Image masking using convolutional networks improves performance classification of radiation pneumonitis for non-small cell lung cancer

Radiation pneumonitis (RP) is a serious side effect of radiotherapy in patients with locally advanced non-small-cell lung cancer (NSCLC). Image cropping reduces training noise and may improve classification accuracy. This study proposes a prediction model for RP grade >= 2 using a convolutional neural network (CNN) model with image cropping. The 3D computed tomography (CT) images cropped in the whole-body, normal lung (nLung), and nLung regions overlapping the region over 20 Gy (nLung boolean AND 20 Gy) used in treatment planning were used as the input data. The output classifies patients as RP grade < 2 or RP grade >= 2. The sensitivity, specificity, accuracy, and area under the curve (AUC) were evaluated using the receiver operating characteristic curve (ROC). The accuracy, specificity, sensitivity, and AUC were 53.9%, 80.0%, 25.5%, and 0.58, respectively, for the whole-body method, and 60.0%, 81.7%, 36.4%, and 0.64, respectively, for the nLung method. For the nLung boolean AND 20 Gy method, the accuracy, specificity, sensitivity, and AUC improved to 75.7%, 80.0%, 70.9%, and 0.84, respectively. The CNN model, in which the input image is segmented in the normal lung considering the dose distribution, can help predict an RP grade >= 2 for NSCLC patients after definitive radiotherapy.

Automated summary

This study proposes a prediction model using a convolutional neural network (CNN) model with image cropping to predict the severity of radiation pneumonitis (RP) in non-small-cell lung cancer (NSCLC) patients after radiotherapy. The CNN model uses 3D computed tomography (CT) images, cropped in different regions, as input data to classify patients into RP grade < 2 or RP grade >= 2. The performance of the model is evaluated using sensitivity, specificity, accuracy, and area under the curve (AUC) measures, with the nLung boolean AND 20 Gy method achieving the highest accuracy, specificity, sensitivity, and AUC.