Keywords: brain atlas; multispectral magnetic resonance imaging; diffusion tensor imaging; unbiased population registration; spatial normalization; label propagation

Postural stability becomes compromised with advancing age, but the neural mechanisms contributing to instability have not been fully explicated. Accordingly, this quantitative physiological and MRI study of sex differences across the adult age range examined the association between components of postural control and the integrity of brain structure and function under different conditions of sensory input and stance stabilization manipulation. The groups comprised 28 healthy men (age 30-73 years) and 38 healthy women (age 34-74 years), who completed balance platform testing, cognitive assessment, and structural MRI. The results supported the hypothesis that excessive postural sway would be greater in older than younger healthy individuals when standing without sensory or stance aids, and that introduction of such aids would reduce sway in both principal directions (anterior-posterior and medial-lateral) and in both the open-loop and closed-loop components of postural control even in older individuals. Sway reduction with stance stabilization, that is, standing with feet apart, was greater in men than women, probably because older men were less stable than women when standing with their feet together. Greater sway was related to evidence for greater brain structural involutional changes, indexed as ventricular and sulcal enlargement and white matter hyperintensity burden. In women, poorer cognitive test performance related to less sway reduction with the use of sensory aids. Thus, aging men and women were shown to have diminished postural control, associated with cognitive and brain structural involution, in unstable stance conditions and with diminished sensory input.

Keywords: Postural control; Posturography; Balance; Cerebellum; Brain; MRI; White matter hyperintensity; Sway; Age; Sex

Neurophysiological, biochemical, and anatomical evidence implicates glutamatergic mechanisms in epileptic seizures. Until recently, however, longitudinal characterization of in vivo glutamate dynamics was not possible. Here, we present data using in vivo magnetic resonance spectroscopy (MRS) optimized for the detection of glutamate to identify changes that evolve following kainic acid (KA)-induced status epilepticus. Wild-type male Wistar rats underwent whole brain MR imaging and single voxel MRS on a clinical 3T scanner equipped with a high-strength insert gradient coil. Scanning took place before and then 3 days, 28 - 32 days, and 42 - 50 days after induction of status epilepticus. Analyses compared 5 seizure (Sz), 5 no-seizure (NoSz; received KA but did not exhibit seizures), and 6 control (Con) animals. This longitudinal study demonstrated reduced glutamate levels in vivo in the dorsal hippocampus 3 days and 1 month following status epilepticus in Sz animals compared with Con animals. Additionally, previous results were replicated: in the Sz group, computed T2 was higher in the ventral hippocampus and limbic cortex 3 days after seizure activity compared with baseline but resolved in both regions at the 1 month scan, suggesting a transient edema. Three days following seizure activity, N-acetylaspartate (NAA) declined and lactate increased in the dorsal hippocampus of the Sz group compared with the Con and NoSz group; both metabolites approached baseline levels by the third scan. Taken together, these results support the conclusion that seizure activity following KA infusion causes loss of glutamatergic neurons.

Keywords: Glutamate; Hippocampus; Cerebellum; Longitudinal; Rat; Status epilepticus

Background: Quantitative fiber tracking derived from diffusion tensor imaging (DTI) was used to determine whether white matter association, projection, or commissural tracts are affected in nondemented individuals with HIV infection and to identify the regional distribution of sparing and impairment of fiber systems.

Methods: DTI measured fractional anisotropy and diffusivity, quantified separately for longitudinal ([lambda]L) diffusivity (index of axonal injury) and transverse ([lambda]T) diffusivity (index of myelin injury), in 11 association and projection white matter tracts and six commissural tracts in 29 men and 13 women with HIV infection and 88 healthy, age-matched controls (42 men and 46 women).

Results: The total group of HIV-infected individuals had higher diffusivity (principally longitudinal) than controls in the posterior sectors of the corpus callosum, internal and external capsules, and superior cingulate bundles. High longitudinal diffusivity, indicative of axonal compromise, was especially prominent in posterior callosal sectors, fornix, and superior cingulate bundle in HIV with AIDS. Unmedicated patients had notably high transverse diffusivity, indicative of myelin compromise, in the occipital forceps, inferior cingulate bundle, and superior longitudinal fasciculus. Pontocerebellar projection fibers were resistant to HIV effects as were commissural fibers coursing through premotor and sensorimotor callosal sectors.

Conclusion: This quantitative survey of brain fiber tract integrity indicates that even nondemented HIV patients can have neuroradiological evidence for damage to association and commissural tracts. These abnormalities were vulnerable to exacerbation with AIDS and possibly mitigated by HAART.

Keywords: AIDS, brain, cognition, dementia, diffusion tensor imaging, HAART, HIV, MRI, white matter

Different brain structures accumulate iron at different rates throughout the adult life span. Typically, striatal and brain stem structures are higher in iron concentrations in older than younger adults, whereas cortical white matter and thalamus have lower concentrations in the elderly than young adults. Brain iron can be measured in vivo with MRI by estimating the relaxivity increase across magnetic field strengths, which yields the Field-Dependent R2 Increase (FDRI) metric. The influence of local iron deposition on susceptibility, manifest as MR phase effects, forms the basis for another approach for iron measurement, Susceptibility-Weighted Imaging (SWI), for which imaging at only one field strength is sufficient. Here, we compared the ability of these two methods to detect and quantify brain iron in 11 young (5 men, 6 women; 21 to 29 years) and 12 elderly (6 men, 6 women; 64 to 86 years) healthy adults. FDRI was acquired at 1.5T and 3.0T, and SWI was acquired at 1.5T. The results showed that both methods detected high globus pallidus iron concentration regardless of age and significantly greater iron in putamen with advancing age. The SWI measures were more sensitive when the phase signal intensities themselves were used to define regions of interest, whereas FDRI measures were robust to the method of region of interest selection. Further, FDRI measures were more highly correlated than SWI iron estimates with published postmortem values and were more sensitive than SWI to iron concentration differences across basal ganglia structures. Whereas FDRI requires more imaging time than SWI, two field strengths, and across-study image registration for iron concentration calculation, FDRI appears more specific to age-dependent accumulation of non-heme brain iron than SWI, which is affected by heme iron and non-iron source effects on phase.

The aim of the present study was to quantify both perfusion and extravasation in the prostate to discriminate tumor from healthy tissue, which might be achieved by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) using a nonspecific low-molecular-weight contrast medium (CM). To determine extravasation as well as tissue perfusion an inversion-prepared dual-contrast sequence employing a parallel acquisition technique (PAT) was designed for interleaved acquisition of T (1)-weighted images for extravasation measurement and T2[Formula: see text]-weighted images for determination of the highly concentrated bolus with a sufficiently high temporal and spatial resolution at an acceptable signal-to-noise ratio. Thirteen patients with proven prostate cancer were examined with the sequence using a combined body-array prostate coil. Before pharmacokinetic evaluation the images were intensity-corrected and, if required, motion-corrected. The pharmacokinetic model used to calculate perfusion, permeability, blood volume, interstitial volume, transit time, and vessel size index included two compartments and a correction of delay and dispersion of the arterial input function. The information provided by the dual-contrast sequence allowed application of a more elaborate model for evaluation and enabled quantification of all parameters. Peripheral prostate tumors were found to differ from peripheral healthy prostate tissue in perfusion (1.38 mL/(min cm(3)) vs. 0.23 mL/(min cm(3)), p = 0.004), mean transit time (2.88 vs. 4.88 s, p = 0.039), and blood volume (1.9 vs. 0.7%, p = 0.019). A inversion-prepared dual-contrast sequence acquiring T (1)- and [Formula: see text]-weighted images with sufficient temporal resolution and signal-to-noise ratio was successfully applied in patients with prostate cancer to quantify all pharmacokinetic parameters of inflow and extravasation of a low-molecular-weight inert tracer.

