Validation of computed tomography for measuring lung weight

Abstract

Background: Lung weight characterises severity of pulmonary oedema and predicts response to mechanical ventilation. The aim of this study was to evaluate the accuracy of quantitative analysis of thorax computed tomography (CT) for measuring lung weight in pigs with or without pulmonary oedema. Methods: Thirty-six pigs were mechanically ventilated with different tidal volumes and positive end-expiratory pressures that did or did not induce pulmonary oedema. After 54 h, they underwent thorax CT (CTin vivo) and were then sacrificed and exsanguinated. Fourteen pigs underwent a second thorax CT (CTpost-exsang.) after exsanguination. Lungs were excised and weighed with a balance (balancepost-exsang.). Agreement between lung weights measured with the balance (considered as reference) and those estimated by quantitative analysis of CT was assessed with Bland-Altman plots. Results: One animal unexpectedly died before CTin vivo. In 35 pigs, lung weight measured with balancepost-exsang. was 371 ± 184 g and that estimated with CTin vivo was 481 ± 189 g (p < 0.001). Bias between methods was −111 g (−35%) and limits of agreement were −176 (−63%) and −46 g (−8%). Measurement error was similar in animals with (−112 ± 45 g; n = 11) or without (−110 ± 27 g; n = 24) pulmonary oedema (p = 0.88). In 14 pigs with thorax CT after exsanguination, lung weight measured with balancepost-exsang. was 342 ± 165 g and that estimated with CTpost-exsang. was 352 ± 160 g (p = 0.02). Bias between methods was −9 g (−4%) and limits of agreement were −36 (−11%) and 17 g (3%). Measurement errors were similar in pigs with (−1 ± 26 g; n = 11) or without (−12 ± 7 g; n = 3) pulmonary oedema (p = 0.12). Conclusions: Compared to the balance, CT obtained in vivo constantly overestimated the lung weight, as it included pulmonary blood (whereas the balance did not). By contrast, CT obtained after exsanguination provided accurate and reproducible results.