Visibility-driven PET-CT visualisation with region of interest (ROI) segmentation

  • Authors:

  • Younhyun Jung;Jinman Kim;Stefan Eberl;Micheal Fulham;David Dagan Feng

  • Affiliations:

  • Biomedical and Multimedia Information Technology (BMIT) Research Group, University of Sydney, Sydney, Australia;Biomedical and Multimedia Information Technology (BMIT) Research Group, University of Sydney, Sydney, Australia;Biomedical and Multimedia Information Technology (BMIT) Research Group, University of Sydney, Sydney, Australia and Department of Molecular Imaging, Royal Prince Alfred Hospital, Sydney, Australia;Biomedical and Multimedia Information Technology (BMIT) Research Group, University of Sydney, Sydney, Australia and Department of Molecular Imaging, Royal Prince Alfred Hospital, Sydney, Australia ...;Biomedical and Multimedia Information Technology (BMIT) Research Group, University of Sydney, Sydney, Australia and Med-X Research Institute, Shanghai Jiao Tong University, Shanghai, China

  • Venue:
  • The Visual Computer: International Journal of Computer Graphics
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multi-modality (MM) positron emission tomography-computed tomography (PET-CT) visualises biological and physiological functions (from PET) as region of interests (ROIs) within a higher resolution anatomical reference frame (from CT). The need to efficiently assess and assimilate the information from these co-aligned volumes simultaneously has stimulated new visualisation techniques that combine 3D volume rendering with interactive transfer functions to enable efficient manipulation of these volumes. However, in typical MM volume rendering visualisation, the transfer functions for the volumes are manipulated in isolation with the resulting volumes being fused, thus failing to exploit the spatial correlation that exists between the aligned volumes. Such lack of feedback makes MM transfer function manipulation complex and time consuming. Further, transfer function alone is often insufficient to select the ROIs when they have similar voxel properties to those of non-relevant regions.In this study, we propose a new ROI-based MM visibility-driven transfer function (m 2-vtf) for PET-CT visualisation. We present a novel `visibility' metric, a fundamental optical property that represents how much of the ROIs are visible to the users, and use it to measure the visibility of the ROIs in PET in relation to how it is affected by transfer function manipulations to its counterpart CT. To overcome the difficulty in ROI selection, we provide an intuitive ROI selection tool based on automated PET segmentation. We further present a MM transfer function automation where the visibility metrics from the PET ROIs are used to automate its CT's transfer function. Our GPU implementation achieved an interactive visualisation of PET-CT with efficient and intuitive transfer function manipulations.