Establishment of fully automated Tc-99m TRODAT-1 expert system

• Main Authors: Yu-Xiang Guan, 官郁翔
• Other Authors: Jason J.S. Lee
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/56166406042300772796

碩士 === 國立陽明大學 === 生物醫學影像暨放射科學系 === 101 === 99mTc TRODAT-1 is a SPECT dopamine transporter (DAT) tracer that probes dopaminergic cell loss in Parkinson’s disease (PD). Semi-quantification by drawing regions of interest (ROIs) manually is a common method to evaluate 99mTc TRODAT-1 SPECT images. However, it is time consuming, low reproducibility and subjective observer. Our aim was to develop the expert system that could automatically quantify TRODAT-1 SPECT images The present study collected 59 controls and 84 patients with PD retrospectively. There were the MR based 99mTc-TRODAT-1 (MBT) template and 99mTc-HMPAO based TRODAT (HBT) template created by control images. Then MBT template, HBT template, 123I-IBZM template and 99mTc-HMPAO template were compared with manual ROI analysis, respectively. The images of striatal phantom were analyzed by expert system and manual ROI analysis, respectively. The specific uptake ratio (SUR) of striatal uptake was delimited into three parts which were normal range, mildly reduced and severely reduced by receiver operating characteristic (ROC) curve. In the comparison of template, the variability and intra-class correlation coefficient (ICC) between MBT template and manual VOI analysis is better than others, which were 9.27% and 0.7, respectively. In phantom study, the Pearson correlation coefficient between expert system and manual ROI analysis for bilateral striatum is 0.951(p<0.05). The cut points of SUR in rough scale for normal range, mildly reduced and severely reduced of striatal uptake is 0.989 (sensitivity=100%, specificity =97.6%) and 0.438 (sensitivity=87.8%, specificity=64.5%) respectively. The expert system is useful for the detection of reduced striatal uptake on 99mTc TRODAT-1 SPECT images automatically, and especially diagnosis in early PD.