Auto Detection of Lung Ground-Glass Opacity Nodules Based on High-Pass Filter and Gaussian Mixture Model

作者：Cao, Lei; Wang, Kai; Xing, Qianqian; Lin, Bingquan; Zhang, Yu^*

来源：Journal of Medical Imaging and Health Informatics, 2016, 6(2): 320-327.

DOI：10.1166/jmihi.2016.1700

• Auto Detection
• Gaussian Mixture Model
• Ground-Glass Opacity Nodules
• High-Pass Filter

The automatic detection for ground-glass opacity (GGO) nodule is exceedingly difficult because of its low contrast and poorly defined boundary. To date, most researches on GGO detection have conducted hybrid detecting solution of threshold, filter, and classifier on private datasets. However, the lack of a gold standard makes the evaluation of these methods difficult. This study provides a different perspective. Considering that the primary aim of the study is to achieve assisted detection, we evaluate our method by comparing our results with the judgments of experts, based on the public database Lung Image Database Consortium (LIDC). First, we extract the GGO candidate regions using the Gaussian mixture model and high-pass filter, and then reduce false positives by applying shape filtering and local feature analysis. Finally, good detection results on 49 GGO nodules of LIDC are achieved. Compared with each expert, the consistency is more than 0.8, some even exceeds 0.9, indicating that the proposed detection method may be of considerable assistance to radiologists.

• 出版日期2016-4
• 单位南方医科大学