S. Cavassila - Academia.edu (original) (raw)
Papers by S. Cavassila
Synopsis A time-domain quantitation algorithm based on a metabolite basis set obtained by quantum... more Synopsis A time-domain quantitation algorithm based on a metabolite basis set obtained by quantum mechanical simulation is proposed. This non-linear least squares algorithm fits a time domain model function, combination of (quantum-mechanically simulated) metabolite signals, to low-SNR in vivo data. The metabolite basis set was created with NMR-SCOPE which can handle various experimental protocols. The present work investigates through Monte Carlo studies the ability of the algorithm to quantify strongly overlapping spectral components in presence of (residual) water and a macromolecule spectrum. Quantitation of short echo-time 1 H human brain signals at 1.5T is demonstrated as well as quantitation of 31 P signals. Method Fitting of model functions to low-SNR in vivo data with strongly overlapping peaks needs invocation of ever more prior knowledge about the model parameters. The fit can be performed in the frequency [1] or time [2] domain using measured spectra of selected metabolite solutions as numerical model functions. Alternatively, one can compute theoretical metabolite signals/spectra quantum-mechanically for the measurement protocol used by a scanner and fit in the frequency domain [3]. The proposed algorithm QUEST (QUantitation based on QUantum ESTimation) fits a combination of (quantum-mechanically simulated) signals of metabolites directly to the in vivo data at hand, in the time-domain domain. QUEST is based on • A non-linear least squares algorithm aiming at finding the model parameters that minimize the distance between the raw signal and the model function. This algorithm allows to automatically compensate for distortions due to the magnetic field heterogeneities with the ideal signals of the metabolite basis set. This has been done by using small extra damping factors and frequency shifts in the fit procedure. • A metabolite basis set. Signals of the metabolites were computed by quantum mechanics with NMR-SCOPE [4] using the spin Hamiltonian parameters given in [5]. NMR-SCOPE, based on the product-operator formalism, can handle various NMR pulse sequences. As for preprocessing, QUALITY deconvolution is applied first (if reference signal available), then water suppression using HLSVD [6] and macromolecule removal by weighting/truncation [7] or correction [8] of initial samples. The quantitation errors are estimated by computing the Cramér-Rao lower bounds. Note, that our CRBs are too small because we have not included the water/macromolecules in the model function. Results QUEST performances are assessed through Monte-Carlo studies, see Fig.1. Then, in vivo 1 H short echo-time signals of human brain at 1.5T obtained with STEAM were quantified with QUEST, see Fig.2. Signals of aspartate (Asp), choline (Cho), GABA, glucose, glutamate (Glu), glutamine (Gln), lactate (Lac), myo-inositol (Ins), Nacetylaspartate (NAA), phosphocreatine (PCr), creatine (Cr), taurine (Tau), plus signals modelling the lipids at 0.9 and 1.3 ppm were included in the QUEST fits. Quantitation of a 31 P signal is shown in Fig.3. .
2008 IEEE International Workshop on Imaging Systems and Techniques, 2008
We apply the Morlet wavelet transform to characterizing Magnetic Resonance Spectroscopy (MRS) sig... more We apply the Morlet wavelet transform to characterizing Magnetic Resonance Spectroscopy (MRS) signals acquired at short echo-time. These signals usually contain contributions from metabolites, water and a baseline which mainly originates from large molecules, known as macromolecules, and lipids. The baseline accommodation is one of the major obstructions in in vivo short echo-time MRS quantification as its shape and intensity are not known a priori. In this paper, the simulated signal of the N-acetylaspartate (NAA) metabolite is used as a test signal to be recovered after adding the in vivo macromolecular signal. The in vivo macromolecule MRS signal was acquired on a horizontal 4.7T Biospec system. By optimizing the inversion time, which represents the delay between the inversion pulse and the first pulse of the PRESS sequence, the metabolites are nullified while the others are maintained. The metabolite-nullified signal from a volume-of-interest centralized in the hippocampus of a healthy mouse, which was a combination of residual water, baseline and noise, was added to the signal of NAA. The amplitude of the metabolite is also varied to visualize the sensitivity of the wavelet transform at different ratios between the intensity of the macromolecular and the metabolite signals. Compared to the simulated signal of NAA, the signal decays much faster. The timescale representation of the wavelet can therefore distinguish the two signals without any additional pre-processing. The amplitude of the metabolite is also correctly derived although at earlier time it still has an effect of the baseline.
