Abstract
Accurate and automatic segmentation of pulmonary nodules in 3D thoracic Computed Tomography (CT) images is of great significance for Computer-Aided medical Diagnosis (CAD) of lung cancer. Currently, this important task remains challenging for lack of the voxel-level annotation and training strategies that balance target/background voxels in thoracic CT images. In this paper, a new region-based network, called Nodule-plus Region-based CNN, is proposed to detect pulmonary nodules in 3D thoracic CT images effectively while synchronously generating an instance segmentation mask for every detected instance. Our new network is constructed with a stack of convolutional blocks in which lateral connections are used to alleviate the difficulty of vanishing gradients. In addition, in order to reduce annotation workload and make best use of unannotated samples, we proposed a new Deep Self-paced Active Learning (DSAL) strategy by combining Active Learning (AL) and Self-Paced Learning (SPL) strategies. For the purpose of evaluating the performance of our proposed Nodule-plus R-CNN, we conduct a series of experiments on the public LIDC-IDRI dataset, and our model achieves 0.66 Dice and 0.96 TP Dice, which are state-of-the-art best results of pulmonary nodule segmentation. When the amount of available annotated samples is limited, our model trained with the DSAL strategy performs much better than that trained with the standard strategy.
Highlights
L UNG cancer is known as one of the most lethal malignancies worldwide
We develop a new Deep Self-paced Active Learning (DSAL) strategy to considerably reduce annotation workload based on the idea of bootstrapping [12], [13]
Our experimental results on the public LIDC-IDRI dataset [1] suggest that the Nodule-plus R-convolutional neural network (CNN) proposed by us achieves state-of-the-art best performance of pulmonary nodule instance segmentation with respect to both the Dice and TP Dice criteria; further, our weakly-supervised segmentation approach (Deep Self-paced Active Learning) is much more effective than common fully supervised methods [9], [20] and other best-known weakly labeled methods [9] when using the same amount of annotated samples
Summary
L UNG cancer is known as one of the most lethal malignancies worldwide. A pulmonary nodule is a small growth (commonly round or oval shape) in the lung, and it has the risk of becoming a site of cancerous tissue. Our experimental results on the public LIDC-IDRI dataset [1] suggest that the Nodule-plus R-CNN proposed by us achieves state-of-the-art best performance of pulmonary nodule instance segmentation with respect to both the Dice and TP Dice criteria; further, our weakly-supervised segmentation approach (Deep Self-paced Active Learning) is much more effective than common fully supervised methods [9], [20] and other best-known weakly labeled methods [9] when using the same amount of annotated samples. We extend our method in [21] in the following ways: (1) proposing a novel deep region-based network (Nodule-plus R-CNN) to improve the state-of-the-art pulmonary nodule instance-level segmentation in [21], (2) evaluating and further analyzing the impacts of different SPL schemes, (3) providing a detailed description of our DSAL strategy, and (4) presenting additional discussions of the experimental results that were not included in the conference version [21]
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
