T Laudadio

Nosologic imaging of the brain: segmentation and classification using MRI and MRSI.

A new technique is presented to create nosologic images of the brain based on magnetic resonance imaging (MRI) and magnetic resonance spectroscopic imaging (MRSI). A nosologic image summarizes the presence of different tissues and lesions in a single image by color coding each voxel or pixel according to the histopathological class it is assigned to. The proposed technique applies advanced methods from image processing as well as pattern recognition to segment and classify brain tumors. First, a registered brain atlas and a subject‐specific abnormal tissue prior, obtained from MRSI data, are used for the segmentation. Next, the detected abnormal tissue is classified based on supervised pattern recognition methods. Class probabilities are also calculated for the segmented abnormal region. Compared to previous approaches, the new framework is more flexible and able to better exploit spatial information leading to improved nosologic images. The combined scheme offers a new way to produce high‐resolution nosologic images, representing tumor heterogeneity and class probabilities, which may help clinicians in decision making. Copyright

Tissue segmentation and classification of MRSI data using canonical correlation analysis.

In this article an accurate and efficient technique for tissue typing is presented. The proposed technique is based on Canonical Correlation Analysis, a statistical method able to simultaneously exploit the spectral and spatial information characterizing the Magnetic Resonance Spectroscopic Imaging (MRSI) data. Recently, Canonical Correlation Analysis has been successfully applied to other types of biomedical data, such as functional MRI data. Here, Canonical Correlation Analysis is adapted for MRSI data processing in order to retrieve in an accurate and efficient way the possible tissue types that characterize the organ under investigation. The potential and limitations of the new technique have been investigated by using simulated as well as in vivo prostate MRSI data, and extensive studies demonstrate a high accuracy, robustness, and efficiency. Moreover, the performance of Canonical Correlation Analysis has been compared to that of ordinary correlation analysis. The test results show that Canonical Correlation Analysis performs best in terms of accuracy and robustness. Magn Reson Med, 2005.

Differentiation between Brain Metastasis and Glioblastoma using MRI and two-dimensional Turbo Spectroscopic Imaging data

In this paper we propose a novel technique to differentiate brain metastases from high-grade gliomas, which represent the most aggressive and common brain lesions. In spite of the significant progresses achieved in the field of MRI in the last decades, the differentiation between these two types of tumors is still a challenge as they show a similar appearance on MRI images, but require a completely different therapeutic treatment. Here, we show that such a differentiation is actually possible and can be obtained by making use of MRI as well as of two-dimensional Turbo Spectroscopic Imaging (2D-TSI) information. Specifically, the proposed technique consists of three steps: we first detect the abnormal region on the MRI by applying a new segmentation-classification algorithm able to produce nosologic images of the brain by exploiting both MRI and 2D-TSI information; afterwards, an accurate and efficient tissue typing technique, based on a statistical method known as canonical correlation analysis, is applied to 2D-TSI data in order to detect the lesion and characterize its nature; finally, the lesion contours provided by the aforementioned methods are compared in order to differentiate glioblastoma from metastasis lesions. The latter step exploits the infiltrative nature of glioblastoma, when compared to metastasis, which can be observed on magnetic resonance spectroscopic imaging (MRSI) but not on MRI. Indeed, our studies show that the tumor region detected in the second step for glioblastoma is significantly broader than the lesion obtained in the first step and observable on the MRI image. On the contrary, such a difference does not occur in the metastasis examples.

Application of the HLSVD technique to the filtering of X-ray diffraction data

A filter based on the Hankel-Lanczos singular value decomposition (HLSVD) technique is presented and applied for the first time to X-ray diffraction (XRD) data. Synthetic and real powder XRD intensity profiles of nanocrystals are used to study the filter performances with different noise levels. Results show the robustness of the HLSVD filter and its capability to extract easily and effciently the useful crystallographic information. These characteristics make the filter an interesting and user-friendly tool for processing of XRD data.

