Non-negative data-driven mapping of structural connections with application to the neonatal brain.

Mapping connections in the neonatal brain can provide insight into the crucial early stages of neurodevelopment that shape brain organisation and lay the foundations for cognition and behaviour. Diffusion MRI and tractography provide unique opportunities for such explorations, through estimation of white matter bundles and brain connectivity. Atlas-based tractography protocols, i.e. a priori defined sets of masks and logical operations in a template space, have been commonly used in the adult brain to drive such explorations. However, rapid growth and maturation of the brain during early development make it challenging to ensure correspondence and validity of such atlas-based tractography approaches in the developing brain. An alternative can be provided by data-driven methods, which do not depend on predefined regions of interest. Here, we develop a novel data-driven framework to extract white matter bundles and their associated grey matter networks from neonatal tractography data, based on non-negative matrix factorisation that is inherently suited to the non-negative nature of structural connectivity data. We also develop a non-negative dual regression framework to map group-level components to individual subjects. Using in-silico simulations, we evaluate the accuracy of our approach in extracting connectivity components and compare with an alternative data-driven method, independent component analysis. We apply non-negative matrix factorisation to whole-brain connectivity obtained from publicly available datasets from the Developing Human Connectome Project, yielding grey matter components and their corresponding white matter bundles. We assess the validity and interpretability of these components against traditional tractography results and grey matter networks obtained from resting-state fMRI in the same subjects. We subsequently use them to generate a parcellation of the neonatal cortex using data from 323 new-born babies and we assess the robustness and reproducibility of this connectivity-driven parcellation.

Probabilistic non-linear registration with spatially adaptive regularisation.

This paper introduces a novel method for inferring spatially varying regularisation in non-linear registration. This is achieved through full Bayesian inference on a probabilistic registration model, where the prior on the transformation parameters is parameterised as a weighted mixture of spatially localised components. Such an approach has the advantage of allowing the registration to be more flexibly driven by the data than a traditional globally defined regularisation penalty, such as bending energy. The proposed method adaptively determines the influence of the prior in a local region. The strength of the prior may be reduced in areas where the data better support deformations, or can enforce a stronger constraint in less informative areas. Consequently, the use of such a spatially adaptive prior may reduce unwanted impacts of regularisation on the inferred transformation. This is especially important for applications where the deformation field itself is of interest, such as tensor based morphometry. The proposed approach is demonstrated using synthetic images, and with application to tensor based morphometry analysis of subjects with Alzheimer's disease and healthy controls. The results indicate that using the proposed spatially adaptive prior leads to sparser deformations, which provide better localisation of regional volume change. Additionally, the proposed regularisation model leads to more data driven and localised maps of registration uncertainty. This paper also demonstrates for the first time the use of Bayesian model comparison for selecting different types of regularisation.

Effects of image reconstruction on fiber orientation mapping from multichannel diffusion MRI: reducing the noise floor using SENSE.

PURPOSE: To examine the effects of the reconstruction algorithm of magnitude images from multichannel diffusion MRI on fiber orientation estimation. THEORY AND METHODS: It is well established that the method used to combine signals from different coil elements in multichannel MRI can have an impact on the properties of the reconstructed magnitude image. Using a root-sum-of-squares approach results in a magnitude signal that follows an effective noncentral-χ distribution. As a result, the noise floor, the minimum measurable in the absence of any true signal, is elevated. This is particularly relevant for diffusion-weighted MRI, where the signal attenuation is of interest. RESULTS: In this study, we illustrate problems that such image reconstruction characteristics may cause in the estimation of fiber orientations, both for model-based and model-free approaches, when modern 32-channel coils are used. We further propose an alternative image reconstruction method that is based on sensitivity encoding (SENSE) and preserves the Rician nature of the single-channel, magnitude MR signal. We show that for the same k-space data, root-sum-of-squares can cause excessive overfitting and reduced precision in orientation estimation compared with the SENSE-based approach. CONCLUSION: These results highlight the importance of choosing the appropriate image reconstruction method for tractography studies that use multichannel receiver coils for diffusion MRI acquisition.

Social network size affects neural circuits in macaques.

