Johannes Slotboom

The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)

In this paper we report the set-up and results of the Multimodal Brain Tumor Image Segmentation Benchmark (BRATS) organized in conjunction with the MICCAI 2012 and 2013 conferences. Twenty state-of-the-art tumor segmentation algorithms were applied to a set of 65 multi-contrast MR scans of low- and high-grade glioma patients - manually annotated by up to four raters - and to 65 comparable scans generated using tumor image simulation software. Quantitative evaluations revealed considerable disagreement between the human raters in segmenting various tumor sub-regions (Dice scores in the range 74%-85%), illustrating the difficulty of this task. We found that different algorithms worked best for different sub-regions (reaching performance comparable to human inter-rater variability), but that no single algorithm ranked in the top for all sub-regions simultaneously. Fusing several good algorithms using a hierarchical majority vote yielded segmentations that consistently ranked above all individual algorithms, indicating remaining opportunities for further methodological improvements. The BRATS image data and manual annotations continue to be publicly available through an online evaluation system as an ongoing benchmarking resource.

Observation of intramyocellular lipids by means of 1H magnetic resonance spectroscopy.

Magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS) are being increasingly used for investigations of human muscle physiology. While MRI reveals the morphology of muscles in great detail (e.g. for the determination of muscle volumes), MRS provides information on the chemical composition of the tissue. Depending on the observed nucleus, MRS allows the monitoring of high-energy phosphates (31P MRS), glycogen (13C MRS), or intramyocellular lipids (1H MRS), to give only a few examples. The observation of intramyocellular lipids (IMCL) by means of 1H MRS is non-invasive and, therefore, can be repeated many times and with a high temporal resolution. MRS has the potential to replace the biopsy for the monitoring of IMCL levels; however, the biopsy still has the advantage that other methods such as those used in molecular biology can be applied to the sample. The present study describes variations in the IMCL levels (expressed in mmol/kg wet weight and ml/100 ml) in three different muscles before and after (0, 1, 2, and 5 d) marathon runs for a well-trained individual who followed two different recovery protocols varying mainly in the diet. It was shown that the repletion of IMCL levels is strongly dependent on the diet post exercise. The monitoring of IMCL levels by means of 1H MRS is extremely promising, but several methodological limitations and pitfalls need to be considered, and these are addressed in the present review.

Characterization of the macromolecule baseline in localized 1H-MR spectra of human brain

Short‐echo‐time magnetic resonance spectra of human brain contain broad contributions from macromolecules. As they are a priori of unknown shape and intensity, they pose a problem if one wants to quantitate the overlying spectral features from low‐molecular‐weight metabolites. On the other hand, the macromolecular contributions may provide relevant clinical information themselves, if properly evaluated. Several methods, based on T1, T2, or spectral shape, have previously been suggested to suppress or edit the macromolecule contributions. Here, a method is presented based on a series of saturation recovery scans and that allows for simultaneous recording of the macromolecular baseline and the fully relaxed metabolite spectrum. In comparison to an inversion recovery technique aimed at nulling signals from long‐T1 components, the saturation recovery method is less susceptible to T1 differences inherent in signals from different metabolites or introduced by pathology. The saturation recovery method was used to quantitate the macromolecular baseline in white and/or gray matter locations of the human brain in 40 subjects. It was found that the content and composition of MR visible macromolecules depends on cerebral location, as well as the age of the investigated subject, while no gender dependence could be found. Magn Reson Med 46:855–863, 2001.

Quantitative 1H-magnetic resonance spectroscopy of human brain: Influence of composition and parameterization of the basis set in linear combination model-fitting

Localized short‐echo‐time 1H‐MR spectra of human brain contain contributions of many low‐molecular‐weight metabolites and baseline contributions of macromolecules. Two approaches to model such spectra are compared and the data acquisition sequence, optimized for reproducibility, is presented. Modeling relies on prior knowledge constraints and linear combination of metabolite spectra. Investigated was what can be gained by basis parameterization, i.e., description of basis spectra as sums of parametric lineshapes. Effects of basis composition and addition of experimentally measured macromolecular baselines were investigated also. Both fitting methods yielded quantitatively similar values, model deviations, error estimates, and reproducibility in the evaluation of 64 spectra of human gray and white matter from 40 subjects. Major advantages of parameterized basis functions are the possibilities to evaluate fitting parameters separately, to treat subgroup spectra as independent moieties, and to incorporate deviations from straightforward metabolite models. It was found that most of the 22 basis metabolites used may provide meaningful data when comparing patient cohorts. In individual spectra, sums of closely related metabolites are often more meaningful. Inclusion of a macromolecular basis component leads to relatively small, but significantly different tissue content for most metabolites. It provides a means to quantitate baseline contributions that may contain crucial clinical information. Magn Reson Med 48:440–453, 2002.

