| HAL-Inria | Publications, software ... of Inria's scientists |
Reports
Region-Of-Interest Selection and Statistical Analysis of Dynamic Breast Magnetic Resonance Imaging Data
Abstract : Magnetic Resonance Imaging (MRI) is emerging as a powerful tool for the diagnosis of breast abnormalities. Early investigation of contrast-enhanced MRI of the breast demonstrated that breast carcinoma consistently enhanced after the intravenous administration of contrast material (Gadolinium). However, many benign lesions also enhance reducing as a result MRI specificity- . Further characterization is necessary to differentiate benign from malignant lesions and analysis of the temporal pattern of contrast uptake has been applied to improve specificity. Selecting a Region-Of-Interest (ROI) is an almost universal step in the process of examining the contrast uptake characteristics of a breast lesion. A limitation of most ROI analysis procedures is that only subjectively selected regions are examined. We propose a ROI selection method based on techniques from multivariate data analysis and statistical image segmentation. In addition to the enhancement rate of the contrast material, we use other characteristics of the signal intensity to group areas of the breast with similar characteristics and then select the group which corresponds to a fast and strong enhancement. We then investigate tools for subsequent analysis of signal intensity time course data in the selected region in order to determine if the enhancing lesion is likely to be benign or malignant. We tested our procedure on cases for which the diagnosis was available. The results are promising and are consistent with the diagnosed nature of the lesions.
https://hal.inria.fr/inria-00072338
Contributor : Rapport de Recherche Inria <>
Submitted on : Tuesday, May 23, 2006 - 8:27:55 PM
Last modification on : Wednesday, October 14, 2020 - 3:49:25 AM
Long-term archiving on: : Sunday, April 4, 2010 - 11:04:29 PM
Contributor : Rapport de Recherche Inria <>
Submitted on : Tuesday, May 23, 2006 - 8:27:55 PM
Last modification on : Wednesday, October 14, 2020 - 3:49:25 AM
Long-term archiving on: : Sunday, April 4, 2010 - 11:04:29 PM