IFMBE Proceedings, 2009
We study the Morlet wavelet transform on characterizing Magnetic Resonance Spectroscopy (MRS) sig... more We study the Morlet wavelet transform on characterizing Magnetic Resonance Spectroscopy (MRS) signals acquired at short echo-time. These MRS signals usually contain contributions from metabolites, water and a baseline which mainly originates from large molecules, known as macromolecules, and lipids. As its shape and intensity are not known a priori, the baseline accommodation becomes one of the major obstructions in in vivo short echo-time MRS quantification. We acquired an in vivo macromolecule MRS signal on a horizontal 4.7T Biospec system by optimizing the inversion time, which represents the delay between the inversion pulse and the first pulse of the PRESS sequence. As a consequence, the metabolites are nullified while the others are maintained. The metabolite-nullified signal from a volume-of-interest centralized in the hippocampus of a healthy mouse was a combination of residual water, baseline and noise. Compared to the simulated signal of creatine, the signal decays much faster. The timescale representation of the wavelet can therefore distinguish the two signals without any additional pre-processing. The amplitude of the metabolite is also correctly derived although at earlier time it still has an effect of the baseline. In addition, we also show that the Morlet wavelet can be used to characterize different lineshapes, e.g. Lorentzian, Gaussian or Voigt, which are generally used to model the MRS signals. That is, the first derivative of the modulus of the wavelet transform relates to the damping effect of the Lorentzian lineshape while its second derivative indicates the second-order broadening of the Gaussian and Voigt. The performance of the wavelet when applied to an in vitro creatine is also presented.
2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro, 2008
In the field of small animal imaging, the interest for phased array coil imaging is growing but h... more In the field of small animal imaging, the interest for phased array coil imaging is growing but high field MR experimental systems with multiple receiver channels are still rare and the upgrade of existing systems is relatively expensive. In this work, a standard 4.7 T Bruker Biospec Avance II spectrometer was modified to allow simultaneous two-channel acquisitions. Modifications were validated on imaging and spectroscopy on metabolite solution phantom as well as on mice brain using a home-made two-channel array coil operating at 200.3 MHz. A dedicated two-channel array coil with two square elements encompassing the mouse brain was designed and built. Compared to a singlechannel surface coil, the mean SNR measured on images in the ROI corresponding to whole mouse brain was improved by about 30% as well as the signal uniformity. For spectroscopic acquisition, the SNR gain in a voxel located close to the coils was improved by about 65%. Modifications realized for proton multiple-channel acquisitions could also be applied for any X-nucleus. Compared to quadrature detection coils, two-channel coils offer the ability to use parallel acquisitions techniques.
NMR in biomedicine, 2001
In order to keep subscribers up-to-date with the latest developments in their field, John Wiley &... more In order to keep subscribers up-to-date with the latest developments in their field, John Wiley & Sons are providing a current awareness service in each issue of the journal. The bibliography contains newly published material in the field of NMR in biomedicine. Each bibliography is divided into 9 sections: 1 Books, Reviews ' Symposia; 2 General; 3 Technology; 4 Brain and Nerves; 5 Neuropathology; 6 Cancer; 7 Cardiac, Vascular and Respiratory Systems; 8 Liver, Kidney and Other Organs; 9 Muscle and Orthopaedic. Within each section, articles are listed in alphabetical order with respect to author. If, in the preceding period, no publications are located relevant to any one of these headings, that section will be omitted.