It has been suggested that variation in brain structure correlates with the sizes of individuals' social networks. Whether variation in social network size causes variation in brain structure, however, is unknown. To address this question, we neuroimaged 23 monkeys that had been living in social groups set to different sizes. Subject comparison revealed that living in larger groups caused increases in gray matter in mid-superior temporal sulcus and rostral prefrontal cortex and increased coupling of activity in frontal and temporal cortex. Social network size, therefore, contributes to changes both in brain structure and function. The changes have potential implications for an animal's success in a social context; gray matter differences in similar areas were also correlated with each animal's dominance within its social network.

A Bayesian framework for global tractography.

We readdress the diffusion tractography problem in a global and probabilistic manner. Instead of tracking through local orientations, we parameterise the connexions between brain regions at a global level, and then infer on global and local parameters simultaneously in a Bayesian framework. This approach offers a number of important benefits. The global nature of the tractography reduces sensitivity to local noise and modelling errors. By constraining tractography to ensure a connexion is found, and then inferring on the exact location of the connexion, we increase the robustness of connectivity-based parcellations, allowing parcellations of connexions that were previously invisible to tractography. The Bayesian framework allows a direct comparison of the evidence for connecting and non-connecting models, to test whether the connexion is supported by the data. Crucially, by explicit parameterisation of the connexion between brain regions, we infer on a parameter that is shared with models of functional connectivity. This model is a first step toward the joint inference on functional and anatomical connectivity.

Free water surface wetlands for wastewater treatment in Sweden: nitrogen and phosphorus removal.

In South Sweden, free water surface wetlands have been built to treat wastewater from municipal wastewater treatment plants. Commonly, nitrogen removal has been the prime aim, though a significant removal of tot-P and BOD7 has been observed. In this study, performance data for 3-8 years from four large (20-28 ha) FWS wetlands have been evaluated. Two of them receive effluent from WWTP with only mechanical and chemical treatment. At the other two, the wastewater has also been treated biologically resulting in lower concentrations of BOD7 and NH4+-N. The wetlands performed satisfactorily and removed 0.7-1.5 ton N ha(-1) yr(-1) as an average for the time period investigated, with loads between 1.7 and 6.3 ton N ha(-1)yr(-1). Treatment capacity depended on the pre-treatment of the water, as reflected in the k20-values for N removal (first order area based model). In the wetlands with no biological pre-treatment, the k20-values were 0.61 and 1.1 m month(-1), whereas for the other two they were 1.7 and 2.5 m month(-1). P removal varied between 10 and 41 kg ha(-1) yr(-1), and was related to differences in loads, P speciation and to the internal cycling of P in the wetlands.

Correction of MR image distortions induced by metallic objects using a 3D cubic B-spline basis set: application to stereotactic surgical planning.

Metallic implants in MRI cause spin-echo (SE) images to be distorted in the slice and frequency-encoding directions. Chang and Fitzpatrick (IEEE Trans Med Imaging 1992;11:319-329) proposed a distortion correction method (termed the CF method) based on the magnitude images from two SE acquisitions that differ only in the polarity of the frequency-encoding and slice-selection gradients. In the present study we solved some problems with the CF method, primarily by modeling the field inhomogeneities as a single 3D displacement field built by 3D cubic B-splines. The 3D displacement field was applied in the actual distortion direction in the slice/frequency-encoding plane. To account for patient head motion, a 3D rigid body motion correction was also incorporated in the model. Experiments on a phantom containing an aneurysm clip showed that the knot spacing between the B-splines is a very important factor in both the final image quality and the processing speed. Depending on the knot spacing and the image volume size, the number of unknowns range from a few thousands to over 100,000, leading to processing times ranging from minutes to days. Optimal knot spacing, a means of increasing the processing speed, and other parameters are investigated and discussed.

Haemodynamic and metabolic disturbances in the acute stage of subarachnoid haemorrhage demonstrated by PET.

