Automatic preparation of radiopharmacokinetic data for in vivo estimation of receptor biochemistry

We present a fully automated region of interest (ROI) and motion correction program for the generation of heart and liver time-activity data resulting from a hepatic functional imaging study using [99mTc-]galactosyl-neoglycoalbumin (99mTc-NGA).

The program automatically draws heart and liver ROI and corrects for lateral movement of the subject. Eighty-four 99mTc-NGA studies, consisting of 32 healthy subjects and 52 patients with liver disease, were processed and submitted to an automated kinetic analysis that estimates the subject's asialoglycoprotein receptor concentration [R]o.

When compared to time-activity data generated by operator-drawn ROIs without motion correction, the average reduced Chi-square of the kinetic analysis decreased significantly (p < 0.001) from 2.20 to 1.37 and the number of studies that satisfied quality control increased from 74 to 81 studies. Receiver operating characteristic of [R]o resulted in greater detectability (0.984 +/- 0.012 compared with 0.965 +/- 0.020) when automatic ROI generation was employed. Using the test criteria of 0.65 microM, the sensitivity of [R]o increased from 0.88 to 0.92 and the specificity increased from 0.96 to 0.97.

Automated definition of liver and heart ROIs with motion correction, that reduces observational noise, increased the success rate of the radiopharmacokinetic analysis from 88% to 96%.