Interstitial Lung Disease Diagnosis

This study developed an artificial neural network to differentiate between 11 interstitial lung diseases using features from chest radiographs and clinical parameters. The network’s output significantly improved radiologists’ diagnostic accuracy, with the average area under the ROC curve increasing from 0.826 without the network to 0.911 with it. The results suggest that this neural network can serve as a valuable “second opinion” tool for radiologists in diagnosing interstitial lung diseases.