Smoothing of MEFV curves by digital filtering of flow as a function of volume.

Noise in flow records limits the degree to which one can characterize local features of maximum expiratory flow-volume (MEFV) curves, i.e., the volume dependence of flow over small volume intervals such as 5-10% of the vital capacity. Such noise includes instrument noise, cardiogenic pulsations, and rapid flow oscillations due to audible turbulence. Noise removal by analog filtering of the flow vs. time signal often is unacceptable because the initial rapid onset of flow can be severely distorted. A digital technique for directly filtering the flow vs. volume record that does not distort the initial rapid flow transient has been developed. Flow values are determined at 256 volume points equally spaced over the vital capacity. This sequence of flow values is passed through a linear phase, finite impulse response, Hamming window filter. The output sequence of filtered flow values, plotted against the corresponding volumes, quantitatively retains the local shape features of the MEFV curve without the noise.