Keywords: Dynamic contrast-enhanced magnetic resonance imaging; Perfusion; Permeability; Dual contrast; Prostate

Iron deposition increases in normal aging, has its greatest presence in structures of the extrapyramidal system, and may contribute to functional decline. MR imaging provides a method for indexing iron deposition in brain structures because of iron’s ferromagnetic properties, which interact with the MRI environment to cause signal intensity attenuation that is quantifiable by comparing images collected at 1.5 and 3.0 T. We tested functional correlates of an MR-based iron index in 10 healthy, elderly individuals previously reported to have a higher iron burden in the putamen and lower in the thalamus than young individuals. Lower scores on the Dementia Rating Scale and longer reaction times on a two-choice attention test correlated with higher iron estimates in the caudate nucleus and putamen; poorer Mini-Mental State Examination and Digit Symbol scores correlated with lower iron estimates in the thalamus. Further analyses based on multiple regression, which considered regional FDRI estimates and volume measures as predictors of performance, identified iron but not the sampled volume as the unique predictor in each case. These exploratory correlations suggest a substrate of performance degradation in aging and have implications for regional signal darkening in an array of MR-based imaging protocols.

Keywords: Iron; Basal ganglia; Extrapyramidal system; Thalamus; MRI; Cognition; Motor; Age; Aging

BACKGROUND: Excessive alcohol use can cause macrostructural tissue shrinkage with regional preference for frontal systems. The extent and locus of alcoholism's effect on white matter microstructure is less known. METHODS: Quantitative fiber tracking derived from diffusion tensor imaging (DTI) assessed the integrity of samples of 11 major white matter bundles in 87 alcoholics (59 men, 28 women) and 88 healthy control subjects (42 men, 46 women). Fiber integrity was expressed as fractional anisotropy (FA) and apparent diffusion coefficient (ADC), quantified separately for longitudinal diffusivity (lambdaL), a putative index of axonal integrity, and transverse diffusivity (lambdaT), a putative index of myelin integrity. RESULTS: Alcoholism affected FA and diffusivity, particularly lambdaT, of several fiber bundles. Frontal and superior sites (frontal forceps, internal and external capsules, fornix, and superior cingulate and longitudinal fasciculi) showed greatest abnormalities in alcoholics relative to control subjects. More posterior and inferior bundles were relatively spared. Lifetime alcohol consumption correlated with regional DTI measures in alcoholic men but not women. When matched for alcohol exposure, alcoholic women showed more DTI signs of white matter degradation than alcoholic men in several fiber bundles. Among all alcoholics, poorer performance on speeded tests correlated with DTI signs of regional white matter degradation. CONCLUSIONS: This survey of multiple brain fiber systems revealed a differential pattern of alcoholism's effect on regional FA and diffusivity with functional consequences attributable in part to compromised fiber microstructure with prominence in signs of myelin degradation. Sex-based differences suggest that women are at enhanced risk for alcoholism-related degradation in selective white matter systems.

Keywords: Alcoholism; brain; diffusion; DTI; fiber tracking; sex differences; white matter

Normal aging is accompanied by decline in selective cognitive and motor functions. A concurrent decline in regional white matter integrity, detectable with diffusion tensor imaging (DTI), potentially contributes to waning function. DTI analysis of white matter loci indicates an anterior-toposterior gradient distribution of declining fractional anisotropy (FA) and increasing diffusivity with age. Quantitative fiber tracking can be used to determine regional patterns of normal aging of fiber systems and test the functional ramifications of the DTI metrics. Here, we used quantitative fiber tracking to examine age effects on commissural (genu and splenium), bilateral association (cingulate, inferior longitudinal fasciculus and uncinate), and fornix fibers in 12 young and 12 elderly healthy men and women and tested functional correlates with concurrent assessment of a wide range of neuropsychological abilities. Principal component analysis of cognitive and motor tests on which the elderly achieved significantly lower scores than the young group was used for data reduction and yielded three factors: Problem Solving, Working Memory, and Motor. Age effects - lower FA or higher diffusivity - in the elderly were prominent in anterior tracts, specifically, genu, fornix, and uncinate fibers. Differential correlations between FA or diffusivity in fiber tracts and scores on Problem Solving, Working Memory, or Motor factors provide convergent validity to the biological meaningfulness of the integrity of the fibers tracked. The observed pattern of relations supports the possibility that regional degradation of white matter fiber integrity is a biological source of age-related functional compromise and may have the potential to limit accessibility to alternative neural systems to compensate for compromised function.

Keywords: Fiber tracking; White matter; DTI; Cognition; Motor; Problem solving; Working memory; Age

Diffusion tensor imaging (DTI) of the brain has become a mainstay in the study of normal aging of white matter, and only recently has attention turned to the use of DTI to examine aging effects in gray matter structures. Of the many changes in the brain that occur with advancing age is increased presence of iron, notable in selective deep gray matter structures. In vivo detection and measurement of iron deposition is possible with magnetic resonance imaging (MRI) because of iron's effect on signal intensity. In the process of a DTI study, a series of diffusion-weighted images (DWI) is collected, and while not normally considered as a major dependent variable in research studies, they are used clinically and they reveal striking conspicuity of the globus pallidus and putamen caused by signal loss in these structures, presumably due to iron accumulation with age. These iron deposits may in turn influence DTI metrics, especially of deep gray matter structures. The combined imaging modality approach has not been previously used in the study of normal aging. The present study used legacy DTI data collected in 10 younger (22-37 years) and 10 older (65-79 years) men and women at 3.0 T and fast spin-echo (FSE) data collected at 1.5 T and 3.0 T to derive an estimate of the field-dependent relaxation rate increase (the “FDRI estimate”) in the putamen, caudate nucleus, globus pallidus, thalamus, and a frontal white matter sample comparison region. The effect of age on the diffusion measures in the deep gray matter structures was distinctly different from that reported in white matter. In contrast to lower anisotropy and higher diffusivity typical in white matter of older relative to younger adults observed with DTI, both anisotropy and diffusivity were higher in the older than younger group in the caudate nucleus and putamen; the thalamus showed little effect of age on anisotropy or diffusivity. Signal intensity measured with DWI was lower in the putamen of elderly than young adults, whereas the opposite was observed for the white matter region and thalamus. As a retrospective study based on legacy data, the FDRI estimates were based on FSE sequences, which underestimated the classical FDRI index of brain iron. Nonetheless, the differential effects of age on DTI metrics in subcortical gray matter structures compared with white matter tracts appears to be related, at least in part, to local iron content, which in the elderly of the present study was prominent in the FDRI estimate of the putamen and visibly striking in the diffusion-weighted image of the basal ganglia structures.

