PRF-RW: a progressive random forest-based random walk approach for interactive semi-automated pulmonary lobes segmentati
- PDF / 5,542,170 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 6 Downloads / 173 Views
ORIGINAL ARTICLE
PRF‑RW: a progressive random forest‑based random walk approach for interactive semi‑automated pulmonary lobes segmentation Qiang Li1,2 · Lei Chen2 · Xiangju Li1 · Xiaofeng Lv3 · Shuyue Xia4 · Yan Kang1,5 Received: 13 April 2019 / Accepted: 25 February 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract The computational detection of lung lobes from computed tomography images is a challenging segmentation problem with important respiratory healthcare applications, including emphysema, chronic bronchitis, and asthma. This paper proposes a progressive random forest-based random walk approach for interactive semi-automated pulmonary lobes segmentation. First, our model performs automated segmentation of the lung lobes in a progressive random forest network, eliminating the need for prior segmentation of lungs, vessels, or airways. Then, an interactive lobes segmentation approach based on random walk mechanism is designed for improving auto-segmentation accuracy. Furthermore, we annotate a new dataset which contains 93 scans (57 men, 36 women; age range: 40–90 years) from the Central Hospital Affiliated with Shenyang Medical College (CHASMC). We evaluate the model on our annotated dataset, LIDC (https://wiki.cancerimagingarchive.net) and LOLA11 (http://lolall.com/) datasets. The proposed model achieved a Dice score of 0.906 ± 0.106 for LIDC, 0.898 ± 0.113 for LOLA11, and 0.921 ± 0.101 for our dataset. Experimental results show the accuracy of the proposed approach, which consistently improves performance across different datasets by a maximum of 8.2% as compared to baselines model. Keywords Random forest · Lobes segmentation · Random walk · Semi-automated segmentation · Machine learning
1 Introduction The human lung is subdivided into five lobes, which are separated from the visceral pleura known as pulmonary fissures. The right lung has three lobes, namely right upper lobe (RUL), right middle lobe (RML), and right lower lobe (RLL). The RUL and RML are divided by the right minor fissure, whereas the right major fissure delimits the lower lobe from the rest of the lung. The left lung has only two lobes, namely, left upper lobe (LUL) and left lower lobe (LLL), which are divided by the left main fissure. Figure 1 shows the five lobes separated by major and minor fissures in a coronal computed tomography (CT) slice. Each of the five * Yan Kang [email protected] 1
Northeastern University, Shenyang, Liaoning, China
2
Neusoft Medical Systems Ltd., Shenyang, Liaoning, China
3
Neusoft Corporation, Shenyang, Liaoning, China
4
The Central Hospital Affiliated to Shenyang Medical College, Shenyang, Liaoning, China
5
Shenzhen Technology University, Shenzhen, China
lobes is functionally independent with separate bronchial and vascular systems. Accurate lung lobes segmentation from chest CT images is a promising method to replace invasive methods (needlebased biopsy techniques, e.g. thoracentesis, thoracic closed drainage, and lung biopsy) [1] for the quantification
Data Loading...