Journal of magnetic resonance. Series B, 1996
A new 3D Fourier imaging method based on sparse radial scanning (SRS-FT) of k space is proposed. ... more A new 3D Fourier imaging method based on sparse radial scanning (SRS-FT) of k space is proposed. It allows acquisition of FIDs and is therefore well suited to imaging objects with very short T2. Use of a Bayesian procedure allows (1) an important reduction of scan time to below that of the projection-reconstruction (PR) method by reducing the number of "Cartesian radial" encoding directions, and (2) a good image quality by estimating missing and corrupted Cartesian samples. SRS-FT images reconstructed from FIDs are compared to conventional FT and PR images.
This study concerns parameterization and sub- sequent subtraction of the fast decaying signals of... more This study concerns parameterization and sub- sequent subtraction of the fast decaying signals of macro- molecules from in vivo MRS signals of tissue metabolites of interest. The parameterization is done with a State Space approach (HSVD) based on singular value decomposition. The method is tested with a simulated non-exponentially damped macromolecule signal and an exponentially damped metabolite signal. and is
Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 1997
ABSTRACT The authors present a new 3D MRI scan method that combines favourable properties of Cart... more ABSTRACT The authors present a new 3D MRI scan method that combines favourable properties of Cartesian and radial sampling. Cartesian sampling requires only Fourier transformation (FT) to obtain an image. Radial sampling enables imaging of objects with very short transverse relaxation times. The authors constrain radial sampling such that coincidence with a Cartesian grid be maximal. Missing Cartesian samples are estimated with a Bayesian procedure based on FT and general prior knowledge. A substantial reduction of scan time is achieved with respect to conventional radial scanning. Application to a `phantom' is shown
2005 IEEE Engineering in Medicine and Biology 27th Annual Conference, 2005
In vivo<sup>1</sup>H short echo-time Magnetic Resonance spectra are m... more In vivo<sup>1</sup>H short echo-time Magnetic Resonance spectra are made up of overlapping spectral components from many metabolites. Typically, they exibit low signal-to-noise ratio. Metabolite concentrations are obtained by quantitating such spectra. Quantitation is difficult due to the superposition of metabolite resonances, macromolecules, lipids and water residue contributions. A fitting algorithm invoking extensive prior knowledge is needed. We quantitated<sup>1</sup>H in vivo mouse brain spectra obtained at 7 Tesla using the time-domain QUEST method combined with in vitro metabolite basis set signals. Brain metabolite concentrations estimated from eight mouse brain signals are compared to previously reported results.
2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro, 2008
In vivo hepatic 1H lineshapes modeled by the complex Voigt function are desirable to reduce syste... more In vivo hepatic 1H lineshapes modeled by the complex Voigt function are desirable to reduce systematic error and obtain accurate fits. However, the optimization procedure becomes challenging when the peak resonances overlap and the proportion of Gaussian to Lorentzian dampings is a priori unknown. In this context, nonlinear least-squares algorithms generally invoked in Magnetic Resonance Spectroscopy quantification are highly sensitive
Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 1997
ABSTRACT Magnetic resonance spectroscopy (MRS) offers a wealth of information to the biochemist o... more ABSTRACT Magnetic resonance spectroscopy (MRS) offers a wealth of information to the biochemist or radiologist. Metabolite concentrations, J-couplings, pH, ion concentrations and gradients, temperature, etc., can all be obtained, in situ, from well-defined volumes in the human body, and in a totally non-invasive way. However, simple methods such as peak area integration or automatic line fitting in the FT MR spectrum are still relied on for routine MRS data analysis. The disadvantages of such methods are tolerated in order to keep processing fast and simple for the spectroscopist. The authors have developed a graphical user interface, in which advanced time domain signal processing methods are combined. They present a complete software package for routine MR data analysis, called MRUI, enabling the use of advanced parameter estimation algorithms with incorporation of prior knowledge via simple menus and spectral displays, in a fashion similar to the spectroscopist's spectrometer software
Localized brain proton spectroscopy can non invasively provide biochemical information from disti... more Localized brain proton spectroscopy can non invasively provide biochemical information from distinct regions of the brain that can be used for disease detection, disease progression monitoring and treatment. Quantitation of Magnetic Resonance Spectroscopy (MRS) signals acquired at short echo time is difficult due to the overlap of metabolite resonances, macromolecules and lipids and to the presence of the water residue. Moreover, the signals have low signal-to-noise ratio. A fitting algorithm invoking extensive prior knowledge is then needed. Metabolites were quantitated using the method QUEST which fits a combination of metabolites signals from a basis set to the in vivo data. The basis set can be obtained either by quantum mechanically simulating the theoretical metabolite signals, or by measuring signals of metabolite aqueous solutions. In this paper, we compare the influence of the basis set on the quantitation results. Short echo-time in vivo signals of rat brains were acquired...
