Improving Detection of Axillary Lymph Nodes by Computer-Aided Kinetic Feature Identification in Positron Emission Tomography
Abstract
The goal of this project is to improve the detection of metastatic axillary breast cancer through sophisticated physiological modeling and statistical signal processing techniques. The major focus of Year 4 (no-cost extension) was the assessment of the developed methodology in feature identification, extraction/estimation, and early detection with digital phantom, animal study and a small amount of patient data. The methods we tested include (i) adding the factor analysis to the conventional ROI averaging method for feature extraction/estimation and (ii) to develop the optimal feature-guided filtering criteria for early lesion detection. The objective is to suppress the interference-plus- noise in dynamic data proceeding to applying the hypothesis test detection criteria. Two types of filters presented in the last annual report, with/without using the physiological features extracted from normal tissues, were further tested with digital phantom, animal study and patient data. The results of phantom study demonstrated that the accuracy of feature extraction in the modified method can be dramatically improved compared to the conventional ROI analysis and that the feature-guided space-temporal filters can enhance the SNR in invisible lesion and make it become detectable. The assessment of non-invasive blood time activity extraction was also performed with patient data selected from our clinical database. All findings in theory and simulations will be continued in the extended year 5 when more clinical data will become available.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2003
- Accession Number
- ADA423474
Entities
People
- Xiaoli Yu
Organizations
- University of Southern California