Research Group at the Univeristy of Illinois under the University of Illinois/NCSA Open Source License.

The project is hosted here. There you can find documentation, code and pre-built Debian packages.

Abstract:

Magnetic resonance (MR) image reconstruction has reached a bottleneck where further speed improvement from the algorithmic perspective is difficult. However, some clinical practices such as real-time surgery monitoring demand faster reconstruction than what is currently available. For such dynamic imaging applications, radial sampling in k-space (i.e. projection acquisition) recently revives due to fast image acquisition, relatively good signal-to-noise ratio, and better resistance to motion artifacts, as compared with the conventional Cartesian scan. Concurrently, using the graphic processing unit (GPU) to improve algorithm performance has become increasingly popular. In this paper, an efficient GPU implementation of the fast Fourier transform (FFT) will first be described in detail, since the FFT is an important part of virtually all MR image reconstruction algorithms.

 Then, we evaluate the speed and image quality for the GPU implementation of two reconstruction algorithms that are suited for projection acquisition. The first algorithm is the look-up table based gridding algorithm. The second one is the filtered backprojection method which is widely used in computed tomography. Our results show that the GPU implementation is up to 100 times faster than a conventional CPU implementation with comparable image quality. 

Abstract:

This paper presents a computational performance analysis of an accelerated medical image registration using Graphics Processing Units (GPUs). In our previous work, a multi-resolution approach using normalized mutual information (NMI) has proven to be useful in medical image registration. In this paper, we propose an acceleration of the NMI procedure using GPU implementation because of the parallel processing capabilities. Registration algorithms were implemented on NVIDIA’s GeForece 9600 GT graphic processor with the Compute Unified Device Architecture (CUDA) programming environment. Experimental results showed that the GPU implementation improves the registration computational performance with a speedup factor of 38. In addition, the maximum speedup can be achieved with diligent data profiling.

Abstract:

In magnetic resonance imaging (MRI), non-Cartesian scan trajectories are advantageous in a wide variety of emerging applications. Advanced reconstruction algorithms that operate directly on non-Cartesian scan data using optimality criteria such as least-squares (LS) can produce significantly better images than conventional algorithms that apply a fast Fourier transform (FFT) after interpolating the scan data onto a Cartesian grid. However, advanced LS reconstructions require significantly more computation than conventional reconstructions based on the FFT. For example, one LS algorithm requires nearly six hours to reconstruct a single three-dimensional image on a modern CPU. Our work demonstrates that this advanced reconstruction can be performed quickly and efficiently on a modern GPU, with the reconstruction of a 643 3D image requiring just three minutes, an acceptable latency for key applications. This paper describes how the reconstruction algorithm leverages the resources of the GeForce 8800 GTX (G80) to achieve over 150 GFLOPS in performance. We find that the combination of tiling the data and storing the data in the G80’s constant memory dramatically reduces the algorithm’s required bandwidth to off-chip memory. The G80’s special functional units provide substantial acceleration for the trigonometric computations in the algorithm’s inner loops. Finally, experiment-driven code transformations increase the reconstruction’s performance by as much as 60% to 80%. 

Abstract:

In minimally invasive image-guided surgical interventions, different imaging modalities, such as magnetic resonance imaging (MRI) or computed tomography (CT), and real-time three-dimensional (3D) ultrasound (US), can provide complementary, multi-spectral image information. Multimodality dynamic image registration is a well-established approach that permits real-time diagnostic information to be enhanced by placing lower-quality real-time images within a high quality anatomical context. For the guidance of cardiac procedures, it would be valuable to register dynamic MRI or CT with intraoperative US. However, in practice, either the high computational cost prohibits such real-time visualization of volumetric multimodal images in a real-world medical environment, or else the resulting image quality is not satisfactory for accurate guidance during the intervention. Modern graphics processing units (GPUs) provide the programmability, parallelism and increased computational precision to begin to address this problem. In this work, we first outline our research on dynamic 3D cardiac MR and US image acquisition, real-time dual-modality registration and US tracking. Then we describe image processing and optimization techniques for 4D (3D + time) cardiac image real-time rendering. We also present our multimodality 4D medical image visualization engine, which directly runs on a GPU in real-time by exploiting the advantages of the graphics hardware. In addition, techniques such as multiple transfer functions for different imaging modalities, dynamic texture binding, advanced texture sampling and multimodality image compositing are employed to facilitate the real-time display and manipulation of the registered dual-modality dynamic 3D MR and US cardiac datasets.