Measurement Science and Technology, 2009
We apply the Morlet wavelet transform (MWT) for quantitatively analyzing proton magnetic resonanc... more We apply the Morlet wavelet transform (MWT) for quantitatively analyzing proton magnetic resonance spectroscopic (MRS) signals, more precisely signals acquired at short echo time. These signals contain many resonating components whose frequencies are characteristic of the observed metabolites, and amplitudes are directly related to the concentrations of these metabolites. With these powerful properties, in vivo MRS can be considered as a unique non-invasive tool to explore biochemical compounds of living tissues. However, the analysis and quantification of these metabolite contributions are difficult due to the low signal-to-noise ratio, the number of overlapping frequencies and the contamination of the signal of interest with water and a baseline originating from macromolecules and lipids. The baseline is a major obstacle for MRS quantification as its shape and intensity are generally not known a priori. In this paper, we present the methodology to quantify the signals by the MWT. We assess the ability of the proposed method to recover parameters such as metabolite amplitudes, frequencies and damping factors while facing successively quantification challenges arising from the non-Lorentzian lineshapes, overlapping frequencies, and noise or baseline. Tests of the method are performed on simulated signals alone or combined with either in vitro acquisition and/or in vivo macromolecular signal acquired on a horizontal 4.7 T scanner. In presence of the macromolecules, the amplitude parameter is correctly derived by the method, thanks to the timescale representation of the wavelet which enables us to distinguish the two signals by their time decays and without any additional pre-processing.
Synopsis A time-domain quantitation algorithm based on a metabolite basis set obtained by quantum... more Synopsis A time-domain quantitation algorithm based on a metabolite basis set obtained by quantum mechanical simulation is proposed. This non-linear least squares algorithm fits a time domain model function, combination of (quantum-mechanically simulated) metabolite signals, to low-SNR in vivo data. The metabolite basis set was created with NMR-SCOPE which can handle various experimental protocols. The present work investigates through Monte Carlo studies the ability of the algorithm to quantify strongly overlapping spectral components in presence of (residual) water and a macromolecule spectrum. Quantitation of short echo-time 1 H human brain signals at 1.5T is demonstrated as well as quantitation of 31 P signals. Method Fitting of model functions to low-SNR in vivo data with strongly overlapping peaks needs invocation of ever more prior knowledge about the model parameters. The fit can be performed in the frequency [1] or time [2] domain using measured spectra of selected metabolite solutions as numerical model functions. Alternatively, one can compute theoretical metabolite signals/spectra quantum-mechanically for the measurement protocol used by a scanner and fit in the frequency domain [3]. The proposed algorithm QUEST (QUantitation based on QUantum ESTimation) fits a combination of (quantum-mechanically simulated) signals of metabolites directly to the in vivo data at hand, in the time-domain domain. QUEST is based on • A non-linear least squares algorithm aiming at finding the model parameters that minimize the distance between the raw signal and the model function. This algorithm allows to automatically compensate for distortions due to the magnetic field heterogeneities with the ideal signals of the metabolite basis set. This has been done by using small extra damping factors and frequency shifts in the fit procedure. • A metabolite basis set. Signals of the metabolites were computed by quantum mechanics with NMR-SCOPE [4] using the spin Hamiltonian parameters given in [5]. NMR-SCOPE, based on the product-operator formalism, can handle various NMR pulse sequences. As for preprocessing, QUALITY deconvolution is applied first (if reference signal available), then water suppression using HLSVD [6] and macromolecule removal by weighting/truncation [7] or correction [8] of initial samples. The quantitation errors are estimated by computing the Cramér-Rao lower bounds. Note, that our CRBs are too small because we have not included the water/macromolecules in the model function. Results QUEST performances are assessed through Monte-Carlo studies, see Fig.1. Then, in vivo 1 H short echo-time signals of human brain at 1.5T obtained with STEAM were quantified with QUEST, see Fig.2. Signals of aspartate (Asp), choline (Cho), GABA, glucose, glutamate (Glu), glutamine (Gln), lactate (Lac), myo-inositol (Ins), Nacetylaspartate (NAA), phosphocreatine (PCr), creatine (Cr), taurine (Tau), plus signals modelling the lipids at 0.9 and 1.3 ppm were included in the QUEST fits. Quantitation of a 31 P signal is shown in Fig.3. .
