This study aims to evaluate the performance of using EIT to infer spatial information for the purpose of diagnostic screening.

We recruited 46 lung disease patients and 24 healthy controls to perform simultaneous EIT and spirometry measurements with 3 or more repetitions. Lung disease patients include 12 ILD, 10 asthma, 8 COPD, 8 bronchiectasis, and 8 with other diseases A regression model using EIT indicators and anthropometrics (including chest circumference, weight, height, BMI and age) is used to predict spirometry indicators such as FEV1, FVC and FEV1/FVC. Furthermore, indicator maps for the same indicators are computed, from which maps are selected based on a classifier, resulting in a smaller set of indicators maps. The number of subjects and number of indicator maps pairs selected include asthma, bronchiectasis, COPD , normal and others. Finally, the number of activated pixels (NA), defined as the pixels whose amplitude is larger than 10% of the maximum, are computed. Boferroni T-test is performed to compare these indicators between normal and disease groups.

The results showed significant correlation of the predicted global spirometry indicators (p<0.0001). Furthermore, the T-test of the spatial indicators (Figure 1B) showed significant differences (p<0.01) between normal subjects and asthma (FVC-NA and FEV1-NA), COPD (FEV1-NA), bronchiectasis (FEV1-NA), FEV1-NA. While the differences were not significant for other indicators.

The results suggest that EIT has the potential to be used to predict global spirometry indicators in addition to providing spatial information that is potentially useful for diagnostic screening.