|
|
|
| e-Pub |
Previous | 
Home
Section: New Results
Vascular Imaging and Arterial Spin Labelling
Patient-specific detection of perfusion abnormalities combining within-subject and between-subject variances in Arterial Spin Labeling.
Participants : Camille Maumet, Pierre Maurel, Jean-Christophe Ferré, Christian Barillot.
In this paper, patient-specific perfusion abnormalities in Arterial Spin Labeling (ASL) were identified by comparing a single patient to a group of healthy controls using a mixed-effect hierarchical General Linear Model (GLM). Two approaches are currently in use to solve hierarchical GLMs: (1) the homoscedastic approach assumes homogeneous variances across subjects and (2) the heteroscedastic approach is theoretically more efficient in the presence of heterogeneous variances but algorithmically more demanding. In practice, in functional magnetic resonance imaging studies, the superiority of the heteroscedastic approach is still under debate. Due to the low signal-to-noise ratio of ASL sequences, within-subject variances have a significant impact on the estimated perfusion maps and the heteroscedastic model might be better suited in this context. In this paper we studied how the homoscedastic and heteroscedastic approaches behave in terms of specificity and sensitivity in the detection of patient-specific ASL perfusion abnormalities. Validation was undertaken on a dataset of 25 patients diagnosed with brain tumors and 36 healthy volunteers. We showed evidence of heterogeneous within-subject variances in ASL and pointed out an increased false positive rate of the homoscedastic model. In the detection of patient-specific brain perfusion abnormalities with ASL, modeling heterogeneous variances increases the sensitivity at the same specificity level [24] .
An a contrario approach for the detection of activated brain areas in fMRI
Participants : Camille Maumet, Pierre Maurel, Jean-Christophe Ferré, Christian Barillot.
BOLD functional MRI (fMRI) is now a widespread imaging technique to study task-related activity in the brain. However, getting the areas of activation at the individual subject level is still an open issue. The standard massively univariate statistical analysis is usually performed after smoothing the data and makes use of a single p-value for final thresholding of the results [1]. In group fMRI studies, the need for compensation of cross-subjects misregistrations clearly justifies the smoothing. However, at the individual level, where neat delineations of the activated areas are of interest, the use of Gaussian smoothing as a pre-processing step is more questionable. In this paper, we propose to study the ability of an a contrario approach, recently adapted for basal perfusion abnormalities detection [2], to correctly detect areas of functional activity [42] .
Robust perfusion maps in Arterial Spin Labeling by means of M-estimators
Participants : Camille Maumet, Pierre Maurel, Jean-Christophe Ferré, Christian Barillot.
Non-invasive measurement of Cerebral Blood Flow (CBF) is now feasible thanks to the introduction of Arterial Spin Labeling (ASL) Magnetic Resonance Imaging (MRI) techniques. To date, due to the low signal-to-noise ratio of ASL, a single acquisition (pair of control/label scans) is not sufficient to estimate perfusion reliably. Instead, the acquisition is usually repeated several times and the perfusion information is calculated by averaging across the repetitions. However, due to its zero breakdown point, the sample mean is very sensitive to outliers. In this paper, we propose to compute ASL CBF maps using Huber's M-estimator, a robust statistical function that is not overly impacted by outliers. This method is compared to an empirical approach, introduced in [1], based on z-score thresholding [43] .
Quantifying CBF from Aterial Spin Labeling via Diverse-TI: sampling diversity or repetitions ?
Participants : Lei Yu, Pierre Maurel, Christian Barillot.
Arterial Spin Labeling (ASL) is a noninvasive perfusion technique which allows the absolute quantification of Cerebral Blood Flow (CBF). The perfusion is obtained from the difference between images with and without magnetic spin labeling of the arterial blood and the captured signal is around 0.5-2% of the magnitude of the labeling images, so the noise is one of the main problems for further data analysis. Classical method, Mono-TI, for CBF quantification is averaging repetitions with only one Inversion Time (TI) - the time delay between labeling and acquisition to allow the labeled blood to arrive the imaging slice. It improves the robustness to noise, however, cannot compensate the variety of Arterial Arrival Time (AAT). In this paper, Diverse-TI is proposed to exploit different TI sampling instants (sampling diversity) to improve the robustness to variety of AAT and simultaneously average repetitions with each TI (sampling repetitions) to improve the robustness to noise. Generally, the sampling diversity is relatively small and can be considered as compressed measurements, thus the Compressive Matched Filter (CMF) enlightened from sparsity is exploited to directly reconstruct CBF and AAT directly from compressed measurements. Meanwhile, regarding the CBF quantification performance, the compromise between the sampling repetition and sampling diversity is discussed and the empirical protocol to determine the sampling diversity is proposed. Simulations are carried out to highlight our discussions. This is a joint work with Remi Gribonval (Panama Team) [56] .
Peripheral angiography and neurovascular imaging
Participants : Hélène Raoult, Jean-Yves Gauvrit, Elise Bannier, Pierre Maurel, Clement Neyton, Christian Barillot, Jean-Christophe Ferré.
Vascular imaging contributions were performed on two different regions during the evaluation period: first on peripheral angiography, then on neurovascular imaging. Arteriography and MR angiography are routinely performed in patients presenting vascular pathologies. Yet, contrast agent injection is contraindicated in patients with renal insufficiency and the underlying risk of developing nephrogenic systemic fibrosis further encourages research on non-contrast enhanced MR angiography techniques (NCE MRA). In this context, we have been working on new MR sequences to reliably detect vascular abnormalities.
A first study [29] was published, where we assessed the feasibility and image quality of an improved non-gated carotid NATIVE TrueFISP NCE MRA sequence providing an extended field of view and a shorter acquisition time as compared to Time-of-Flight (TOF) imaging. A second study [48] was recently accepted for publication in Radiology on intracranial NCE MRA for arteriovenous malformation imaging with a high temporal resolution over 2 cardiac cycles. Combined with image post-processing, it allows improved depiction of venous drainage necessary to evaluate hemorrhagic risk and quantification. This ongoing work was just submitted.