2008 IEEE International Workshop on Imaging Systems and Techniques, 2008
We apply the Morlet wavelet transform to characterizing Magnetic Resonance Spectroscopy (MRS) sig... more We apply the Morlet wavelet transform to characterizing Magnetic Resonance Spectroscopy (MRS) signals acquired at short echo-time. These signals usually contain contributions from metabolites, water and a baseline which mainly originates from large molecules, known as macromolecules, and lipids. The baseline accommodation is one of the major obstructions in in vivo short echo-time MRS quantification as its shape and intensity are not known a priori. In this paper, the simulated signal of the N-acetylaspartate (NAA) metabolite is used as a test signal to be recovered after adding the in vivo macromolecular signal. The in vivo macromolecule MRS signal was acquired on a horizontal 4.7T Biospec system. By optimizing the inversion time, which represents the delay between the inversion pulse and the first pulse of the PRESS sequence, the metabolites are nullified while the others are maintained. The metabolite-nullified signal from a volume-of-interest centralized in the hippocampus of a healthy mouse, which was a combination of residual water, baseline and noise, was added to the signal of NAA. The amplitude of the metabolite is also varied to visualize the sensitivity of the wavelet transform at different ratios between the intensity of the macromolecular and the metabolite signals. Compared to the simulated signal of NAA, the signal decays much faster. The timescale representation of the wavelet can therefore distinguish the two signals without any additional pre-processing. The amplitude of the metabolite is also correctly derived although at earlier time it still has an effect of the baseline.
IFMBE Proceedings, 2009
We study the Morlet wavelet transform on characterizing Magnetic Resonance Spectroscopy (MRS) sig... more We study the Morlet wavelet transform on characterizing Magnetic Resonance Spectroscopy (MRS) signals acquired at short echo-time. These MRS signals usually contain contributions from metabolites, water and a baseline which mainly originates from large molecules, known as macromolecules, and lipids. As its shape and intensity are not known a priori, the baseline accommodation becomes one of the major obstructions in in vivo short echo-time MRS quantification. We acquired an in vivo macromolecule MRS signal on a horizontal 4.7T Biospec system by optimizing the inversion time, which represents the delay between the inversion pulse and the first pulse of the PRESS sequence. As a consequence, the metabolites are nullified while the others are maintained. The metabolite-nullified signal from a volume-of-interest centralized in the hippocampus of a healthy mouse was a combination of residual water, baseline and noise. Compared to the simulated signal of creatine, the signal decays much faster. The timescale representation of the wavelet can therefore distinguish the two signals without any additional pre-processing. The amplitude of the metabolite is also correctly derived although at earlier time it still has an effect of the baseline. In addition, we also show that the Morlet wavelet can be used to characterize different lineshapes, e.g. Lorentzian, Gaussian or Voigt, which are generally used to model the MRS signals. That is, the first derivative of the modulus of the wavelet transform relates to the damping effect of the Lorentzian lineshape while its second derivative indicates the second-order broadening of the Gaussian and Voigt. The performance of the wavelet when applied to an in vitro creatine is also presented.