OBJECTIVE: To investigate the occurrence of early haemodynamic and metabolic changes in the acute stage of subarachnoid haemorrhage (SAH). MATERIAL AND METHODS: Eleven SAH patients were studied. Regional cerebral blood flow (CBF), cerebral metabolic rate of oxygen (CMRO2) and oxygen extraction ratio (OER) were measured with positron emission tomography (PET) 22-53 h after haemorrhage, utilizing 15O-labelled water bolus and the 15O-inhalation technique. Ten volumes of interest (VOIs) representing vascular territories were outlined in each patient according to a standardized procedure. The occurrence of irreversible ischaemia, penumbra, oligaemia, hyperperfusion and normal haemodynamics according to PET criteria was investigated. These pathophysiological categories were related to final tissue outcome as determined by follow-up computed tomography (CT). RESULTS: All five tissue subtypes were represented in the vascular region VOIs; oligaemia was the predominant pathophysiological pattern. When global changes were analysed, blood flow was reduced in three, oxygen metabolism was reduced in four, and OER was increased in four of seven unsedated patients, respectively. The sedated patients all had markedly reduced CBF and CMRO2 and OER in the high or supranormal range. CONCLUSION: Haemodynamic and metabolic disturbances proved to be common after SAH. These abnormalities probably reflect the primary brain injury caused by the initial haemorrhage. The impact of secondary insults such as acute hydrocephalus, brain oedema, vasospasm, seizures, hypotension and hypoxaemia are likely to be dependent on the degree of primary injury, which can be assessed by PET.

On the effects of gating in diffusion imaging of the brain using single shot EPI.

To reduce image artifacts in diffusion tensor imaging using single shot EPI, cardiac gating may be performed to prevent brain motion induced signal void in the DWI data. In this work the necessity of performing cardiac gating with single shot echo planar imaging has been explored using F statistics of the variance in DWI data. Peripheral measurement of the cardiac cycle has been employed because of its greater convenience, and hence use, compared to ECG gating. Four volunteers have been studied. Six different trigger delays have been evaluated in the range 18-500 ms relative the peripheral pulse wave. Brain motion was found significant in and inferior to the corpus callosum area. A trigger delay of 500 ms was found to be optimal. However, a long delay allows only for one slice per R-R interval. Therefore, a minimum trigger delay may be the best choice in terms of SNR per unit of experiment time.

Abnormal brain white matter in schizophrenia: a diffusion tensor imaging study.

Fractional anisotropy and the mean diffusion coefficient were measured in the cerebral volume in 20 schizophrenic and 24 healthy subjects, men and women, using diffusion tensor imaging. In addition, 3D SPGR was used for segmentation of brain tissue into grey and white matter and cerebrospinal fluid. In schizophrenic patients, fractional anisotropy was reduced in the splenium of the corpus callosum and in adjacent occipital white matter. The segmentation revealed no tissue deficits in the volume of reduced fractional anisotropy. The mean diffusion was increased in the total white and grey matter volume of the schizophrenic patients compared with the healthy subjects. The findings support the view that global and regional white matter abnormalities occur in chronic schizophrenia.

A global estimator unbiased by local changes.

The global activity is an important confound when analyzing PET data in that its inclusion in the statistical model can substantially reduce error variance and increase sensitivity. However, by defining global activity as the average over all voxels one introduces a bias that is collinear with experimental factors. This leads to an underestimation of true activations and the introduction of artefactual deactivations. We propose a novel estimator for the global activity based on the notion of finding a maximally nonlocal mode in a multivariate characterization of the data, while maximizing the locality of the remaining modes. The approach uses singular value decomposition (SVD) to find a provisional set of modes, which are subsequently rotated such that a metric based on the above heuristic is maximized. This metric is a version of the stochastic sign change (SSC) criterion that has been used previously for normalizing medical images with focal defects. The estimator was evaluated on simulated and real functional imaging (PET) data. The simulations show that the bias of the global mean, introduced by focal activations, is reduced by 80--90% with the new estimator. Comparison with a previous unbiased estimator, using the empirical data, yielded similar results. The advantage of the new estimator is that it is not informed of experimental design and relies only on general assumptions regarding the nature of the signal.

Modeling geometric deformations in EPI time series.

Even after realignment there is residual movement-related variance present in fMRI time-series, causing loss of sensitivity and, potentially, also specificity. One cause is the differential deformation of the sampling matrix, by field inhomogeneities, at different object positions, i.e., a movement-by-inhomogeneity interaction. This has been addressed previously by using empirical field measurements. In the present paper we suggest a forward model of how data is affected by an inhomogeneous field at different object positions. From this model we derive a method to solve the inverse problem of estimating the field inhomogeneities and their derivatives with respect to object position, directly from the EPI data and estimated realignment parameters. The field is modeled as a linear combination of cosine basis fields, which facilitates a fast way of implementing the necessary matrix operations. Simulations suggest that the solution is tractable and that the fields are estimable given the deformed images and knowledge of the relative positions at which they have been acquired. An experiment on a subject performing voluntary movements in the scanner yielded plausible estimates of the deformation fields and their application to "unwarp" the time series significantly reduced movement-related variance.

