Limitations of Kr-81m for quantitation of ventilation scans.

Predictions of Kr-81m behavior in the lungs have been based on a single-compartment model of well-mixed gas. To determine the validity of this model, anesthetized, paralyzed dogs were ventilated mechanically over a wide range of ventilations with Kr-81m in the inspirate. Averaged data obtained from 1.5- to 2-min images were compared with the model's predictions. Krypton-81m concentration appeared more linearly related to ventilation than the model predicts. This suggests that the Kr-81m may not be distributed to the entire resident gas volume or that its distribution changes with lung volume. Estimation of end-expiratory volume (FRC)--based on the tidal volume and the maximum-to-minimum count ratio over the lung during gated acquisitions--underestimated the true FRC by about 40%. The magnitude of error depended upon the combination of tidal volume and frequency and the inspiratory time. Thus, Kr-81m does not mix well with resident lung gas, and the well-mixed, single-compartment model is not a good predictor of Kr-81m behavior in the lung.