2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro, 2008
In the field of small animal imaging, the interest for phased array coil imaging is growing but h... more In the field of small animal imaging, the interest for phased array coil imaging is growing but high field MR experimental systems with multiple receiver channels are still rare and the upgrade of existing systems is relatively expensive. In this work, a standard 4.7 T Bruker Biospec Avance II spectrometer was modified to allow simultaneous two-channel acquisitions. Modifications were validated on imaging and spectroscopy on metabolite solution phantom as well as on mice brain using a home-made two-channel array coil operating at 200.3 MHz. A dedicated two-channel array coil with two square elements encompassing the mouse brain was designed and built. Compared to a singlechannel surface coil, the mean SNR measured on images in the ROI corresponding to whole mouse brain was improved by about 30% as well as the signal uniformity. For spectroscopic acquisition, the SNR gain in a voxel located close to the coils was improved by about 65%. Modifications realized for proton multiple-channel acquisitions could also be applied for any X-nucleus. Compared to quadrature detection coils, two-channel coils offer the ability to use parallel acquisitions techniques.
NMR in biomedicine, 2001
In order to keep subscribers up-to-date with the latest developments in their field, John Wiley &... more In order to keep subscribers up-to-date with the latest developments in their field, John Wiley & Sons are providing a current awareness service in each issue of the journal. The bibliography contains newly published material in the field of NMR in biomedicine. Each bibliography is divided into 9 sections: 1 Books, Reviews ' Symposia; 2 General; 3 Technology; 4 Brain and Nerves; 5 Neuropathology; 6 Cancer; 7 Cardiac, Vascular and Respiratory Systems; 8 Liver, Kidney and Other Organs; 9 Muscle and Orthopaedic. Within each section, articles are listed in alphabetical order with respect to author. If, in the preceding period, no publications are located relevant to any one of these headings, that section will be omitted.
Journal of magnetic resonance. Series B, 1996
A new 3D Fourier imaging method based on sparse radial scanning (SRS-FT) of k space is proposed. ... more A new 3D Fourier imaging method based on sparse radial scanning (SRS-FT) of k space is proposed. It allows acquisition of FIDs and is therefore well suited to imaging objects with very short T2. Use of a Bayesian procedure allows (1) an important reduction of scan time to below that of the projection-reconstruction (PR) method by reducing the number of "Cartesian radial" encoding directions, and (2) a good image quality by estimating missing and corrupted Cartesian samples. SRS-FT images reconstructed from FIDs are compared to conventional FT and PR images.
This study concerns parameterization and sub- sequent subtraction of the fast decaying signals of... more This study concerns parameterization and sub- sequent subtraction of the fast decaying signals of macro- molecules from in vivo MRS signals of tissue metabolites of interest. The parameterization is done with a State Space approach (HSVD) based on singular value decomposition. The method is tested with a simulated non-exponentially damped macromolecule signal and an exponentially damped metabolite signal. and is
Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 1997
ABSTRACT The authors present a new 3D MRI scan method that combines favourable properties of Cart... more ABSTRACT The authors present a new 3D MRI scan method that combines favourable properties of Cartesian and radial sampling. Cartesian sampling requires only Fourier transformation (FT) to obtain an image. Radial sampling enables imaging of objects with very short transverse relaxation times. The authors constrain radial sampling such that coincidence with a Cartesian grid be maximal. Missing Cartesian samples are estimated with a Bayesian procedure based on FT and general prior knowledge. A substantial reduction of scan time is achieved with respect to conventional radial scanning. Application to a `phantom' is shown
2005 IEEE Engineering in Medicine and Biology 27th Annual Conference, 2005
In vivo<sup>1</sup>H short echo-time Magnetic Resonance spectra are m... more In vivo<sup>1</sup>H short echo-time Magnetic Resonance spectra are made up of overlapping spectral components from many metabolites. Typically, they exibit low signal-to-noise ratio. Metabolite concentrations are obtained by quantitating such spectra. Quantitation is difficult due to the superposition of metabolite resonances, macromolecules, lipids and water residue contributions. A fitting algorithm invoking extensive prior knowledge is needed. We quantitated<sup>1</sup>H in vivo mouse brain spectra obtained at 7 Tesla using the time-domain QUEST method combined with in vitro metabolite basis set signals. Brain metabolite concentrations estimated from eight mouse brain signals are compared to previously reported results.