Keywords: Brain; Aging; Iron; DTI; DWI; MRI; Diffusion

In order to understand the connectivity of neuronal networks, their constituent neurons should ideally be studied in a common framework. Since morphological data from physiologically characterized and stained neurons usually arise from different individual brains, this can only be performed in a virtual standardized brain that compensates for interindividual variability. The desert locust, Schistocerca gregaria, is an insect species used widely for the analysis of olfactory and visual signal processing, endocrine functions, and neural networks controlling motor output. To provide a common multi-user platform for neural circuit analysis in the brain of this species, we have generated a standardized three-dimensional brain of this locust. Serial confocal images from whole-mount locust brains were used to reconstruct 34 neuropil areas in ten brains. For standardization, we compared two different methods: an iterative shape-averaging (ISA) procedure by using affine transformations followed by iterative nonrigid registrations, and the Virtual Insect Brain (VIB) protocol by using global and local rigid transformations followed by local nonrigid transformations. Both methods generated a standard brain, but for different applications. Whereas the VIB technique was designed to visualize anatomical variability between the input brains, the purpose of the ISA method was the opposite, i.e., to remove this variability. A novel individually labeled neuron, connecting the lobula to the midbrain and deutocerebrum, has been registered into the ISA atlas and demonstrates its usefulness and accuracy for future analysis of neural networks. The locust standard brain is accessible at http://www.3d-insectbrain.com.

Keywords: Virtual Insect Brain protocol; Iterative shape averaging; Three-dimensional reconstruction; Standard brain; Virtual neuroanatomy; Schistocerca gregaria (Insecta)

Background: Structural magnetic resonance imaging (MRI) reveals widespread brain damage manifest as tissue shrinkage and complementary ventriculomegaly in human alcoholism. For an animal model to parallel the human condition, high alcohol exposure should produce similar radiologically detectable neuropathology. Our previous structural MRI study demonstrated only modest brain dysmorphology of the alcohol-preferring (P) rat with average blood alcohol levels (BALs) of 125 mg/dl achieved with voluntary consumption. Here, we tested the hypothesis that wild-type Wistar rats, exposed to vaporized alcohol ensuring higher BALs than typically achieved with voluntary consumption in rodents, would model MRI findings in the brains of humans with chronic alcoholism.

Methods: The longitudinal effects of vaporized alcohol exposure on the brains of 10 wild-type Wistar rats compared with 10 sibling controls were investigated with structural MRI, conducted before (MRI 1) and after (MRI 2) 16 of alcohol exposure and after an additional 8 weeks at a higher concentration of alcohol (MRI 3).

Results: Two rats in the alcohol group died prior to MRI 2. The remaining vapor-exposed rats (n = 8) achieved BALs of 293 mg/dl by MRI 2 and 445 mg/dl by MRI 3. Whereas the controls gained 17% of their body weight from MRI 1 to MRI 3, the alcohol-exposed group lost 6%. MRI, quantified with atlas-based parcellation, revealed a profile of significant ventricular expansion, after alcohol vapor exposure, in 9 contiguous slices, extending from the dorsolateral to ventrolateral ventricles. In particular, from MRI 1 to MRI 2, this ventricular volume expanded by an average of 6.5% in the controls and by 27.1% in the alcohol-exposed rats but only an additional 1.5% in controls and 2.4% in alcohol-exposed rats from MRI 2 to MRI 3. The midsagittal volume of the full anterior-to-posterior extent of the corpus callosum grew between the first 2 MRIs in both groups followed by regression in the alcohol group by MRI 3. Although group differences were statistically significant, among animals there was substantial variability of the effects of alcohol exposure on brain morphology; some animals showed profound effects, whereas others were essentially unaffected.

Conclusions: The ventricular dilatation and callosal shrinkage produced in wild-type rats following involuntary alcohol exposure yielded a modestly successful model of neurodysmorphology phenotypes of human alcoholism. As is the case for the human condition, however, in which some individuals express greater alcoholism-related neuropathology than others, some rats may be more susceptible than others to extreme alcohol exposure.

Keywords: Alcohol; Rat; Magnetic Resonance Imaging; Vaporized Alcohol; Brai;, Corpus Callosum; Ventricles

In Drosophila, ˜50 classes of olfactory receptor neurons (ORNs) send axons to 50 corresponding glomeruli in the antennal lobe. Uniglomerular projection neurons (PNs) relay olfactory information to the mushroom body (MB) and lateral horn (LH). Here, we combine single-cell labeling and image registration to create high-resolution, quantitative maps of the MB and LH for 35 input PN channels and several groups of LH neurons. We find (1) PN inputs to the MB are stereotyped as previously shown for the LH; (2) PN partners of ORNs from different sensillar groups are clustered in the LH; (3) fruit odors are represented mostly in the posterior-dorsal LH, whereas candidate pheromone-responsive PNs project to the anterior-ventral LH; (4) dendrites of single LH neurons each overlap with specific subsets of PN axons. Our results suggest that the LH is organized according to biological values of olfactory input.

We introduce an algorithm for segmenting brain magnetic resonance (MR) images into anatomical compartments such as the major tissue classes and neuro-anatomical structures of the gray matter. The algorithm is guided by prior information represented within a tree structure. The tree mirrors the hierarchy of anatomical structures and the sub-trees correspond to limited segmentation problems. The solution to each problem is estimated via a conventional classifier. Our algorithm can be adapted to a wide range of segmentation problems by modifying the tree structure or replacing the classifier. We evaluate the performance of our new segmentation approach by revisiting a previously published statistical group comparison between first-episode schizophrenia patients, first-episode affective psychosis patients, and comparison subjects. The original study is based on 50 MR volumes in which an expert identified the brain tissue classes as well as the superior temporal gyrus, amygdala, and hippocampus. We generate analogous segmentations using our new method and repeat the statistical group comparison. The results of our analysis are similar to the original findings, except for one structure (the left superior temporal gyrus) in which a trend-level statistical significance (p=0.07) was observed instead of statistical significance.

Keywords: Automatic Segmentation; Data Tree; Expectation-Maximization; Parcellation; Statistical Group Comparison Study; MRI

A new method for averaging multidimensional images is presented, which is based on signed Euclidean distance maps computed for each of the pixel values. We refer to the algorithm as “shape-based averaging” (SBA) because of its similarity to Raya and Udupa's shape-based interpolation method. The new method does not introduce pixel intensities that were not present in the input data, which makes it suitable for averaging nonnumerical data such as label maps (segmentations). Using segmented human brain magnetic resonance images, SBA is compared to label voting for the purpose of averaging image segmentations in a multiclassifier fashion. SBA, on average, performed as well as label voting in terms of recognition rates of the averaged segmentations. SBA produced more regular and contiguous structures with less fragmentation than did label voting. SBA also was more robust for small numbers of atlases and for low atlas resolutions, in particular, when combined with shape-based interpolation. We conclude that SBA improves the contiguity and accuracy of averaged image segmentations.

Keywords: Combination of segmentations; shape-based averaging (SBA); shape-based interpolation (SBI); signed Euclidean distance transform

We investigated whether changes in memory or static balance in chronic alcoholics, occurring with abstinence or relapse, are associated with changes in lateral and fourth ventricular volume. Alcoholics meeting DSM-IV criteria for Alcohol Dependence (n = 15) and non-alcoholic controls (n = 26) were examined twice at a mean interval of 2 years with standard Wechsler Abbreviated Scale of Intelligence (WASI), Wechsler Memory Scale Revised (WMS-R) tests, an ataxia battery, and structural MRI. At study entry, alcoholics had been abstinent on average for over 4 months and achieved lower scores than controls on WASI General IQ Index, WMS-R General Memory Index, and the ataxia battery. The 10 alcoholics who maintained sobriety at retest did not differ at study entry in socio-demographic measures, alcohol use, or WASI and WMS-R summary scores from the five relapsers. At follow-up, abstainers improved more than controls on the WMS-R General Memory Index. Ataxia tended to improve in abstainers relative to controls. Associations were observed between memory and lateral ventricular volume change and between ataxia and fourth ventricular volume change in alcoholics but not in the controls. Both memory and ataxia can improve with sustained sobriety, and brain behavior associations suggest selective brain structural substrates for the changes observed.

