Computerized Identification of Bronchiectasis Using a 3D Quantitative CT Protocol

Objective: This study proposes a three-dimensional automatic approach to identify bronchiectasis. The performance of the quantitative CT protocol is evaluated on bronchiectasis patients. Methods: Chest CT images from 121 patients with bronchiectasis were analyzed using a quantitative CT protocol. The radiologist features were determined by the extent of bronchiectasis; the degree of bronchiectasis; and the time costs by comparing the discrepancy between the radiologists' results and that of computerized scheme. Results: The mean score of bronchiectasis extent between the computerized scheme and the radiologists were 0.10±2.30 and 0.33±1.25 (r = 0.78, p = .00) respectively. When the degree of bronchiectasis was considered, the value of bronchial dilatation (VBD) reported by quantitative CT analyze was 2.13±3.20; and the relationship between internal diameter of dilated bronchi and that of its adjacent pulmonary artery (IDB/IDA) measured by radiologists was 1.92±2.46 (r = 0.73, p = .00). The computational cost in time of segmenting bronchiectasis was 4.50 s±0.02, which was much less than the time cost of the radiologists. (10.29 m±20.67, t = 18.78, p = .00). 24-hour sputum volume was associated with the extent of bronchiectasis determined by both the radiologists and the quantitative CT method (r = 0.64, p = .00, r = 0.58, p = .00, respectively). Conclusion: This study showed that a three-dimensional computerized scheme could achieve a reasonable performance in bronchial segmentation and detection. Three-dimensional bronchial detection can establish the relationship between the morphologic changes and clinical activity.