Combining MRI with PET for partial volume correction improves image-derived input functions in mice.

Accurate kinetic modelling using dynamic PET requires knowledge of the tracer concentration in plasma, known as the arterial input function (AIF). AIFs are usually determined by invasive blood sampling, but this is prohibitive in murine studies due to low total blood volumes. As a result of the low spatial resolution of PET, image-derived input functions (IDIFs) must be extracted from left ventricular blood pool (LVBP) ROIs of the mouse heart. This is challenging because of partial volume and spillover effects between the LVBP and myocardium, contaminating IDIFs with tissue signal. We have applied the geometric transfer matrix (GTM) method of partial volume correction (PVC) to 12 mice injected with 18F-FDG affected by a Myocardial Infarction (MI), of which 6 were treated with a drug which reduced infarction size [1]. We utilised high resolution MRI to assist in segmenting mouse hearts into 5 classes: LVBP, infarcted myocardium, healthy myocardium, lungs/body and background. The signal contribution from these 5 classes was convolved with the point spread function (PSF) of the Cambridge split magnet PET scanner and a non-linear fit was performed on the 5 measured signal components. The corrected IDIF was taken as the fitted LVBP component. It was found that the GTM PVC method could recover an IDIF with less contamination from spillover than an IDIF extracted from PET data alone. More realistic values of Ki were achieved using GTM IDIFs, which were shown to be significantly different (p<0.05) between the treated and untreated groups.

Quantitative BOLD: the effect of diffusion.

PURPOSE: To make the quantitative blood oxygenation level-dependent (qBOLD) method more suitable for clinical application by accounting for proton diffusion and reducing acquisition times. MATERIALS AND METHODS: Monte Carlo methods are used to simulate the signal from diffusing protons in the presence of a blood vessel network. A diffusive qBOLD model was then constructed using a lookup table of the results. Acquisition times are reduced by parallel imaging and by employing an integrated fieldmapping method, rather than running an additional sequence. RESULTS: The addition of diffusion to the model is shown to have a significant impact on predicted signal formation. This is found to affect all fitted parameters when the model is applied to real data. Parallel imaging and integrated fieldmapping allowed the GESSE (gradient echo sampling of a spin echo) acquisition to be made in less than 10 minutes while maintaining high signal-to-noise ratio (SNR). CONCLUSION: By incorporating integrated field mapping and parallel imaging techniques, GESSE data were acquired within clinically acceptable acquisition times. These data fit closely to the diffusive qBOLD model, providing more realistic and robust measurements of T(2) and blood oxygenation than the static model.

Application of a probabilistic double-fibre structure model to diffusion-weighted MR images of the human brain.

A Markov chain Monte Carlo (MCMC) algorithm has been reported which is capable of determining the probabilistic orientation of two-fibre populations from high angular resolution diffusion-weighted data (HARDI). We present and critically discuss the application of this algorithm to in vivo human datasets acquired in clinically realistic times. We show that by appropriate model selection areas of multiple fibre populations can be identified that correspond with those predicted from known anatomy. Quantitative maps of fibre orientation probability are derived and shown for one- and two-fibre models of neural architecture. Fibre crossings in the pons, the internal capsule and the corona radiata are shown. In addition, we demonstrate that the relative proportion of anisotropic signal may be a more appropriate measure of anisotropy than summary measures derived from the tensor model such as fractional anisotropy in areas with multi-fibre populations.

Monte Carlo simulation and scatter correction of the GE advance PET scanner with SimSET and Geant4.

For Monte Carlo simulations to be used as an alternative solution to perform scatter correction, accurate modelling of the scanner as well as speed is paramount. General-purpose Monte Carlo packages (Geant4, EGS, MCNP) allow a detailed description of the scanner but are not efficient at simulating voxel-based geometries (patient images). On the other hand, dedicated codes (SimSET, PETSIM) will perform well for voxel-based objects but will be poor in their capacity of simulating complex geometries such as a PET scanner. The approach adopted in this work was to couple a dedicated code (SimSET) with a general-purpose package (Geant4) to have the efficiency of the former and the capabilities of the latter. The combined SimSET+Geant4 code (SimG4) was assessed on the GE Advance PET scanner and compared to the use of SimSET only. A better description of the resolution and sensitivity of the scanner and of the scatter fraction was obtained with SimG4. The accuracy of scatter correction performed with SimG4 and SimSET was also assessed from data acquired with the 20 cm NEMA phantom. SimG4 was found to outperform SimSET and to give slightly better results than the GE scatter correction methods installed on the Advance scanner (curve fitting and scatter modelling for the 300-650 keV and 375-650 keV energy windows, respectively). In the presence of a hot source close to the edge of the field of view (as found in oxygen scans), the GE curve-fitting method was found to fail whereas SimG4 maintained its performance.

A new fast accurate nonlinear medical image registration program including surface preserving regularization.

