Research press release


Nature Medicine

Lung cancer identified earlier with AI




D Tseたちは、肺結節の悪性度をヒトが加わらずに判定できる深層学習モデルを開発し、4万2290枚のCTスキャン画像を使って訓練した。このAI搭載システムは、テストに使われた6716例で微小な悪性の肺結節を94%の正診率で検出できた。以前に撮影したCT画像が得られない場合には、このモデルの成績は、診断を行った6人の放射線科医全員の成績を上回り、以前の画像がある場合には医師たちと同程度だった。


An artificial intelligence framework able to detect malignant lung nodules on chest computed tomography (CT) scans with performance meeting or exceeding that of expert radiologists is reported this week in Nature Medicine. This deep-learning model offers an automated image evaluation system to enhance the accuracy of early lung cancer diagnosis that could inform clinical interventions.

Lung cancer is the most common cause of cancer-related death in the United States, resulting in an estimated 160,000 deaths in 2018. Large clinical trials across the United States and Europe have shown that chest screening can identify the cancer and reduce death rates. However, high error rates and the limited applicability of this approach, alongside other clinical factors, mean that many lung cancers are being detected at advanced stages when they are hard to treat.

Daniel Tse and colleagues developed a deep-learning model and trained it on 42,290 CT scan images to predict the malignancy of lung nodules without the need for human involvement. They found that the artificial-intelligence-powered system was able to spot minuscule malignant lung nodules with an accuracy of 94% in 6,716 test cases. The model outperformed all six radiologists when previous CT imaging was not available and performed as well as the radiologists when there was prior imaging.

The authors caution that these findings need to be clinically validated in large patient populations, but they suggest this model may assist in improving the management and outcome of patients with lung cancer.

doi: 10.1038/s41591-019-0447-x


メールマガジンリストの「Nature 関連誌今週のハイライト」にチェックをいれていただきますと、毎週各ジャーナルからの最新の「注目のハイライト」をまとめて皆様にお届けいたします。