Re-Identification and Growth Detection of Pulmonary Nodules without Image Registration Using 3D Siamese Neural Networks
This addresses the time-consuming and error-prone task for clinical radiologists in monitoring lung cancer progression, representing a domain-specific incremental improvement.
The study tackled the problem of lung cancer follow-up by developing a 3D Siamese neural network to re-identify and detect growth of pulmonary nodules in CT scans without image registration, achieving a nodule detection sensitivity of 94.7%, matching accuracy of 88.8%, and growth detection sensitivity of 92.0% with 88.4% precision.
Lung cancer follow-up is a complex, error prone, and time consuming task for clinical radiologists. Several lung CT scan images taken at different time points of a given patient need to be individually inspected, looking for possible cancerogenous nodules. Radiologists mainly focus their attention in nodule size, density, and growth to assess the existence of malignancy. In this study, we present a novel method based on a 3D siamese neural network, for the re-identification of nodules in a pair of CT scans of the same patient without the need for image registration. The network was integrated into a two-stage automatic pipeline to detect, match, and predict nodule growth given pairs of CT scans. Results on an independent test set reported a nodule detection sensitivity of 94.7%, an accuracy for temporal nodule matching of 88.8%, and a sensitivity of 92.0% with a precision of 88.4% for nodule growth detection.