NMF in MR Spectroscopy

Nowadays, magnetic resonance spectroscopy (MRS) represents a powerful nuclear magnetic resonance (NMR) technique in oncology since it provides information on the biochemical profile of tissues, thereby allowing clinicians and radiologists to identify in a non-invasive way the different tissue types characterising the sample under investigation. The main purpose of the present chapter is to provide a review of the most recent and significant applications of non-negative matrix factorisation (NMF) to MRS data in the field of tissue typing methods for tumour diagnosis. Specifically, NMF-based methods for the recovery of constituent spectra in ex vivo and in vivo brain MRS data, for brain tissue pattern differentiation using magnetic resonance spectroscopic imaging (MRSI) data and for automatic detection and visualisation of prostate tumours, will be described. Furthermore, since several NMF implementations are available in the literature, a comparison in terms of pattern detection accuracy of some NMF algorithms will be reported and discussed, and the NMF performance for MRS data analysis will be compared with that of other blind source separation (BSS) techniques.

Nosologic Imaging of Brain Tumors Using MRI and MRSI

A new technique is presented to create nosologic images of the brain based on magnetic resonance imaging and magnetic resonance spectroscopic imaging (MRSI). A nosologic image summarizes the presence of different tissues and lesions in a single image by color coding each voxel or pixel according to the histopathological class it is assigned to. The proposed technique applies advanced methods from image processing as well as pattern recognition to segment and classify brain tumors. First, a registered brain atlas and a subject-specific abnormal tissue prior, obtained from MRSI data, are used for the segmentation. Next, the detected abnormal tissue is classified based on supervised pattern recognition methods. Class probabilities are also calculated for the segmented abnormal region. Compared to previous approaches, the new framework is more flexible and able to better exploit spatial information leading to improved nosologic images. The combined scheme offers a new way to produce high-resolution nosologic images, representing tumor heterogeneity and class probabilities, which may help clinicians in decision making.

Incorporating in vivo and ex vivo NMR sources of information for modeling robust brain tumor classifiers

The purpose of this paper is to investigate the potential and limitations of using multimodal sources of information coming from in vivo NMR and ex vivo NMR data for detecting brain tumors. Supervised pattern recognition methods, whose performance directly depends on the prior available observations used in building them, are proposed. We show that high resolution magic angle spinning (HR-MAS) data act as complementary information for classifying magnetic resonance spectroscopic imaging (MRSI) data. In particularly, when considering rare brain tumors, since it is unlikely to acquire sufficient cases to define their metabolite profiles using only in vivo NMR information, HR-MAS can support the classification procedure. We describe different approaches to combine HRMAS data with in vivo MRSI and magnetic resonance imaging (MRI) data and investigate which parameters influence the classification results by means of extensive simulations and in vivo studies.

Model independent pre-processing of X-ray powder diffraction profiles

Precise knowledge of X-ray diffraction profile shape is crucial in the investigation of the properties of matter in crystals powder. Line-broadening analysis is the fourth pre-processing step in most of the full powder pattern fitting softwares. The final result of line-broadening analysis strongly depends on three further steps: noise filtering, removal of background signal, and peak fitting. In this work a new model independent procedure for two of the aforementioned steps (background suppression and peak fitting) is presented. The former is dealt with by using morphological mathematics, while the latter relies on the Hankel-Lanczos singular value decomposition technique. Real X-ray powder diffraction (XRPD) intensity profiles of Ceria samples are used to test the performance of the proposed procedure. Results show the robustness of this approach and its capability of efficiently improving the disentangling of instrumental broadening. These features make the proposed approach an interesting and user-friendly tool for the pre-processing of XRPD data.

MAGNETIC RESONANCE SPECTROSCOPIC IMAGING: A SURVEY OF QUANTIFICATION AND CLASSIFICATION ALGORITHMS

Abstract The purpose of this paper is to provide a survey of a few but significant methods currently used by the Nuclear Magnetic Resonance (NMR) community to quantify and classify MR Spectroscopic Imaging (MRSI) signals. In particular, several time–domain algorithms, able to extract from MRSI data useful information aboute metabolite concentrations, will be outlined. Furthermore, some methods able to exploit this information to detect tumors and classify their grade of malignancy will also be given. Finally, a very recent tissue typing technique, based on the use of a statistical method called canonical correlation analysis, will be described.