2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro, 2008
In vivo hepatic 1H lineshapes modeled by the complex Voigt function are desirable to reduce syste... more In vivo hepatic 1H lineshapes modeled by the complex Voigt function are desirable to reduce systematic error and obtain accurate fits. However, the optimization procedure becomes challenging when the peak resonances overlap and the proportion of Gaussian to Lorentzian dampings is a priori unknown. In this context, nonlinear least-squares algorithms generally invoked in Magnetic Resonance Spectroscopy quantification are highly sensitive
Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 1997
ABSTRACT Magnetic resonance spectroscopy (MRS) offers a wealth of information to the biochemist o... more ABSTRACT Magnetic resonance spectroscopy (MRS) offers a wealth of information to the biochemist or radiologist. Metabolite concentrations, J-couplings, pH, ion concentrations and gradients, temperature, etc., can all be obtained, in situ, from well-defined volumes in the human body, and in a totally non-invasive way. However, simple methods such as peak area integration or automatic line fitting in the FT MR spectrum are still relied on for routine MRS data analysis. The disadvantages of such methods are tolerated in order to keep processing fast and simple for the spectroscopist. The authors have developed a graphical user interface, in which advanced time domain signal processing methods are combined. They present a complete software package for routine MR data analysis, called MRUI, enabling the use of advanced parameter estimation algorithms with incorporation of prior knowledge via simple menus and spectral displays, in a fashion similar to the spectroscopist's spectrometer software
Localized brain proton spectroscopy can non invasively provide biochemical information from disti... more Localized brain proton spectroscopy can non invasively provide biochemical information from distinct regions of the brain that can be used for disease detection, disease progression monitoring and treatment. Quantitation of Magnetic Resonance Spectroscopy (MRS) signals acquired at short echo time is difficult due to the overlap of metabolite resonances, macromolecules and lipids and to the presence of the water residue. Moreover, the signals have low signal-to-noise ratio. A fitting algorithm invoking extensive prior knowledge is then needed. Metabolites were quantitated using the method QUEST which fits a combination of metabolites signals from a basis set to the in vivo data. The basis set can be obtained either by quantum mechanically simulating the theoretical metabolite signals, or by measuring signals of metabolite aqueous solutions. In this paper, we compare the influence of the basis set on the quantitation results. Short echo-time in vivo signals of rat brains were acquired...
Measurement Science and Technology, 2009
We apply the Morlet wavelet transform (MWT) for quantitatively analyzing proton magnetic resonanc... more We apply the Morlet wavelet transform (MWT) for quantitatively analyzing proton magnetic resonance spectroscopic (MRS) signals, more precisely signals acquired at short echo time. These signals contain many resonating components whose frequencies are characteristic of the observed metabolites, and amplitudes are directly related to the concentrations of these metabolites. With these powerful properties, in vivo MRS can be considered as a unique non-invasive tool to explore biochemical compounds of living tissues. However, the analysis and quantification of these metabolite contributions are difficult due to the low signal-to-noise ratio, the number of overlapping frequencies and the contamination of the signal of interest with water and a baseline originating from macromolecules and lipids. The baseline is a major obstacle for MRS quantification as its shape and intensity are generally not known a priori. In this paper, we present the methodology to quantify the signals by the MWT. We assess the ability of the proposed method to recover parameters such as metabolite amplitudes, frequencies and damping factors while facing successively quantification challenges arising from the non-Lorentzian lineshapes, overlapping frequencies, and noise or baseline. Tests of the method are performed on simulated signals alone or combined with either in vitro acquisition and/or in vivo macromolecular signal acquired on a horizontal 4.7 T scanner. In presence of the macromolecules, the amplitude parameter is correctly derived by the method, thanks to the timescale representation of the wavelet which enables us to distinguish the two signals by their time decays and without any additional pre-processing.