Image-derived input functions for PET brain studies

To assess the robustness of a previously introduced method to obtain accurate image-derived input functions (IDIF) for three other tracers. Dynamic PET and online blood data of five repeat [(11)C]PIB (Pittsburgh Compound-B) ([(11)C]PIB), six repeat (R)-[(11)C]verapamil, and ten single (R)-[(11)C]PK11195 studies were used. IDIFs were extracted from partial volume corrected scans using the four hottest pixels per plane method. Results obtained with IDIFs were compared with those using standard online measured arterial input functions (BSIF). IDIFs were used both with and without calibration based on manual blood samples. For (R)-[(11)C]verapamil, accurate IDIFs were obtained using noncalibrated IDIFs (slope 0.96 +/- 0.17; R (2) 0.92 +/- 0.07). However, calibration was necessary to obtain IDIFs comparable to the BSIF for both [(11)C]PIB (slope 1.04 +/- 0.05; R (2) 1.00 +/- 0.01) and (R)-[(11)C]PK11195 (slope 0.96 +/- 0.05; R (2) 0.99 +/- 0.01). The need for calibration may be explained by the sticking property of both tracers, indicating that BSIF may be affected by sticking and therefore may be unreliable. The present study shows that a previously proposed method to extract IDIFs is suitable for analysing [(11)C]PIB, (R)-[(11)C]verapamil and (R)-[(11)C]PK11195 studies, thereby obviating the need for online arterial sampling.