Fear conditioning and brain activity: a positron emission tomography study in humans.

Regional cerebral blood flow (rCBF) was measured with H2 (15)O positron emission tomography in 8 healthy women before and after fear conditioning (i.e., paired shocks) and unpaired shocks to videotape cues. Conditioning was supported by enhanced peripheral nervous system recordings and subjective ratings. Fear conditioning increased rCBF in the central gray of the midbrain; bilaterally in the hypothalamus, the thalamus, and the left striatum; and in the right and left anterior cingulate and right prefrontal cortices. Regional CBF was attenuated bilaterally in the right and left prefrontal, temporal (including the amygdala), parietal, and occipital cortices, and in the left orbitofrontal cortex. When compared with unpaired shock presentations, fear conditioning resulted in elevated rCBF in the left cerebellum. Hence, in the present paradigm, only neural activity in the left cerebellum solely reflected processes associated with true Pavlovian conditioning.

Improved efficiency for MRI-SPET registration based on mutual information.

Mutual information has been proposed as a criterion for image registration. The criterion is calculated from a two-dimensional grey-scale histogram of the image pair being registered. In this paper we study how sparse sampling can be used to increase speed performance using the registration algorithm of Maes et al. (IEEE Trans Med Imaging 1997; 16: 187-198) with a focus on registration of MRI-SPET brain images. In particular we investigate how sparse sampling and parameters such as the number of bins used for the grey-scale histograms and smoothing of the data prior to registration affect accuracy and robustness of the registration. The method was validated using both simulated and human data. Our results show that sparse sampling introduced local maxima into the mutual information similarity function when the number of bins used for the histograms was large. To speed up registration while retaining robustness, smoothing of the data prior to registration was used and a coarse to fine subsampling protocol, where the number of bins in the histograms were dependent on the subsampling factor, was employed, For the simulated data, the method was able to recover known transformations with an accuracy of about 1 mm. Using the human data, there were no significant differences in the recovered transformation parameters when the suggested subsampling scheme was used compared with when no subsampling was used, but there was a more than tenfold increase in speed. Our results show that, with the appropriate choice of parameters, the method can accurately register MRI-SPET brain images even when very efficient sampling protocols are used.

A metabolic threshold of irreversible ischemia demonstrated by PET in a middle cerebral artery occlusion-reperfusion primate model.

OBJECTIVE: to evaluate the predictive value of measurements of regional cerebral blood flow (CBF), oxygen metabolism (CMRO2) and oxygen extraction ratio (OER) for assessment of the fate of ischemic brain tissue. MATERIALS AND METHODS: Sequential PET measurements were performed during middle cerebral artery occlusion (MCAO; 2 h) and 12-24 h (mean 18 h) of reperfusion in a primate model (Macaca mulatta, n = 8). A penumbra region was delineated on the MCAO PET image (OER > 125% and CMRO2> or = 45% of the values observed in the contralateral hemisphere, respectively) and an infarction region was delineated on the last PET image (CMRO2 <45% of the values observed in the contralateral hemisphere). The penumbra regions delineated during MCAO and the infarction regions delineated at the final PET, were copied on to the images from all other PET sessions for measurements of CBF, CMRO2 and OER. Ratios were calculated by dividing the mean values obtained by the values of the corresponding contralateral region. RESULTS: Histopathology verified the adequacy of the criteria applied in the last PET for delineation of the infarction region. The penumbra region and infarction region were separated in all cases, except in two cases where a minimal overlap was seen. CBF and OER showed considerable variation over time and there was no consistent difference between the penumbra and infarction regions. CMRO2 showed a more stable pattern and the difference between penumbra and infarction regions was maintained from the time of MCAO throughout the entire reperfusion phase. With CMRO2 as predictor, all 50 observations could be correctly predicted as penumbra or infarction when using an optimal threshold ratio value estimated to be in the interval of 61% to 69% of the corresponding contralateral region. CBF and OER proved to have low power as predictors. CONCLUSIONS: The results indicate that CMRO2 is the best predictor of reversible or irreversible brain damage and the critical metabolic threshold level appears to be a reduction of oxygen metabolism to between 61% and 69% of the corresponding contralateral region.

Image registration using a symmetric prior--in three dimensions.