Factors that determine penumbral tissue loss in acute ischaemic stroke

The goal of acute stroke treatment with intravenous thrombolysis or endovascular recanalization techniques is to rescue the penumbral tissue. Therefore, knowing the factors that influence the loss of penumbral tissue is of major interest. In this study we aimed to identify factors that determine the evolution of the penumbra in patients with proximal (M1 or M2) middle cerebral artery occlusion. Among these factors collaterals as seen on angiography were of special interest. Forty-four patients were included in this analysis. They had all received endovascular therapy and at least minimal reperfusion was achieved. Their penumbra was assessed with perfusion- and diffusion-weighted imaging. Perfusion-weighted imaging volumes were defined by circular singular value decomposition deconvolution maps (Tmax > 6 s) and results were compared with volumes obtained with non-deconvolved maps (time to peak > 4 s). Loss of penumbral volume was defined as difference of post- minus pretreatment diffusion-weighted imaging volumes and calculated in per cent of pretreatment penumbral volume. Correlations between baseline characteristics, reperfusion, collaterals, time to reperfusion and penumbral volume loss were assessed using analysis of covariance. Collaterals (P = 0.021), reperfusion (P = 0.003) and their interaction (P = 0.031) independently influenced penumbral tissue loss, but not time from magnetic resonance (P = 0.254) or from symptom onset (P = 0.360) to reperfusion. Good collaterals markedly slowed down and reduced the penumbra loss: in patients with thrombolysis in cerebral infarction 2 b-3 reperfusion and without any haemorrhage, 27% of the penumbra was lost with 8.9 ml/h with grade 0 collaterals, whereas 11% with 3.4 ml/h were lost with grade 1 collaterals. With grade 2 collaterals the penumbral volume change was -2% with -1.5 ml/h, indicating an overall diffusion-weighted imaging lesion reversal. We conclude that collaterals and reperfusion are the main factors determining loss of penumbral tissue in patients with middle cerebral artery occlusions. Collaterals markedly reduce and slow down penumbra loss. In patients with good collaterals, time to successful reperfusion accounts only for a minor fraction of penumbra loss. These results support the hypothesis that good collaterals extend the time window for acute stroke treatment.

Examining the gateway to the limbic system with diffusion tensor imaging: the perforant pathway in dementia.

Current treatments for Alzheimers disease (AD) are only able to slow the progression of mental deterioration, making early and reliable diagnosis an essential part of any promising therapeutic strategy. In the initial stages of AD, the first neuropathological alterations occur in the perforant pathway (PP), a large neuronal fiber tract located at the entrance to the limbic system. However, to date, there is no sensitive diagnostic tool for performing in vivo assessments of this structure. In the present bimodal magnetic resonance imaging (MRI) study, we examined 10 elderly controls, 10 subjects suffering from mild cognitive impairment (MCI), and 10 AD patients in order to evaluate the sensitivity of diffusion tensor imaging (DTI), a new MRI technique, for detecting changes in the PP. Furthermore, the diagnostic explanatory power of DTI data of the PP should be compared to high-resolution MRI volumetry and intervoxel coherences (COH) of the hippocampus and the entorhinal cortex, two limbic regions also involved in the pathophysiology of early AD. DTI revealed a marked decrease in COH values in the PP region of MCI (right side: 26%, left side: 29%, as compared to controls) and AD patients (right side: 37%, left side: 43%, as compared to controls). Reductions in COH values of the PP region were significantly correlated with cognitive impairment. DTI data of the PP zone were the only parameter differing significantly between control subjects and MCI patients, while the volumetric measures and the COH values of the hippocampus and the entorhinal cortex did not. DTI of medial temporal brain regions is a promising non-invasive tool for the in vivo diagnosis of the early/preclinical stages of AD.

Restoration of electrophysiological signals distorted by inductive effects of magnetic field gradients during MR sequences.

A generally applicable method for almost complete suppression of signal artifacts on electrophysiological signals caused by B0‐gradient switching (gradient noise) is presented. The method is demonstrated for electrocardiograms (ECGs) but can also be used for other electrophysiological signals. It takes advantage of the fact that under certain conditions, the effect of switching the B0‐field gradient upon an electrophysiological signal can be modeled as a linear time‐invariant system and fully characterized by pulse response functions. It is shown how the systems pulse response functions of the X, Y, and Z gradients can be determined and how gradient noise can be eliminated efficiently. The elimination of gradient noise by the proposed method causes in the current arrangement a constant delay of 128 msec, which is acceptable for patient monitoring and magnetic resonance sequence triggering. Magn Reson Med 41:715–721, 1999.

