A robust convolutional neural network for lung nodule detection in the presence of foreign bodies

Abstract

Lung cancer is a major cause of death worldwide. As early detection can improve outcome, regular screening is of great interest, especially for certain risk groups. Besides low-dose computed tomography, chest X-ray is a potential option for screening. Convolutional network (CNN) based computer aided diagnosis systems have proven their ability of identifying nodules in radiographies and thus may assist radiologists in clinical practice. Based on segmented pulmonary nodules, we trained a CNN based one-stage detector (RetinaNet) with 257 annotated radiographs and 154 additional radiographs from a public dataset. We compared the performance of the convolutional network with the performance of two radiologists by conducting a reader study with 75 cases. Furthermore, the potential use for screening on patient level and the impact of foreign bodies with respect to false-positive detections was investigated. For nodule location detection, the architecture achieved a performance of 43 true-positives, 26 false-positives and 22 false-negatives. In comparison, performance of the two readers was 42 ± 2 true-positives, 28 ± 0 false-positives and 23 ± 2 false-negatives. For the screening task, we retrieved a ROC AUC value of 0.87 for the reader study test set. We found the trained RetinaNet architecture to be only slightly prone to foreign bodies in terms of misclassifications: out of 59 additional radiographs containing foreign bodies, false-positives in two radiographs were falsely detected due to foreign bodies.

An Overview of Artificial Intelligence in Oncology

“…There are examples of studies showing that combining AI and human power improved mammography screening for breast cancer [ 17 , 18 ] Detection: AI can aid in the identification of cancerous lesions that could otherwise be missed by humans. For instance, it can be used to find lung nodules [ 19 ] or brain metastases on MRI readouts [ 20 ]. Detection relies on the the use of bounding boxes to detect a lesion or object of interest.…”

Deep learning applied to automatic disease detection using chest X‐rays

“…DDN also outperformed human observers in generic lung nodule detection. Schultheiss 32 employed RetinaNet with a pretrained ResNet‐101 backbone using a dataset of 411 CXRs, 257 with annotated pulmonary nodules. It achieved AUROC 0.87 and outperformed human observers (43 vs. 42 TP, 26 vs. 28 FP and 22 vs. 23 FN).…”

Per-Pixel Lung Thickness and Lung Capacity Estimation on Chest X-Rays using Convolutional Neural Networks

“…Due to the design of the loss function, there will be some extrathoracic pixels marked as lung pixels, which are removed by multiplying the prediction with a lung mask. To obtain the lung mask, we utilize a U-Net lung segmentation network trained with JSRT dataset [29] and JSRT mask data [30], as trained in some of our previous work [31]. Small connected segmentation components (area smaller than 4100 pixels), which are usually extrathoracic segmentation predictions, were removed from the lung mask.…”