Shape Analysis in Molecular Imaging

Abstract

Molecular imaging is a new research discipline enabling the visualization, characterization and quantification of biologic processes taking place at the cellular and subcellular levels within intact living subjects. Applications of molecular imaging techniques will benefit various clinical practices including classification and tracking of chemotherapy and treatment planning of radiotherapy, as well as drug discovery and development. Molecular imaging typically includes two or three dimensional imaging with quantification over time, and is often applied on molecular imaging modalities, such as Positron Emission Tomography (PET), Single Photon Emission Computed Tomography (SPECT) etc. Image series acquired with spatiotemporal distribution of molecular biomarkers must be carefully analyzed to estimate the underlying physiology-related metabolic parameters. Shape analysis is one of the most powerful tools to analyze the geometrical properties from similar shapes or different groups, and can be applied to estimate both the concentration of biomarkers and interaction between biomarkers and tissue/organs. However, some limitations from molecular imaging modalities and clinical practices still hinder the quantitative accuracy of shape analysis, e.g. the low spatial and temporal resolution in PET scan, the inaccuracy of blood samplings from patients, the low Signal-to-Noise (SNR) ratio of measurement data in dynamic PET/CT scan. In this chapter, firstly, we will introduce the definition of molecular imaging, the clinical advantages and limitations of various molecular imaging modalities, secondly, we will review the challenges in data analysis based on the data processing procedure, and explain how data corrections affect the accuracy of static and dynamic PET imaging, thirdly, the general frameworks of image processing in PET and SPECT are reviewed with focus on image reconstruction, at last, we will show some recent advancements and give examples of clinical applications.