Towards Accurate Pulmonary Nodule Detection by Representing Nodules as Points With High-Resolution Network

Abstract

Almost all successful nodule detectors rely heavily on a fixed set of anchor boxes. In this paper, inspired by the success of the keypoint estimation method in natural image detection, we propose an anchor-free framework for accurate pulmonary nodule detection. We first present a novel representation for detecting nodules, in terms of their 3D center locations, which reduces the number of hyper-parameters and the corresponding computation related to anchors, thus making the nodule detection pipeline much simpler. Then, an effective two-stream network is introduced to reduce the false positive nodule candidates, by aggregating information from the image stream and motion-history stream. Experiments show that the proposed approach achieves a sensitivity of 96.1%, with 8 false positives per scan, and a CPM score of 90.6%, on the publicly available LUNA16 dataset, which outperforms other state-of-the-art methods. By testing on the SPIE-AAPM dataset with models pre-trained on the LUNA16, our proposed method yields 92.8% sensitivity with 8 false positives per scan. This demonstrates the effectiveness and generalization ability of our method.

Pulmonary Nodule Detection Based on Faster R-CNN With Adaptive Anchor Box

“…It can also be observed from Table 4 that our proposed system achieves the CPM score of 88.2%, being at the middle among the other works. Our methods are only 2.4% less accurate than the top performer at 90.6% [45], while still have a considerable gap with the other end ( [23] and [46] at 79.0% and 79.6%, respectively).…”

“…Table 5 presents FROC performance and CPM score of our proposed system and compares the results with other methods on SPIE-AAPM dataset. From the table, it can be observed that even with unseen data from SPIE-AAPM dataset, our method achieves a sensitivity of 89.3% and a CPM score of 84.8% which outperforms other nodule detection methods [45], [50] on the same dataset. These results demonstrate the generalization of our proposed model.…”

Computergestützte Detektion solider Lungenherde

“…Hier haben sich in den letzten Jahren verschiedene Modelle entwickelt, welche mithilfe von neuronalen Netzwerken versuchen, die Sensitivität und Spezifität zu erhöhen [3,4].…”