Recently inexpensive graphical processing units (GPUs) have become established as a viable alternative to traditional CPUs for many medical image processing applications. GPUs offer the potential of very significant improvements in performance at low cost and with low power consumption. One way in which GPU programs differ from traditional CPU programs is that increasingly elaborate calculations per voxel may not impact of the overall processing time because memory accesses can dominate execution time. This paper presents a new GPU based elastic image registration program named Ezys. The Ezys image registration algorithm belongs to the wide class of diffeomorphic demons but uses surface preserving image smoothing and regularization filters designed for a GPU that would be computationally expensive on a CPU. We describe the methods used in Ezys and present results from two important neuroscience applications. Firstly inter-subject registration for transfer of anatomical labels and secondly longitudinal intra-subject registration to quantify atrophy in individual subjects. Both experiments showed that Ezys registration compares favorably with other popular elastic image registration programs. We believe Ezys is a useful tool for neuroscience and other applications, and also demonstrates the value of developing of novel image processing filters specifically designed for GPUs.

Attenuation correction methods suitable for brain imaging with a PET/MRI scanner: a comparison of tissue atlas and template attenuation map approaches.

UNLABELLED: Modeled attenuation correction (AC) will be necessary for combined PET/MRI scanners not equipped with transmission scanning hardware. We compared 2 modeled AC approaches that use nonrigid registration with rotating (68)Ge rod-based measured AC for 10 subjects scanned with (18)F-FDG. METHODS: Two MRI and attenuation map pairs were evaluated: tissue atlas-based and measured templates. The tissue atlas approach used a composite of the BrainWeb and Zubal digital phantoms, whereas the measured templates were produced by averaging spatially normalized measured MR image and coregistered attenuation maps. The composite digital phantom was manually edited to include 2 additional tissue classes (paranasal sinuses, and ethmoidal air cells or nasal cavity). In addition, 3 attenuation values for bone were compared. The MRI and attenuation map pairs were used to generate subject-specific attenuation maps via nonrigid registration of the MRI to the MR image of the subject. SPM2 and a B-spline free-form deformation algorithm were used for the nonrigid registration. To determine the accuracy of the modeled AC approaches, radioactivity concentration was assessed on a voxelwise and regional basis. RESULTS: The template approach produced better spatial consistency than the phantom-based atlas, with an average percentage error in radioactivity concentration across the regions, compared with measured AC, of -1.2% ± 1.2% and -1.5% ± 1.9% for B-spline and SPM2 registration, respectively. In comparison, the tissue atlas method with B-spline registration produced average percentage errors of 0.0% ± 3.0%, 0.9% ± 2.9%, and 2.9% ± 2.8% for bone attenuation values of 0.143 cm(-1), 0.152 cm(-1), and 0.172 cm(-1), respectively. The largest errors for the template AC method were found in parts of the frontal cortex (-3%) and the cerebellar vermis (-5%). Intersubject variability was higher with SPM2 than with B-spline. Compared with measured AC, template AC with B-spline and SPM2 achieved a correlation coefficient (R(2)) of 0.99 and 0.98, respectively, for regional radioactivity concentration. The corresponding R(2) for the tissue atlas approach with B-spline registration was 0.98, irrespective of the bone attenuation coefficient. CONCLUSION: Nonrigid registration of joint MRI and attenuation map templates can produce accurate AC for brain PET scans, particularly with measured templates and B-spline registration. Consequently, these methods are suitable for AC of brain scans acquired on combined PET/MRI systems.

Quantitative phenomenological model of the BOLD contrast mechanism.

Different theoretical models of the BOLD contrast mechanism are used for many applications including BOLD quantification (qBOLD) and vessel size imaging, both in health and disease. Each model simplifies the system under consideration, making approximations about the structure of the blood vessel network and diffusion of water molecules through inhomogeneities in the magnetic field created by deoxyhemoglobin-containing blood vessels. In this study, Monte-Carlo methods are used to simulate the BOLD MR signal generated by diffusing water molecules in the presence of long, cylindrical blood vessels. Using these simulations we introduce a new, phenomenological model that is far more accurate over a range of blood oxygenation levels and blood vessel radii than existing models. This model could be used to extract physiological parameters of the blood vessel network from experimental data in BOLD-based experiments. We use our model to establish ranges of validity for the existing analytical models of Yablonskiy and Haacke, Kiselev and Posse, Sukstanskii and Yablonskiy (extended to the case of arbitrary time in the spin echo sequence) and Bauer et al. (extended to the case of randomly oriented cylinders). Although these models are shown to be accurate in the limits of diffusion under which they were derived, none of them is accurate for the whole physiological range of blood vessels radii and blood oxygenation levels. We also show the extent of systematic errors that are introduced due to the approximations of these models when used for BOLD signal quantification.

Gaussian process modeling for image distortion correction in echo planar imaging.

An enhanced method for correction of image distortion due to B(0)-field inhomogeneities in echo planar imaging (EPI) is presented. The algorithm is based on the measurement of the point spread function (PSF) associated with each image voxel using a reference scan. The expected distortion map in the phase encode direction is then estimated using a nonparametric inference algorithm known as Gaussian process modeling. The algorithm is shown to be robust to the presence of regions of low signal-to-noise in the image and large inhomogeneities.