Daniel Gounot

Automatic change detection in multimodal serial MRI: application to multiple sclerosis lesion evolution

The automatic analysis of subtle changes between MRI scans is an important tool for assessing disease evolution over time. Manual labeling of evolutions in 3D data sets is tedious and error prone. Automatic change detection, however, remains a challenging image processing problem. A variety of MRI artifacts introduce a wide range of unrepresentative changes between images, making standard change detection methods unreliable. In this study we describe an automatic image processing system that addresses these issues. Registration errors and undesired anatomical deformations are compensated using a versatile multiresolution deformable image matching method that preserves significant changes at a given scale. A nonlinear intensity normalization method is associated with statistical hypothesis test methods to provide reliable change detection. Multimodal data is optionally exploited to reduce the false detection rate. The performance of the system was evaluated on a large database of 3D multimodal, MR images of patients suffering from relapsing remitting multiple sclerosis (MS). The method was assessed using receiver operating characteristics (ROC) analysis, and validated in a protocol involving two neurologists. The automatic system outperforms the human expert, detecting many lesion evolutions that are missed by the expert, including small, subtle changes.

IN VIVO DETERMINATION OF MULTIEXPONENTIAL T2 RELAXATION IN THE BRAIN OF PATIENTS WITH MULTIPLE SCLEROSIS

In vivo measurement of T2 relaxation times in multiple sclerosis (MS) lesions by magnetic resonance imaging (MRI) is potentially useful for the evaluation of the disease activity. Seven patients with definite MS were investigated over a period of three years (19 examinations), using a whole-body MRI scanner operating at 0.15 T with a specially designed high-power radio-frequency head coil. A modified CPMG sequence with a 180 degree pulse interval of TE = 6 msec and 128 echoes was used for the T2 relaxation measurement of the areas of increased signal (AIS) and white matter (WM). A biexponential T2 analysis of each pixel of the spin-echo images was computed. The T2 relaxation processes were found to be a monoexponential function in WM. The T2 relaxation times of apparently normal white matter in MS patients was significantly longer than in control subjects. The T2 relaxation curves of the AIS were found in most cases to fit a biexponential function characterized by a short and a long T2. T2 long relaxation times of AIS were spread out over a wide range (150-560 msec). The study of T2 long histograms shows that some AIS can be divided into two or three parts depending on the T2 long values. Each of these parts may correspond to a pathological process such as edema, demyelination and gliosis. Evolution of T2 relaxation times over a period of time cannot as yet be correlated with modifications in the clinical state.

The BOLD response and the gamma oscillations respond differently than evoked potentials: an interleaved EEG-fMRI study

BackgroundThe integration of EEG and fMRI is attractive because of their complementary precision regarding time and space. But the relationship between the indirect hemodynamic fMRI signal and the more direct EEG signal is uncertain. Event-related EEG responses can be analyzed in two different ways, reflecting two different kinds of brain activity: evoked, i.e. phase-locked to the stimulus, such as evoked potentials, or induced, i.e. non phase-locked to the stimulus such as event-related oscillations. In order to determine which kind of EEG activity was more closely related with fMRI, EEG and fMRI signals were acquired together, while subjects were presented with two kinds of rare events intermingled with frequent distractors. Target events had to be signaled by pressing a button and Novel events had to be ignored.ResultsBoth Targets and Novels triggered a P300, of larger amplitude in the Novel condition. On the opposite, the fMRI BOLD response was stronger in the Target condition. EEG event-related oscillations in the gamma band (32–38 Hz) reacted in a way similar to the BOLD response.ConclusionsThe reasons for such opposite differential reactivity between oscillations / fMRI on the one hand, and evoked potentials on the other, are discussed in the paper. Those results provide further arguments for a closer relationship between fast oscillations and the BOLD signal, than between evoked potentials and the BOLD signal.

Where arousal meets attention: a simultaneous fMRI and EEG recording study

In this fMRI study, we looked for the regions supporting interaction between cortical arousal and attention during three conditions: detection, observation, and rest. Arousal measurements were obtained from the EEG low-frequency (LF) power (5-9.5 Hz) recorded continuously together with fMRI. Whatever the condition, arousal was positively correlated with the fMRI signal of the right dorsal-lateral prefrontal and superior parietal cortices, closely overlapping regions involved in the maintenance of attention. Although the inferior temporal areas also presented a correlation with arousal during detection, path analysis suggests that this influence may be indirect, through the top-down influence of the previously mentioned network. However, those visual-processing areas could account for the correlation between arousal and performances. Lastly, the medial frontal cortex, frontal opercula, and thalamus were inversely correlated with arousal but only during detection and observation so that they could account for the control of arousal.

Evaluation of RARE-MR urography in the assessment of ureterohydronephrosis

Objective The goal of this prospective study was to evaluate the value of the fast imaging sequence called RARE-MR urography (RMU) for the diagnosis of ureterohydronephrosis. Materials and Methods Sixty-nine patients underwent this procedure. The results were compared with those obtained by intravenous urography (IVU) and ultrasonography (US). Results The accuracy of RMU in the detection of urinary tract dilatation and the localization of the level of obstruction was excellent (100%). The determination of the type of obstruction, intrinsic versus extrinsic, was 80% by IVU and 60% for RMU. The RMU sequence alone could not specify the nature of the obstruction. Functional information about the obstructed collecting system could not be obtained. Conclusion The RMU technique may be considered in the following circumstances: contraindications to IVU (allergy to contrast medium, severe renal failure), impairment of renal excretion, and failure to locate the level of obstruction by US. The absence of ionizing radiations favors the promotion of this procedure to study ureterohydronephrosis during pregnancy.

