Development of a novel artificial intelligence algorithm to detect pulmonary nodules on chest radiography

Abstract

Background: In this study, we aimed to develop a novel artificial intelligence (AI) algorithm to support pulmonary nodule detection, which will enable physicians to efficiently interpret chest radiographs for lung cancer diagnosis.

Methods: We analyzed chest X-ray images obtained from a health examination center in Fukushima and the National Institutes of Health (NIH) Chest X-ray 14 dataset. We categorized these data into two types: type A included both Fukushima and NIH datasets, and type B included only the Fukushima dataset. We also demonstrated pulmonary nodules in the form of a heatmap display on each chest radiograph and calculated the positive probability score as an index value.

Results: Our novel AI algorithms had a receiver operating characteristic (ROC) area under the curve (AUC) of 0.74, a sensitivity of 0.75, and a specificity of 0.60 for the type A dataset. For the type B dataset, the respective values were 0.79, 0.72, and 0.74. The algorithms in both the type A and B datasets were superior to the accuracy of radiologists and similar to previous studies.

Conclusions: The proprietary AI algorithms had a similar accuracy for interpreting chest radiographs when compared with previous studies and radiologists. Especially, we could train a high quality AI algorithm, even with our small type B data set. However, further studies are needed to improve and further validate the accuracy of our AI algorithm.

Keywords: artificial intelligence (AI); chest radiography; deep learning; lung cancer.

Fig. 1.

Schematic view of cross-validation. We randomly divided the dataset into five groups that included positive and negative data at equal rates. Then, we validated one group as test data and used the other groups as training data. Next, we assigned each group as test data and obtained five sets of results (Results 1-5). Finally, we calculated the average accuracy for each set.

Fig. 2.

Receiver operating characteristic (ROC) curves for the type A (a) and type B (b) datasets.

Fig. 3.

Three examples from datasets of a health examination center in this study. The proposed AI algorithm correctly detected pulmonary nodules and localized the areas in the image that were most indicative of pulmonary nodules (a-1, a-2). The AI algorithm also detected pulmonary nodules that were missed by the physicians (b-1, b-2). A false-positive display is shown in c-1 and c-2, which requires improvement. The positive probability scores of these cases are shown in a-2, b-2, and c-2, respectively.