A general approach for automatic segmentation of pneumonia, pulmonary nodule, and tuberculosis in CT images
Lu Wang,
He Zhou,
Nan Xu,
Yuchan Liu,
Xiran Jiang,
Shu Li,
Chaolu Feng,
Hainan Xu,
Kexue Deng,
Jiangdian Song
Affiliations
Lu Wang
Department of Library, Shengjing Hospital of China Medical University, Shenyang, Liaoning 110004, China; School of Health Management, China Medical University, Shenyang, Liaoning 110122, China
He Zhou
School of Health Management, China Medical University, Shenyang, Liaoning 110122, China
Nan Xu
School of Health Management, China Medical University, Shenyang, Liaoning 110122, China
Yuchan Liu
Department of Radiology, The First Affiliated Hospital of University of Science and Technology of China (USTC), Division of Life Sciences and Medicine, USTC Hefei, Anhui 230036, China
Xiran Jiang
School of Intelligent Medicine, China Medical University, Shenyang, Liaoning 110122, China
Shu Li
School of Health Management, China Medical University, Shenyang, Liaoning 110122, China
Chaolu Feng
Key Laboratory of Intelligent Computing in Medical Image (MIIC), Ministry of Education, Shenyang, Liaoning 110169, China
Hainan Xu
Department of Obstetrics and Gynecology, Pelvic Floor Disease Diagnosis and Treatment Center, Shengjing Hospital of China Medical University, Shenyang, Liaoning 110004, China; Corresponding author
Kexue Deng
Department of Radiology, The First Affiliated Hospital of University of Science and Technology of China (USTC), Division of Life Sciences and Medicine, USTC Hefei, Anhui 230036, China; Corresponding author
Jiangdian Song
School of Health Management, China Medical University, Shenyang, Liaoning 110122, China; Corresponding author
Summary: Proposing a general segmentation approach for lung lesions, including pulmonary nodules, pneumonia, and tuberculosis, in CT images will improve efficiency in radiology. However, the performance of generative adversarial networks is hampered by the limited availability of annotated samples and the catastrophic forgetting of the discriminator, whereas the universality of traditional morphology-based methods is insufficient for segmenting diverse lung lesions. A cascaded dual-attention network with a context-aware pyramid feature extraction module was designed to address these challenges. A self-supervised rotation loss was designed to mitigate discriminator forgetting. The proposed model achieved Dice coefficients of 70.92, 73.55, and 68.52% on multi-center pneumonia, lung nodule, and tuberculosis test datasets, respectively. No significant decrease in accuracy was observed (p > 0.10) when a small training sample size was used. The cyclic training of the discriminator was reduced with self-supervised rotation loss (p < 0.01). The proposed approach is promising for segmenting multiple lung lesion types in CT images.
