The multiple inert gas elimination (MIGE) technique, is a powerful test for maldistribution of ventilation/perfusion ratios VPR in the lung. Previous methods of MIGE data analysis, however, have suffered from certai theoretical difficulties which limit their usefulness in recovering VPR distribution. The present investigation proposes new, improved methods of analyzing MIGE data. Advanced technique of signal and image processing, systems engineering and information theory, which have proven successful in solving other physical and engineering problems of a similar nature but are hitherto unknown in respiratory physiology, will be introduced to the solution of the VPR problem. Our objective are to understand quantitatively the theoretical limits of VPR recoverability and to explore alternative methods for its improvement. The following approaches will be taken: 1) the method of """"""""maximum entropy"""""""" for spectral estimation will be employed to directly recover VPR distribution, avoiding degradationg caused by parametric assumptions normally required in earlier procedures; 2) the confidence limits of the entire distribution will be derived by using optimization technique; 3) the quantitative effects of incomplete and noisy measurements on the resulting resolution will be examined from the perspective of sampling theory and spectral analysis; and 4) variability of the recovered distribution will be minimized through improved experimental design which maximizes the information content. The analytical results will be thoroughly evaluated using both hypothetical and empirical data. The result will complement the experimental MIGE procedure and will provide a theoretical basic for more extensive experimental validation in the future.
| Wang, X H; Poon, C S (1991) Spectral vs. compartmental averaging of VA/Q distributions: confidence limits. J Appl Physiol 70:1290-9 |