The right parahippocampal gyrus contributes to the formation and maintenance of bound information in working memory.

Working memory is devoted to the temporary storage and on-line manipulation of information. Recently, an integrative system termed the episodic buffer has been proposed to integrate and hold information being entered or retrieved from episodic memory. Although the brain system supporting such an integrative buffer is still in debate, the medial temporal lobe appears to be a promising candidate for the maintenance of bound information. In the current work, binding was assessed by comparing two conditions in which participants had to retain three letters and three spatial locations presented either bound or separate. At the behavioral level, lower performance was found for bound information than for separate information. When contrasting the two conditions, activation in the right parahippocampal gyrus was greater for the encoding and maintenance of bound information. No activation was observed in the medial temporal lobe during the retrieval of bound information. Together, our results suggest that the parahippocampal gyrus may underlie the integrative and maintenance functions of the episodic buffer.

Phonological processing in relation to reading: An fMRI study in deaf readers

Without special education, early deprivation of auditory speech input, hinders the development of phonological representations and may alter the neural mechanisms of reading. By using fMRI during lexical and rhyming decision tasks, we compared in hearing and pre-lingually deaf subjects the neural activity in functional regions of interest (ROIs) engaged in reading. The results show in deaf readers significantly higher activation in the ROIs relevant to the grapho-phonological route, but also in the posterior medial frontal cortex (pMFC) and the right inferior frontal gyrus (IFG). These adjustments may be interpreted within the dual route model of reading as an alternative strategy, which gives priority to rule-based letter-to-sound conversion. Activation in the right IFG would account for compensation mechanisms based on phonological recoding and inner speech while activation in the posterior medial frontal cortex (pMFC) may relate to the cognitive effort called for by the alternative strategy. Our data suggest that the neural mechanisms of reading are shaped by the auditory experience of speech.

In vitro validation of MR measurements of arterial pulse-wave velocity in the presence of reflected waves.

A magnetic resonance imaging projective velocity encoding sequence was used to determine the pulse‐wave velocity in an artery model. To this end, a well‐defined flow phantom simulating flow propagation in large arteries was used. In order to validate the measurement method in the presence of large reflected waves, these were deliberately created in the phantom. The projective sequence was applied to two measurement sites and the wave velocity was determined from the spatial and temporal separations of the foot of the velocity waveform. A theoretical model describing reflection and attenuation phenomena was compared with experimental velocity waveforms. The model showed that reflections and attenuation can explain the important changes in velocity waveforms. The model also confirmed that in the presence of reflecting waves, the foot of the waveform can be used as a characteristic point for measurements through changes in the waveform. J. Magn. Reson. Imaging 2001;14:120–127.

Abnormal Medial Temporal Activity for Bound Information During Working Memory Maintenance in Patients With Schizophrenia

Alterations of binding in long‐term memory in schizophrenia are well established and occur as a result of aberrant activity in the medial temporal lobe (MTL). In working memory (WM), such a deficit is less clear and the pathophysiological bases remain unstudied. Seventeen patients with schizophrenia and 17 matched healthy controls performed a WM binding task while undergoing functional magnetic resonance imaging. Binding was assessed by contrasting two conditions comprising an equal amount of verbal and spatial information (i.e., three letters and three spatial locations), but differing in the absence or presence of a link between them. In healthy controls, MTL activation was observed for encoding and maintenance of bound information but not for its retrieval. Between‐group comparisons revealed that patients with schizophrenia showed MTL hypoactivation during the maintenance phase only. In addition, BOLD signals correlated with behavioral performance in controls but not in patients with schizophrenia. Our results confirm the major role that the MTL plays in the pathophysiology of schizophrenia. Short‐term and long‐term relational memory deficits in schizophrenia may share common cognitive and functional pathological bases. Our results provide additional information about the episodic buffer that represents an integrative interface between WM and long‐term memory.

Nuclear magnetic resonance T2 relaxation times in multiple sclerosis

An original method was used to carry out the mathematical analysis of T2 transverse magnetization decay curves and the measure of T2 relaxation times on multiple sclerosis (MS) patients. The presumably normal white matter (WM) of these patients presented higher T2 relaxation times (98.6 msec), in comparison with that found in a population sample (88 msec). In this case, magnetization decay curves remain mostly monoexponential and are characterized by a single T2. On the other hand, areas of increased signal (AIS) curves are always better fitted by a biexponential function characterized by a short (82 msec) and a long (greater than 200 msec) T2. The spreading out of long T2 varies from one AIS to another in the same patient and among different patients; values of long T2 also vary with time, but without any correlation with the clinical state. In fact, no correlation was been established between relaxation times and clinical parameters. Quantitative MRI therefore enables a different approach to interpret MRI images; results suggest that several histobiochemical parameters play a role in the pathogenesis of an AIS and that MS is a dynamic and constantly evolving disease.