Related Experiment Video
Updated: Aug 27, 2025
08:05
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
14.3K
A deep learning-based post-processing method for automated pulmonary lobe and airway trees segmentation using chest
Haiqun Xing1, Xin Zhang2, Yingbin Nie2
1Department of Nuclear Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Science & Peking Union Medical College, Beijing Key Laboratory of Molecular Targeted Diagnosis and Therapy in Nuclear Medicine, Beijing, China.
Quantitative Imaging in Medicine and Surgery
|October 3, 2022
Summary
This study presents an automated deep learning model combined with post-processing for segmenting lung lobes and airways in PET/CT scans, improving disease localization accuracy.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Radiology
Background:
- Accurate localization of lung disease is crucial for effective treatment.
- Chest computed tomography (CT) images acquired during positron emission tomography/computed tomography (PET/CT) scans are vital for diagnosis.
- Segmenting pulmonary anatomical regions, including lobes and airways, aids in disease assessment.
Purpose of the Study:
- To develop and validate an automated deep learning model combined with post-processing for segmenting six pulmonary anatomical regions.
- To enhance the accuracy of lung disease localization using CT imaging from PET/CT scans.
- To evaluate the performance of the combined model in segmenting five pulmonary lobes and airway trees.
Main Methods:
- A retrospective cohort of 640 patients undergoing PET/CT imaging was enrolled.
- A convolutional neural network (CNN) with DenseVNet architecture and post-processing algorithms was employed for pulmonary segmentation.
- Performance was assessed using Dice coefficient, Hausdorff distance, and Jaccard coefficient, comparing the combined model against ground truth.
Main Results:
- The combined deep learning and post-processing model demonstrated superior performance compared to a single deep learning model.
- In the test set, the Dice coefficient for all lobes was 0.972, and for airway trees was 0.849.
- Excellent agreement was observed between the model's segmentations and ground truth across all regions.
Conclusions:
- The proposed model effectively segments five pulmonary lobes and airway trees on chest CT images from PET/CT scans.
- The combined approach significantly improves segmentation accuracy over deep learning alone.
- This automated segmentation supports more precise lung disease localization.
