Controlling familywise error rate for matched subspace detection in dynamic FDG PET
Zheng Li1, Quanzheng Li, Dimitrios Pantazis
1Signal and Image Processing Institute, University of Southern California, Los Angeles, CA 90089, USA.
IEEE Transactions on Medical Imaging
|September 29, 2009
Summary
This study introduces a new statistical method to improve the detection of small tumors in dynamic fluorodeoxyglucose (FDG) positron emission tomography (PET) scans. The approach enhances the accuracy of identifying metastatic lesions by controlling false positives.
Area of Science:
- Medical Imaging
- Nuclear Medicine
- Oncology
Background:
- Detection of small lesions in fluorodeoxyglucose (FDG) positron emission tomography (PET) is challenging due to limitations in image resolution and signal-to-noise ratio.
- Previous work introduced a matched subspace detection method utilizing time activity curves to differentiate tumors from background in dynamic FDG PET.
Purpose of the Study:
- To describe and evaluate a novel thresholding method for controlling the familywise error rate (FWER) in matched subspace detection statistical maps.
- To enhance the identification of potential secondary or metastatic tumors in dynamic FDG PET imaging.
Main Methods:
- The proposed method involves segmenting PET images into homogeneous regions.
- Statistical maps are normalized to a zero mean unit variance Gaussian random field.
- Thresholding is performed using a random field theory maximum statistic approach, estimating spatial smoothness to control FWER.
Main Results:
- The thresholding method was evaluated using digital phantoms derived from clinical dynamic images.
- The approach was applied to clinical PET data from a breast cancer patient with metastatic disease, demonstrating its practical application.
Conclusions:
- The developed thresholding method effectively controls the familywise error rate for matched subspace detection statistical maps in dynamic FDG PET.
- This technique offers improved accuracy for detecting small metastatic lesions, aiding in cancer diagnosis and management.