Mechanical Thrombectomy for Acute Ischemic Stroke Thrombus–Device Interaction, Efficiency, and Complications In Vivo

Background and Purpose— Mechanical thrombectomy is a promising new modality of interventional stroke treatment. The various devices differ with regard to where they apply force on the thrombus, taking a proximal approach such as aspiration devices or a distal approach such as basket-like devices. The study compares the in vivo effectiveness and thrombus–device interaction of these 2 approaches. Methods— Angiography and embolization with a radioopaque whole blood thrombus was performed in 10 swine. Mechanical thrombectomy was performed in 20 cranial vessels using a proximal aspiration device (Vasco35) and a distal basket-like device (Catch) with and without proximal balloon occlusion. Fifty-six retrieval attempts were made. Results— The proximal device allowed fast repeated application with a low risk of thromboembolic events (3%) and vasospasm, but it had a significantly lower success rate (39.4%) in retrieving thrombotic material than the distal device (DD) (82.6%; odds ratio, 7.3; 95% CI, 2.0 to 26.4). The compaction of the thrombus during retrieval with DD increased the risk of vessel wall irritation significantly (P<0.01) and complicated retrieval into the guiding catheter. The number of embolic events was significantly higher with DD (26%; odds ratio, 11.3; 95% CI, 1.35 to 101.6) unless proximal balloon occlusion was used. Conclusions— The proximal and the distal approaches to mechanical thrombectomy proved to be effective at achieving recanalization of cranial vessels. The proximal device is faster in application and allowed repeated attempts with a low complication rate. The DD is more successful at removing thrombotic material, but its method of application and attendant thrombus compaction increase the risk of thromboembolic events and vasospasms.

Neonatal body composition : Dual-energy X-ray absorptiometry, magnetic resonance imaging, and three-dimensional chemical shift imaging versus chemical analysis in piglets

An animal study to evaluate dual-energy x-ray absorptiometry (DXA) and magnetic resonance (MR) imaging and spectroscopy for measurement of neonatal body composition was performed. Twenty-three piglets with body weights ranging from 848 to 7550 g were used. After measuring total body water, animals were killed and body composition was assessed using DXA and MR (1.5 T; MR imaging, T1-weighted sagittal spin-echo sequence; MR spectroscopy, three-dimensional chemical shift imaging) as well as chemical carcass analysis (standard methods) after homogenization. Body composition by chemical analysis (percent of body weight, mean ± SD) was as follows: body water, 75.3 ± 3.9%; total protein, 13.9 ± 8.8%; and total fat, 6.5 ± 3.7%. Absolute content of fat and total ash was 7–674 and 35–237 g, respectively. Mean hydration of fat-free mass was 0.804 ± 0.011 g/kg and decreased with increasing body weight (r2 = 0.419) independent of age. Using DXA, bone mineral content was highly correlated with calcium content (r2= 0.992), and calcium per bone mineral content was 44.1 ± 4.2%. DXA fat mass correlated with total fat (r2 = 0.961). Using MR, spectroscopy and chemical analysis were highly correlated with fat-to-water ratio (r2 = 0.984) and absolute fat content (r2 = 0.988). Total fat by MR imaging volumetry showed a lower correlation (r2 = 0.913) and overestimated total fat by a factor of 2.46. Conversion equations for DXA were developed (total fat = 1.31 × fat mass measured by DXA − 68.8; calcium = 0.402 × bone mineral content + 1.7), which improved precision and accuracy of DXA measurements. In conclusion, both DXA and MR spectroscopy give accurate and precise estimates of neonatal body composition and may become valuable tools for the noninvasive assessment of neonatal growth and nutritional status.

Methods and reproducibility of cardiac/respiratory double‐triggered 1H‐MR spectroscopy of the human heart

Localized 1H‐MR spectroscopy is sensitive to motion and has mostly been applied to the brain. For the human heart, cardiac and respiratory motion lead to displacements on the order of the localized voxel and lead to substantial variations of voxel content, lineshape, water suppression, and signal phase and amplitude. Combined respiratory and cardiac double triggering can avoid these complications to a large extent. Three methods of double triggering are evaluated, with reproducibility established in nine subjects for a method based on respiratory modulation of the ECG amplitude and a visual feedback mechanism. Quantitated with respect to water, within‐subject reproducibilities for this setup were 9% for trimethylammonium compounds, 10% for creatine/phosphocreatine, and 13% for lipids. ANOVA showed significant differences between subjects which may relate to natural variability between subjects or exact location within the heart. Unresolved issues for this technique are its susceptibility to precise placement of ECG electrodes and the reasons for failure in 20% of examination. With this technique it is possible to investigate open questions in cardiac pathophysiology, such as the creatine content in chronic heart disease. Variants of this triggering method may also improve cardiovascular MRI methods relying on data acquired in several heartbeats. Magn Reson Med 42:903–910, 1999.