Method and system of multivariate analysis on volume-wise data of reference structure normalized images for improved quality in positron emission tomography studies

US8233689B2 · kind B2 · utility

Assignee

• GE HEALTHCARE LIMITED · GB

Inventors

• Pasha Razifar · Uppsala, SE
• Mats Bergstrom · Malmö, SE
• Bengt Langstrom · Uppsala, SE

Key dates

• Filing date: Aug 31, 2006
• Grant date: Jul 31, 2012
• Expiry date: Apr 17, 2029

Classification

• Technology area (CPC G): Physics
• CPC primary: G06T2207/30016
• WIPO field: Measurement
• WIPO sector: Instruments

Abstract

A method and system are provided to mask out the background in dynamic PET images, to perform pre-normalization on the masked dynamic PET images, and to apply multivariate image analysis (e.g., principal component analysis PCA) on the masked pre-normalized dynamic PET images in order to improve the quality of the dynamic PET images and the PET study. A masking operation applies PCA to untreated dynamic PEET images before any pre-normalization in order to mask out the background pixels. This masking operation uses the Otsu method. A first normalization method is background noise pre-normalization where pixel values are corrected for background noise. A second normalization method is kinetic pre-normalization where the contrast within an image is improved. Multivariate analysis such as PCA may be applied on the whole volume to find the largest variance in the structure.