Keywords: MRI; Ventricles; Recovery; Ataxia; Brain behavior association

Nonrigid registration and atlas-based parcellation methods were used to compare the volume of the ventricular system and the cross-sectional area of the midsagittal corpus callosum on brain MRIs from 272 subjects in four groups: patients with HIV infection, with and without alcoholism comorbidity, alcoholics, and controls. Prior to testing group differences in regional brain metrics, each measure was corrected by regression analysis for significant correlations with supratentorial cranial volume and age, observed in 121 normal control men and women, whose age spanned six decades. Disregarding HIV disease severity, we observed a graded pattern of modest enlargement of the total ventricular system (0.28 SD for uncomplicated HIV, 0.65 SD for HIV comorbid with alcoholism, and 0.72 SD for the alcoholism group). The pattern of callosal thinning showed a similar but small (˜0.5 SD) graded effect. A different pattern emerged, however, when HIV severity in the context of alcoholism comorbidity was factored into the analysis. Substantially greater volume abnormalities were present in individuals with a history of an AIDS-defining event or low CD4+ T cell counts (<=200 mm) irrespective of alcoholism comorbidity, and the effect of HIV severity was disproportionately exacerbated by alcoholism comorbidity, with 1 SD size deficit in the genu of corpus callosum and nearly 2 SD greater volume of the frontal and body regions of the ventricles for the AIDS + alcohol comorbid group. The differences in brain volumes between the AIDS groups with vs. without alcoholism could not be attributed to differences in HIV disease severity, defined by CD4+ count, viral load, or Karnofsky score. The substantial effect of the alcoholism-AIDS interaction on ventricular and callosal dysmorphology, in the context of the modest changes observed in non-AIDS, nonalcohol abusing HIV-infected individuals, highlight the need to consider alcohol use disorders as a major risk factor for neuropathology among HIV-infected persons.

Substantial changes in brain morphology mark the course of alcoholism from development through dependence, recovery, and relapse. These changes can be characterized with deformation-based morphometry, which quantifies shape differences between anatomical structures, either in different subjects (cross-sectional) or in the same subject over time (longitudinal). Here we present analyses of image data from a longitudinal MRI study on the effects of alcoholism on brain structure. Images were acquired from alcoholic women (n=7, mean age 47.8 8.3 years) and age-matched control women (n=16, mean age 51.2 7.5 years). From each subject, we acquired two structural MR brain images, separated by approximately two years (mean 21.6 7 months). We performed two types of morphometry using log-Jacobian maps of inter-subject and intra-subject nonrigid coordinate transformations, justified by the invariance of relevant statistics (mean, standard deviation, z-score, and t-test) under changes of the spatial and temporal reference coordinate system. With all images from one time point, a cross-sectional inter-subject morphometry determined group differences between alcoholics and normal controls. We compared these results with longitudinal intra-subject morphometry based on two images per subject acquired at different times (approximately two years apart). Inter- and intra-subject analysis produced partially conflicting results. Whereas the intra-subject analysis indicated faster ventricular volume increases in the alcoholics (+11% per year) than in the controls (+2% per year), the inter-subject analysis showed, on average, smaller absolute ventricle volumes in the alcoholics than in the controls (-33% relative volume). These differences were confirmed by manual planimetry and were statistically significant whether tested based on difference or change, integrated over the volume of the ventricles. Other changes and group differences were consistent between the two analyses, e.g., reduction of white matter (including corpus callosum) and increase in CSF volume, and these are in agreement with established effects of alcoholism on brain structure. We conclude that intra-subject morphometry of longitudinal data is preferable to inter-subject morphometry for detecting dynamic changes due to a disease, especially when only small samples are available. Our analysis demonstrates that the distinction between group differences observed at a point in time vs. over time is not merely academic but can substantially reduce the validity of the outcomes of actual morphometric studies. This discrepancy in results underscores the importance of distinguishing between volume differences and volume changes in morphometric analyses.

Keywords: Deformation-based morphometry; Magnetic resonance images; Longitudinal analysis; Cross-sectional analysis; Alcoholism

The objective was the evaluation of single photon emission computed tomography (SPECT) with integrated low dose computed tomography (CT) in comparison with a retrospective fusion of SPECT and high-resolution CT and a side-by-side analysis for lesion localisation in patients with neuroendocrine tumours. Twenty-seven patients were examined by multidetector CT. Additionally, as part of somatostatin receptor scintigraphy (SRS), an integrated SPECT-CT was performed. SPECT and CT data were fused using software with a registration algorithm based on normalised mutual information. The reliability of the topographic assignment of lesions in SPECT-CT, retrospective fusion and side-by-side analysis was evaluated by two blinded readers. Two patients were not enrolled in the final analysis because of misregistrations in the retrospective fusion. Eighty-seven foci were included in the analysis. For the anatomical assignment of foci, SPECT-CT and retrospective fusion revealed overall accuracies of 91 and 94% (side-by-side analysis 86%). The correct identification of foci as lymph node manifestations (n=25) was more accurate by retrospective fusion (88%) than from SPECT-CT images (76%) or by side-by-side analysis (60%). Both modalities of image fusion appear to be well suited for the localisation of SRS foci and are superior to side-by-side analysis of non-fused images especially concerning lymph node manifestations.

Keywords: Image fusion; Single photon emission computed tomography; Computed tomography; Neuroendocrine tumour; Somatostatine receptor scintigraphy

The anatomical substrates of neural nets are usually composed from reconstructions of neurons that were stained in different preparations. Realistic models of the structural relationships between neurons require a common framework. Here we present 3-D reconstructions of single projection neurons (PN) connecting the antennal lobe (AL) with the mushroom body (MB) and lateral horn, groups of intrinsic mushroom body neurons (type 5 Kenyon cells), and a single mushroom body extrinsic neuron (PE1), aiming to compose components of the olfactory pathway in the honeybee. To do so, we constructed a digital standard atlas of the bee brain. The standard atlas was created as an average-shape atlas of 22 neuropils, calculated from 20 individual immunostained whole-mount bee brains. After correction for global size and positioning differences by repeatedly applying an intensity-based nonrigid registration algorithm, a sequence of average label images was created. The results were qualitatively evaluated by generating average gray-value images corresponding to the average label images and judging the level of detail within the labeled regions. We found that the first affine registration step in the sequence results in a blurred image because of considerable local shape differences. However, already the first nonrigid iteration in the sequence corrected for most of the shape differences among individuals, resulting in images rich in internal detail. A second iteration improved on that somewhat and was selected as the standard. Registering neurons from different preparations into the standard atlas reveals 1) that the m-ACT neuron occupies the entire glomerulus (cortex and core) and overlaps with a local interneuron in the cortical layer; 2) that, in the MB calyces and the lateral horn of the protocerebral lobe, the axon terminals of two identified m-ACT neurons arborize in separate but close areas of the neuropil; and 3) that MB-intrinsic clawed Kenyon cells (type 5), with somata outside the calycal cups, project to the peduncle and lobe output system of the MB and contact (proximate) the dendritic tree of the PE1 neuron at the base of the vertical lobe. Thus the standard atlas and the procedures applied for registration serve the function of creating realistic neuroanatomical models of parts of a neural net. The Honeybee Standard Brain is accessible at www.neurobiologie.fu-berlin.de/beebrain.

