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.
Highlights
Lung cancer is a major cause of death worldwide
We evaluated its accuracy for screening and nodule detection tasks
We trained and investigated a Convolutional network (CNN) for pulmonary nodule detection and compared the predictions made by the CNN to the results of two professional radiologists
Summary
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. For successful lung cancer screening, keeping the rate of false-negatives and false-positives as low as possible is mandatory. For mammography and chest X-ray classification, networks which are trained with case-level labels showed promising results[9,10,11,12]. Such systems can only provide disease locations by the use of techniques such as saliency maps[13]. Current state of the art methods for pneumothorax d etection[17] or mammography s creenin[6] make use of box-annotations, which can be derived from pixel wise annotations Both aforementioned studies use a RetinaNet architecture, a one stage d etector[18, 19], which is characterized by a faster inference time than two stage detectors[20, 21]. We compared its performance to the participants of a reader study
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
