Hepatic perfusion index: a multicentre trial.

: The reproducibility and accuracy of the hepatic perfusion index (HPI) was examined by consideration of in vitro and in vivo factors. A phantom was used to simulate liver blood flow and data acquired on nine gamma cameras. Dynamic hepatic scintigraphy was undertaken on 28 patients at two centres and values obtained for the HPI. Results from the phantom study showed good agreement between the nine cameras and also with the HPI values predicted from the measured phantom flow rates. The results of the patient study indicated a high degree of conformity between observers (r = 0.95, S.E = 0.03) but poorer correlation between the HPI values from the two centres (r = 0.67, S.E. = 0.09). These results imply that centres wishing to use HPI clinically should establish their own range of normality.