Keywords: virtual neuroanatomy; insect brain; brain atlas; brain reference system; mushroom bodies; antennal lobe; olfactory system

Accurate and fast localization of a pre-defined target region inside the patient is an important component of many image-guided therapy procedures. This problem is commonly solved by registration of intra-operative 2D projection images to 3D pre-operative images. If the patient is not fixed during the intervention, the 2D image acquisition is repeated several times during the procedure, and the registration problem can be cast instead as a 3D tracking problem. To solve the 3D problem, we propose in this paper to apply 2D region tracking to first recover the components of the transformation that are in-plane to the projections. The 2D motion estimates of all projections are backprojected into 3D space, where they are then combined into a consistent estimate of the 3D motion. We compare this method to intensity-based 2D-3D registration and a combination of 2D motion backprojection followed by a 2D-3D registration stage. Using clinical data with a fiducial marker-based gold-standard transformation, we show that our method is capable of accurately tracking vertebral targets in 3D from 2D motion measured in x-ray projection images. Using a standard tracking algorithm (hyperplane tracking), tracking is achieved at video frame rates but fails relatively often (32% of all frames tracked with target registration error (TRE) better than 1.2mm, 82% of all frames tracked with TRE better than 2.4mm). With intensity-based 2D-2D image registration using normalized mutual information (NMI) and pattern intensity (PI), accuracy and robustness are substantially improved. NMI tracked 82% of all frames in our data with TRE better than 1.2mm and 96% of all frames with TRE better than 2.4mm. This comes at the cost of a reduced frame rate, 1.7s average processing time per frame and projection device. Results using PI were slightly more accurate, but required on average 5.4s time per frame. These results are still substantially faster than 2D-3D registration. We conclude that motion backprojection from 2D motion tracking is an accurate and efficient method for tracking 3D target motion, but tracking 2D motion accurately and robustly remains a challenge.

Keywords: Real-time target tracking; 2D-3D registration; 2D-2D registration; motion backprojection; frameless stereotactic radiosurgery

Three different systematic approaches to generate multiple classifiers in atlas-based biomedical image segmentation are compared. Different atlases, as well as different parametrization of the registration algorithm, lead to different atlas-based classifiers. The classifier outputs are combined and compared to a manual ground truth segmentation. Classifier combination consistently improved classification accuracy with the biggest improvement from multiple atlases. We conclude that multi-classifier techniques have a natural application to atlas-based segmentation and increase classification accuracy in real-world segmentation problems.

Keywords: atlas-based segmentation; multiple classifier system; bagging; nonrigid registration

Computation of digitally reconstructed radiograph (DRR) images is the rate-limiting step in most current intensity-based algorithms for the registration of three-dimensional (3D) images to two-dimensional (2D) projection images. This paper introduces and evaluates the progressive attenuation field (PAF), which is a new method to speed up DRR computation. A PAF is closely related to an attenuation field (AF). A major difference is that a PAF is constructed on the fly as the registration proceeds; it does not require any precomputation time, nor does it make any prior assumptions of the patient pose or limit the permissible range of patient motion. A PAF effectively acts as a cache memory for projection values once they are computed, rather than as a lookup table for precomputed projections like standard AFs. We use a cylindrical attenuation field parameterization, which is better suited for many medical applications of 2D-3D registration than the usual two-plane parameterization. The computed attenuation values are stored in a hash table for time-efficient storage and access. Using clinical gold-standard spine image data sets from five patients, we demonstrate consistent speedups of intensity-based 2D-3D image registration using PAF DRRs by a factor of ten over conventional ray casting DRRs with no decrease of registration accuracy or robustness.

Keywords: 2D-3D Image Registration; Digitally Reconstructed Radiograph; DRR; Attenuation Field; Precomputation; Computational Performance

Confocal microscopy (CM) is a powerful image acquisition technique that is well established in many biological applications. It provides three-dimensional acquisition with high spatial resolution and can acquire several different channels of complementary image information. Due to the specimen extraction and preparation process, however, the shapes of imaged objects may differ considerably from their in vivo appearance. Magnetic resonance microscopy (MRM) is an evolving variant of magnetic resonance imaging, which achieves microscopic resolutions using a high magnetic field and strong magnetic gradients. Compared to CM imaging, MRM allows for in situ imaging and is virtually free of geometrical distortions. We propose in this paper to combine the advantages of both methods by unwarping CM images using a MRM reference image. Our method incorporates a sequence of image processing operators applied to the MRM image, followed by a two-stage intensity-based registration to compute a nonrigid coordinate transformation between the CM images and the MRM image. We present results obtained using CM images from the brains of 20 honey bees and a MRM image of an in situ bee brain.

Keywords: Confocal microscopy imaging; magnetic resonance microscopy; nonrigid image registration; geometrical distortion

RATIONALE AND OBJECTIVES. The 2D-3D registration of a CT image to one or more x-ray projection images has a number of image-guided therapy applications. Fiducial marker-based methods are in general fast, accurate, and robust, but marker implantation is not always possible, is often considered too invasive to be clinically acceptable, and entails risk. There is also the unresolved issue of whether it is acceptable to leave markers permanently implanted. Intensity-based registration methods do not require the use of markers and can be automated because geometric features such as points and surfaces do not need to be segmented from the images. For spine images, however, intensity-based methods are susceptible to local optima in the cost function and thus need initial transformations that are close to the correct transformation.

MATERIALS AND METHODS. In this paper, we propose a hybrid similarity measure for 2D-3D registration that is a weighted combination of an intensity-based similarity measure (mutual information) and a point-based measure using one fiducial marker. We evaluate its registration accuracy and robustness using gold-standard clinical spine image data from four patients.

RESULTS. The mean registration error for successful registrations over the four patients was 1.3 and 1.1 mm for the intensity-based and hybrid similarity measures, respectively. Whereas the percentage of successful intensity-based registrations (registration error less than 2.5 mm) decreased rapidly as the initial transformation got further from the correct transformation, the incorporation of a single marker produced successful registrations more than 99% of the time independent of the initial transformation.

CONCLUSION. The use of one fiducial marker reduces 2D-3D spine image registration error slightly and improves robustness substantially. The findings are potentially relevant for image-guided therapy. If one marker is sufficient to obtain clinically acceptable registration accuracy and robustness, as the preliminary results using the proposed hybrid similarity measure suggest, the marker can be placed on a spinous process, which could be accomplished without penetrating muscle or using fluoroscopic guidance, and such a marker could be removed relatively easily.

