Discriminative Localization in CNNs for Weakly-Supervised Segmentation of Pulmonary Nodules

By Xinyang Feng, Jie Yang, Andrew F. Laine, Elsa D. Angelini. 2017

166 citations recorded by the source corpus.

Automated detection and segmentation of pulmonary nodules on lung computed tomography (CT) scans can facilitate early lung cancer diagnosis. Existing supervised approaches for automated nodule segmentation on CT scans require voxel-based annotations for training, which are labor- and time-consuming to obtain. In this work, we propose a weakly-supervised method that generates accurate voxel-level nodule segmentation trained with image-level labels only. By adapting a convolutional neural network (CNN) trained for image classification, our proposed method learns discriminative regions from the activation maps of convolution units at different scales, and identifies the true nodule location with a novel candidate-screening framework. Experimental results on the public LIDC-IDRI dataset demonstrate that, our weakly-supervised nodule segmentation framework achieves competitive performance compared to a fully-supervised CNN-based segmentation method.

Subfields: Pulmonary and Respiratory Medicine, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Pulmonary and Respiratory Medicine

1. Ronald M. Summers
2. Qi-Yan Dou
3. Xiaosong Wang
4. P. Heng
5. Lequan Yu
6. Hao Chen
7. Le Lu
8. Ning Guo

In Radiology, Nuclear Medicine and Imaging

1. P. Heng
2. Hao Chen
3. Lequan Yu
4. Qi-Yan Dou
5. Jing Qin
6. Xiaosong Wang
7. Ronald M. Summers
8. Lubomir M. Hadjiiski

Citations in the corpus

• Visual Feature Attribution Using Wasserstein GANs (2017, background): “Pinpointing all evidence of a class is important for a variety of tasks such as weakly supervised localisation or segmentation of structures [41, 43, 65], and better understanding disease effects, and physiological or pathological processes in medical images [67, 18, 12, 19, 13, 27, 54, 30, 31, 63].”
• Deep learning in medical imaging and radiation therapy. (2018, background): “311,312 For example, Feng et al. 313 trained a system for lung nodule segmentation with a binary label if a nodule was present for a given image slice.”
• LViT: Language Meets Vision Transformer in Medical Image Segmentation (2022, background): “In order to address the issue of under-annotated data, some approaches have gone beyond traditional supervised learning by training their models using both labeled and more widely available unlabeled data, such as semi-supervised learning [5], [8] and weakly-supervised learning [9].”
• Embracing Imperfect Datasets: A Review of Deep Learning Solutions for Medical Image Segmentation (2019, methodology): “Feng et al. (2017) propose a 2-stage approach consisting of a coarse image segmentation followed by a fine instance-level segmentation.”
• SV-RCNet: Workflow Recognition From Surgical Videos Using Recurrent Convolutional Network (2018, background): “networks on an image and intercept the learned information of networks [37], [38].”
• Deep Learning Techniques for Medical Image Segmentation: Achievements and Challenges (2019, methodology): “For instance, creating multiple views of a patch (augmentation) rather than having a single view is proven to have a positive effect in overfitting [25].”
• Weakly-Supervised Learning-Based Feature Localization in Confocal Laser Endomicroscopy Glioma Images (2018, methodology): “Each model constructs a DFM to create a segmentation map: M1, similar to [14]; M2 – DFM and NFM of CAM 1,2, and 3 are first laterally inhibited and then collaterally integrated; M3 – CAM 1,2, and 3 are collaterally integrated; M4, M5, M6 – by laterally inhibiting the DFM and NFM of CAM 1, 2, and 3, ”
• Deep Learning-based Image Segmentation on Multimodal Medical Imaging (2019, methodology): “Further, for the task of automatic tumor segmentation, CNNs have been applied to segmentation of tumors in brain [5], [12], [13], liver [14], breast [15], lung [16], [17],”
• Transformation-Consistent Self-Ensembling Model for Semisupervised Medical Image Segmentation (2019, background): “The lack of the labeled data motivates the study of methods that can be trained with limited supervision, such as semisupervised learning [4]–[6], weakly supervised learning [7]–[9], and unsupervised domain adaptation [10]–[12].”
• Deep Learning in Medical Imaging (2019, background): “50 51-53 Since these studies have not yet surpassed the segmentation performance of supervised learning, it is considered future value as a technology that can overcome the severe imbalances in medical imaging.”