This paper describes a Bayesian method for three-dimensional registration of brain images. A finite element approach is used to obtain a maximum a posteriori estimate of the deformation field at every voxel of a template volume. The priors used by the MAP estimate penalize unlikely deformations and enforce a continuous one-to-one mapping. The deformations are assumed to have some form of symmetry, in that priors describing the probability distribution of the deformations should be identical to those for the inverses (i.e., warping brain A to brain B should not be different probablistically from warping B to A). A gradient descent algorithm is presented for estimating the optimum deformations.

Registration of neuroimaging data: implementation and clinical applications.

Image registration brings images into a form in which each voxel corresponds to a predetermined anatomic entity and is necessary for comparisons of data across scans. Intrasubject registration is a matter of translating and rotating one image volume into correspondence with another. Intersubject registration is more difficult because it requires the removal of individual anatomy dependence from the data. This article describes, with the help of clinical examples, automated methods for intrasubject registration of scans within and between modalities, and intersubject registration used for registering a three-dimensional brain atlas with a patient's brain scan.

Tyrosine transport is regulated differently in patients with schizophrenia.

Previous PET studies of tyrosine transport have suggested that the transport of tyrosine from blood to brain compartment is not dependent on its plasma concentration in patients with schizophrenia. In order to examine this relationship, the transport constant (K1) of tyrosine was determined in five patients with schizophrenia and five normals. L-[1-11C]Tyrosine was injected i.v. and arterial blood samples were taken during PET scanning. The tyrosine transport was assessed during baseline conditions and after oral administration of L-tyrosine at a dose (175 mg/kg) that significantly elevated the plasma levels. K1 was determined from tracer kinetic modelling. The transport rate dropped in the normals after tyrosine loading, which is consistent with the prevailing notion that the brain transport system for neutral amino acids works close to saturation, whereas it was virtually unchanged in the schizophrenics. The results demonstrated that tyrosine transport was not saturated in the patients with schizophrenia and thus could lead to elevated brain concentrations of tyrosine.

A multivariate approach to registration of dissimilar tomographic images.

We devised a method to allow for retrospective registration of tomographic images with very different information content, the main emphasis being on sets of positron emission tomography images obtained with different tracers. A multivariate cost-function based on information theory was used as an index of "goodness-of-alignment". The cost-function makes no assumptions regarding the form of the relationship between the two image sets, and is hence very general. Image volumes, with known relative spatial orientation, were simulated for tracers with different uptake patterns and the method was validated by assessing its ability to recover these known orientations. The method was able to recover the known transformations with an accuracy of about 1 mm and 1 degrees along and around all axes, even for tracer combinations with radically different uptake patterns and with large initial misalignment. With the suggested method it is feasible to retrospectively align examinations obtained with different tracers and/or modalities.

Brain regions associated with episodic retrieval in normal aging and Alzheimer's disease.

OBJECTIVE: To examine patterns of brain activation during verbal episodic retrieval in normal elderly subjects and patients in an early phase of AD. BACKGROUND: It is established that 1) a profound episodic memory impairment is a cardinal symptom of AD; and 2) some of the earliest brain changes in this disease occur in regions critical to episodic memory, such as the hippocampus and neighboring regions. Yet, it remains largely unknown whether the episodic memory deficit seen in AD is paralleled by concomitant alterations in brain activity during actual task performance in these or other brain areas. METHODS: Using PET, blood flow was assessed in normal elderly subjects and patients with early AD during two retrieval conditions involving completion of word stems: baseline and cued recall. RESULTS: The patients with AD showed a marked performance deficit in cued recall, although the two groups were indistinguishable in the baseline task condition. Both groups showed bilateral activity in orbital and dorsolateral prefrontal cortex, left precuneus, and right cerebellum, as well as decreased activity in distinct left temporal regions during cued recall. The normal elderly alone activated the left parietal cortex and the left hippocampal formation during episodic retrieval. By contrast, AD-related increases in activity during cued recall were observed in the left orbital prefrontal cortex and left cerebellum. CONCLUSIONS: The similar patterns of activations in the two groups suggest that a large distributed network involved in episodic memory retrieval functions relatively normally in early AD. Those retrieval activations seen in the normal elderly, as opposed to the patients, may reflect AD-related failures in semantic processing and successful recollection of the target information, respectively. Finally, the AD-related increases in activity were interpreted in terms of compensatory reactions to the difficulties in performing the episodic memory task.