Keywords: 2D-3D image registration; fiducial marker; image-guided surgery; radiosurgery

Generation of digitally reconstructed radiographs (DRRs) is computationally expensive and is typically the rate-limiting step in the execution time of intensity-based 2D-3D registration algorithms. We address this computational issue by extending the technique of light field rendering from the computer graphics community. The extension of light fields, which we call attenuation fields (AFs), allows most of the DRR computation to be performed in a preprocessing step; after this precomputation step, DRRs can be generated substantially faster than with conventional ray casting. We derive expressions for the physical sizes of the two planes of an AF necessary to generate DRRs for a given X-ray camera geometry and all possible object motion within a specified range. An AF can require substantial memory, which we address by compressing it using vector quantization. We compare DRRs generated using AFs (AF-DRRs) to those generated using ray casting (RC-DRRs) for a typical C-arm geometry and CT images of several anatomic regions. They are quantitatively very similar: the median peak signal-to-noise ratio of AF-DRRs vs. RC-DRRs is greater than 43dB in all cases. We perform intensity-based 2D-3D registration using AF-DRRs and RC-DRRs and evaluate registration accuracy using gold-standard clinical spine image data from four patients. The registration accuracy and robustness of the two methods is virtually identical whereas the execution speed using AF-DRRs is an order of magnitude faster.

Keywords: Digitally reconstructed radiographs; light fields; intensity-based 2D-3D image registration; image-guided therapy

This study evaluated the use of image fusion in the preoperative staging of neuroendocrine tumors (NET) of the pancreas and the gastrointestinal tract (GIT). Thirty-eight patients suffering from a metastasized NET with location of the primary in the pancreas (n=15) or the GIT (n=23) were examined by somatostatin receptor scintigraphy (SRS) and computed tomography (CT). Consecutive image registration and fusion were performed using custom-built software integrated in AVS/Express (Advanced Visual Systems, Waltham, MA, USA). Registration was performed by a voxel-based algorithm based on normalized mutual information. Image fusion was feasible in 36/38 patients. A total of 87 foci were assigned to anatomical regions (e.g. gut, pancreas, liver, lymph node or others) by two independent observers in both SRS and SRS/CT fusion images. The assignments used a binary ranking system (1=definite, 0=not definite). These results were then retrospectively compared to the classification of the foci, based on postoperative histology or clinical follow-up. Imaging by SRS allowed a definite anatomical assignment in 57% (50/87) and 61% (53/87) of all lesions in the case of observers A and B, respectively. Image fusion improved the topographic assignment to 91% (79/87) and to 93% (81/87). The number classified as definite by both observers increased from 54% (47/87) to 86% (77/87). The increase in definite assignments was highly significant for both observers (P<0.0001 for each). In the case of foci classified as liver metastases, image fusion allowed improved assignment to the corresponding liver segment from 45% (18/40) to 98% (39/40) and from 58% (23/40) to 100% (40/40) by observers A and B, respectively. Furthermore, the improved assignment of foci classified as lesions by image fusion was relevant for therapy in 7/36 patients (19%). Therefore, the image fusion technique presented herein appears to be a very useful method for clinical routine.

Keywords: Neuroendocrine tumors; Somatostatin receptor scintigraphy; Computed tomography; Image registration; Image fusion

Background: This study describes a novel computer-generated anatomic symmetry plane as a framework for the quantitative description of sphenoid sinus anatomy. The aim of this study was to (1) determine relationships and distances between a midline sphenoid reference point (called the central sphenoid point [CSP]) and lateral sphenoid wall structures and (2) assess the incidence of anterior clinoid process (ACP) pneumatization and pterygoid recess (PR) pneumatization.

Methods: Axial computed tomography (CT) scans (1-mm slice thickness) were obtained on a VolumeZoom CT scanner (Siemens Medical, Erlangen, Germany). Mathematically derived anatomic symmetry planes were created using custom postprocessing software. A standardized review of each CT scan using surgical planning software (CBYON Suite version 2.6; CBYON, Mountain View, CA) was performed. The CSP was defined as a reference point in the midline sagittal plane at the intersection of the vertical sellar face and the horizontal sellar floor.

Results: A total of 128 sides in 64 cadaveric specimens were available for review. The incidences of ACP pneumatization and PR pneumatization were 23.4 and 37.5%. The mean distances from the CSP to the left optic canal midpoint, the left ACP entrance point, and the left PR lateral wall were 17.2, 15.6, and 27.6 mm, respectively. The corresponding distances from the CSP on the right side were 17.3, 15.8, and 28.0 mm, respectively. Measurements from the maxillary spine to the optic canal midpoint, ACP entrance point, and PR lateral wall on each side were performed also.

Conclusion: This approach provides both quantitative and qualitative understanding of sphenoid osteology and may be coupled with intraoperative surgical navigation to reduce the risks of sphenoid surgery. Both PR and ACP pneumatization are surprisingly common. Because the CSP-derived relationships may be referenced during endoscopic surgical navigation, they may provide greater clinical utility than traditional alternatives. This paradigm may facilitate a greater understanding of sphenoid anatomy and enhance surgical safety and precision.

Keywords: symmetry; ENT

PURPOSE: To evaluate the feasibility and the clinical benefits of retrospective digital image fusion (PET, SPECT, CT and MRI). MATERIALS AND METHODS: In a prospective study, a total of 273 image fusions were performed and evaluated. The underlying image acquisitions (CT, MRI, SPECT and PET) were performed in a way appropriate for the respective clinical question and anatomical region. Image fusion was executed with a software program developed during this study. The results of the image fusion procedure were evaluated in terms of technical feasibility, clinical objective, and therapeutic impact. RESULTS: The most frequent combinations of modalities were CT/PET (n = 156) and MRI/PET (n = 59), followed by MRI/SPECT (n = 28), CT/SPECT (n = 22) and CT/MRI (n = 8). The clinical questions included following regions (more than one region per case possible): neurocranium (n = 42), neck (n = 13), lung and mediastinum (n = 24), abdomen (n = 181), and pelvis (n = 65). In 92.6% of all cases (n = 253), image fusion was technically successful. Image fusion was able to improve sensitivity and specificity of the single modality, or to add important diagnostic information. Image fusion was problematic in cases of different body positions between the two imaging modalities or different positions of mobile organs. In 37.9% of the cases, image fusion added clinically relevant information compared to the single modality. CONCLUSION: For clinical questions concerning liver, pancreas, rectum, neck, or neurocranium, image fusion is a reliable method suitable for routine clinical application. Organ motion still limits its feasibility and routine use in other areas (e. g., thorax).

Keywords: Retrospective image fusion; CT; MRI; PET; SPECT; differential diagnosis

Differential diagnosis of pancreatic lesions still remains a problem. Whereas CT provides high spatial resolution, PET detects malignant lesions with high sensitivity. The objective of this study was to evaluate the clinical benefit of PET/CT image fusion in the diagnostic workup of pancreatic cancer. Methods: One hundred four patients with suspected pancreatic lesion underwent triple-phase multidetector CT and 18F-FDG PET scanning. Voxel-based retrospective registration and fusion of CT and PET were performed with recently developed software. CT, PET, and fused images were assessed by 2 radiologists with regard to the detection of malignancies, possible infiltration of adjacent tissue or lymph nodes, or distant metastases. Results: Fusion of CT and PET images was technically successful in 96.2%. In 2 cases, paraaortic lymph node infiltration was detected only by image fusion; in a further 8 cases, lymph node metastases were confirmed with improved localization. In 5 patients, additional pancreatic tumors or distant metastases only suspected during PET scanning were confirmed. Image fusion improved the sensitivity of malignancy detection from 76.6% (CT) and 84.4% (PET) to 89.1% (image fusion). Compared with CT alone, image fusion increased the sensitivity of detecting tissue infiltration to 68.2%, but at the cost of decreased specificity. Conclusion: The most important supplementary finding supplied by image fusion is a more precise correlation with focal tracer hot spots in PET. Image fusion improved the sensitivity of differentiating between benign and malignant pancreatic lesions with no significant change in specificity. All image modalities failed to stage lymph node involvement.

Keywords: PET; CT, spiral; image manipulation or reconstruction; pancreas; computer applications, detection

This paper evaluates strategies for atlas selection in atlas-based segmentation of three-dimensional biomedical images. Segmentation by intensity-based non-rigid registration to atlas images is applied to confocal microscopy images acquired from the brains of 20 bees. This paper evaluates and compares four different approaches for atlas image selection: registration to an individual atlas image (IND), registration to an average shape atlas image (AVG), registration to the most similar image from a database of individual atlas images (SIM), and registration to all images from a database of individual atlas images with subsequent multi-classifier decision fusion (MUL). The MUL strategy is a novel application of multi-classifier techniques, which are common in pattern recognition, to atlas-based segmentation. For each atlas selection strategy, the segmentation performance of the algorithm was quantified by means of the similarity index between the automatic segmentation result and a manually generated gold standard. The best segmentation accuracy was achieved using the MUL paradigm, which resulted in a mean similarity index value between manual and automatic segmentation of 0.86 (AVG, 0.84; SIM, 0.82; IND, 0.81). The superiority of the MUL paradigm is statistically significant (two-sided paired t-test, P<0.001). Our findings show that atlas selection is an important issue in atlas-based segmentation and that in particular multi-classifier techniques can substantially increase the segmentation accuracy.

Keywords: Atlas-based segmentation; Atlas selection; Non-rigid image registration; Bee brain; Confocal microscopy imaging

We present a technique for modeling liver motion during the respiratory cycle using intensity-based non-rigid registration of gated magnetic resonance (MR) images. Three-dimensional MR images of the abdomens of four volunteers were acquired at end-inspiration, end-expiration, and eight time points in between using respiratory gating. The deformation fields between the images were computed using intensity-based rigid and non-rigid registration algorithms. Global motion is modeled by a rigid transformation while local motion is modeled by a free-form deformation based on B-splines. Much of the liver motion was cranial-caudal translation, and thus the rigid transformation captured much of the motion. However, there was still substantial residual deformation (approximately 10mm averaged over the entire liver in four volunteers, and 34mm at one place in the liver of one volunteer). The computed organ motion model can potentially be used to determine an appropriate respiratory-gated radiotherapy window during which the position of the target is known within a specified excursion.

Keywords: non-rigid image registration; free-form spline-based deformation; liver motion; respiration-gated MR images

It is well known in the pattern recognition community that the accuracy of classifications obtained by combining decisions made by independent classifiers can be substantially higher than the accuracy of the individual classifiers. We have previously shown this to be true for atlas-based segmentation of biomedical images. The conventional method for combining individual classifiers weights each classifier equally (vote or sum rule fusion). In this paper, we propose two methods that estimate the performances of the individual classifiers and combine the individual classifiers by weighting them according to their estimated performance. The two methods are multiclass extensions of an expectation-maximization (EM) algorithm for ground truth estimation of binary classification based on decisions of multiple experts (Warfield , 2004). The first method performs parameter estimation independently for each class with a subsequent integration step. The second method considers all classes simultaneously. We demonstrate the efficacy of these performance-based fusion methods by applying them to atlas-based segmentations of three-dimensional confocal microscopy images of bee brains. In atlas-based image segmentation, multiple classifiers arise naturally by applying different registration methods to the same atlas, or the same registration method to different atlases, or both. We perform a validation study designed to quantify the success of classifier combination methods in atlas-based segmentation. By applying random deformations, a given ground truth atlas is transformed into multiple segmentations that could result from imperfect registrations of an image to multiple atlas images. In a second evaluation study, multiple actual atlas-based segmentations are combined and their accuracies computed by comparing them to a manual segmentation. We demonstrate in both evaluation studies that segmentations produced by combining multiple individual registration-based segmentations are more accurate for the two classifier fusion methods we propose, which weight the individual classifiers according to their EM-based performance estimates, than for simple sum rule fusion, which weights each classifier equally.

Keywords: Atlas-based segmentation; classifier performance; expectation-maximization (EM) parameter estimation; mixture of experts; multiclassifier decision fusion

AIM: Minimally invasive resection of hyperfunctional parathyroid glands is an alternative to open surgery. However, it requires a precise preoperative localization. This study evaluated the diagnostic use of magnetic resonance (MR) imaging, parathyroid scintigraphy, and consecutive image fusion. PATIENTS, METHODS: 17 patients (9 women, 8 men; age: 29-72 years; mean: 51.2 years) with primary hyperparathyroidism were included. Examination by MRI used unenhanced T1- and T2-weighted sequences as well as contrast-enhanced T1-weighted sequences. (99m)Tc-MIBI scintigraphy consisted of planar and SPECT (single photon emission tomography) imaging techniques. In order to improve the anatomical localization of a scintigraphic focus, SPECT-data were fused with the corresponding MR-data using a modified version of the Express 5.0 software (Advanced Visual Systems, Waltham, MA). Results of image fusion were then compared to histopathology. RESULTS: In 14/17 patients, a single parathyroid adenoma was found. There were 3 cases with hyperplastic glands. MRI detected 10 (71%), scintigraphy 12 (86%) adenomas. Both modalities detected 1/3 patients with hyperplasia. Image fusion improved the anatomical assignment of the 13 scintigraphic foci in five patients and was helpful in the interpretation of inconclusive MR-findings in two patients. CONCLUSIONS: Both MRI and (99m)Tc-MIBI scintigraphy sensitively detect parathyroid adenomas but are less reliable in case of hyperplastic glands. In case of a scintigraphic focus, image fusion considerably improves its topographic assignment. Furthermore, it facilitates the evaluation of inconclusive MRI findings.

Keywords: Parathyroid gland, adenoma, scintigraphy, magnetic resonance imaging, image fusion

One major problem with nonrigid image registration techniques is their high computational cost. Because of this, these methods have found limited application to clinical situations where fast execution is required, e.g., intra-operative imaging. This paper presents a parallel implementation of a nonrigid image registration algorithm. It takes advantage of shared-memory multiprocessor computer architectures using multithreaded programming by partitioning of data and partitioning of tasks, depending on the computational subproblem. For three different biomedical applications (intra-operative brain deformation, contrast-enhanced MR mammography, inter-subject brain registration), the scaling behavior of the algorithm is quantitatively analyzed. The method is demonstrated to perform the computation of intra-operative brain deformation in less than a minute using 64 CPUs on a 128-CPU shared-memory supercomputer (SGI Origin 3800). It is shown that its serial component is no more than 2 percent of the total computation time, allowing a speedup of at least a factor of 50. In most cases, the theoretical limit of the speedup is substantially higher (up to 132-fold in the application examples presented in this paper). The parallel implementation of our algorithm is therefore capable of solving nonrigid registration problems with short execution time requirements and may be considered an important step in the application of such techniques to clinically important problems such as the computation of brain deformation during cranial image-guided surgery.

Keywords: nonrigid image registration; high-performance computing; multithreaded computations; parallel performance; intra-operative brain deformation; contrast-enhanced MR mammography; motion correction; inter-subject registration; brain atlas

In this paper, we extend a previously reported intensity-based non-rigid registration algorithm by using a novel regularization term to constrain the deformation. Global motion is modeled by a rigid transformation while local motion is described by a free-form deformation based on B-splines. An information theoretic measure, normalized mutual information, is used as an intensity-based image similarity measure. Registration is performed by searching for the deformation that minimizes a cost function consisting of a weighted combination of the image similarity measure and a regularization term. The novel regularization term is a local volume-preservation (incompressibility) constraint, which is motivated by the assumption that soft tissue is incompressible for small deformations and short time periods. The incompressibility constraint is implemented by penalizing deviations of the Jacobian determinant of the deformation from unity. We apply the non-rigid registration algorithm with and without the incompressibility constraint to pre- and post-contrast MR breast images from 17 patients. Without using a constraint, the volume of contrast-enhancing lesions decreases by 1 to 78% (mean 26%). Image improvement (motion artifact reduction) obtained using the new constraint is compared to that obtained using a smoothness constraint based on the bending energy of the coordinate grid by blinded visual assessment of maximum intensity projections of subtraction images. For both constraints, volume preservation improves, and motion artifact correction worsens, as the weight of the constraint penalty term increases. For a given volume change of the contrast-enhancing lesions (2% of the original volume), the incompressibility constraint reduces motion artifacts better than or equal to the smoothness constraint in 13 out of 17 cases (better in 9, equal in 4, worse in 4). The preliminary results suggest that incorporation of the incompressibility regularization term improves intensity-based free-form non-rigid registration of contrast-enhanced MR breast images by greatly reducing the problem of shrinkage of contrast-enhancing structures while simultaneously allowing motion artifacts to be substantially reduced.

Keywords: Non-rigid registration; free-form deformation; B-splines; mutual information; volume preservation; contrast-enhanced MR breast images

OBJECTIVE: Despite the growing popularity of frameless image-guided surgery systems, stereotactic frame systems are widely accepted by neurosurgeons and are commonly used to perform biopsies, functional procedures, and stereotactic radiosurgery. We investigated the accuracy of the Brown-Roberts-Wells stereotactic frame system when the mechanical load on the frame changes between preoperative imaging and the intervention because of different patient position: supine during imaging, prone during intervention.

METHODS: We analyzed computed tomographic images acquired from 14 patients who underwent stereotactic biopsy, deep brain stimulator implantation, or radiosurgery. Two images were acquired for each patient, one with the patient in the supine position and one in the prone position. The prone images were registered to the respective supine images by use of an intensity-based registration algorithm, once using only the frame and once using only the head. The difference between the transformations produced by these two registrations describes the movement of the patient's head with respect to the frame.

RESULTS: The maximum frame-based registration error between the supine and prone positions was 2.8 mm; it was more than 2 mm in two patients and more than 1.5 mm in six patients. Anteroposterior translation is the dominant component of the difference transformation for most patients. In general, the magnitude of the movement increased with brain volume, which is an index of head weight.

CONCLUSION: To minimize frame-based registration error caused by a change in the mechanical load on the frame, stereotactic procedures should be performed with the patient in the identical position during imaging and intervention.

Keywords: Image-guided surgery; Mechanical stress; Prone position; Stereotactic head frame system; Stereotactive radiosurgery; Stereotactic techniques; Stereotaxy; Supine position

The authors have evaluated eight different similarity measures used for rigid body registration of serial magnetic resonance (MR) brain scans. To assess their accuracy the authors used 33 clinical three-dimensional (3-D) serial MR images, with deformable extradural tissue excluded by manual segmentation and simulated 3-D MR images with added intensity distortion. For each measure the authors determined the consistency of registration transformations for both sets of segmented and unsegmented data. They have shown that of the eight measures tested, the ones based on joint entropy produced the best consistency. In particular, these measures seemed to be least sensitive to the presence of extradural tissue. For these data the difference in accuracy of these joint entropy measures, with or without brain segmentation, was within the threshold of visually detectable change in the difference images.

Keywords: biomedical MRI; brain; entropy; image registration; image segmentation; medical image processing; 3-D serial MR brain image registration; MRI; brain segmentation; extradural tissue; joint entropy; magnetic resonance imaging; medical diagnostic imaging; transformation estimates consistency; voxel similarity measures

Aim: Catheter-placement in liver metastases is difficult when the lesions are not visible on plain images. We evaluated the use of image fusion using CT and PET data, providing information on anatomy and liver lesions, respectively. Methods: Plain CT of the liver and whole body FDG-PET were performed in 28 patients with colorectal carcinoma in preparation for laser-induced thermotherapy. Fusions of image data and 2D-visualisation were performed and evaluated with regard to quality of the registration, number of detected lesions and influence of the procedure on laser-induced thermotherapy. Image fusion was performed using an algorithm which was developed by our group. Results: 84 focal liver lesions were visible on CT, 107 on PET images. This means that CT guided catheter placement would have to be performed "blindly" in 23 lesions. In 6 patients previously unknown extra-hepatic tumor deposits were seen on PET images (local recurrence, 2 extrahepatic metastases, 2 second primaries). Conclusion: Hot spots from metastases, which are difficult to locate on PET images alone, may be identified on images generated by fusion of CT and PET data. These images facilitate an interventional approach to liver lesions which are not visible on plain CT. As PET allows for whole body screening, it helps identifying patients who would not profit from laser-induced thermotherapy.

Keywords: Image fusion - Laser-induced interstitial thermotherapy (LITT) - Computed tomography (CT) - Magnetic resonance imaging (MRI) - Positron emission tomography (PET) - Liver metastases

OBJECTIVE: A procedure for acquisition, automated registration, and fusion of functional and anatomical magnetic resonance images is presented. Its accuracy is quantitatively assessed using a publicly available gold standard. A patient case is used to illustrate the technique's clinical usefulness in image-guided neurosurgery.

MATERIALS AND METHODS: Before and after functional MRI (fMRI) acquisition, additional anatomical images were acquired at spatial locations identical to those of the functional images (5-10 slices) for the purpose of voxel-based image registration. Registration accuracy of the anatomical volumes and high-resolution 3D MRI volumes (MP-RAGE imaging) was quantified using adapted data (8 patients) originating from the Vanderbilt Retrospective Registration Evaluation Project (NIH project 1 R01 NS33926-02). Selecting three subsets of slices from that data (5 slices/6 mm slice distance, 10 slices/3 mm distance, and 10 slices/6 mm distance), the small number of images available from fMRI acquisition was taken into account. Accuracies in registering these sparse data sets were then compared to the accuracy achieved using complete data. For clinical patient data (16 patients), fMRI images were fused with MP-RAGE images, thereby integrating anatomical images with information about the locations of functional areas. The resulting images were used for planning and navigation during tumor resections using an operating microscope (MKM, Zeiss).

RESULTS: Quantitative analysis showed no loss of registration accuracy due to a reduced number of slices, regardless of whether 5 or 10 slices were used. For small-volume coverage in the anatomical images (thickness 24 mm), registration of one patient failed, and this could easily be identified by visual inspection. No failures were experienced when 54 mm was covered. In the clinical environment, all 16 interventions using fused fMRI and MRI data were successful.

CONCLUSIONS: Automatic registration of functional and high-resolution anatomical MRI was found to be sufficiently accurate and reliable for use in stereotactic neurosurgery.

Keywords: functional MRI; MP-RAGE; registration accuracy; image fusion