- 期刊
- OpenAccess
The Efficacy of Artificial Intelligence in Detecting Lung Nodules on Low-Dose Computed Tomograms
摘要
PURPOSE. This study aimed to investigate the performance of vessel-suppressed computer-aided detection (VS-CAD) artificial intelligence (AI) in detecting and characterizing lung nodules on low-dose computed tomography (LDCT) images. MATERIALS AND METHODS. From September 2014 to July 2019, 70 patients with 111 lung nodules detected by LDCT who subsequently underwent pulmonary wedge resection were enrolled. The pre-operative LDCT images of the 70 patients were retrospectively and separately interpreted by a third-year radiology resident and a senior thoracic radiologist by using both the traditional method and an AI-based method. The AI-based system consisted of a vessel-suppression function for producing a vessel-suppressed (VS) image and a deep-learning-based VS-CAD analyzer. In the AI-based method, the VS-LDCT images were used as the primary source for interpretation and the original LDCT images were used for reference. The reading time and detected target nodules for each reading mode were recorded for comparison. RESULTS. A total of 111 nodules from 70 patients were confirmed by surgical pathology. All the 111 nodules were successfully detected by the two observers using the traditional method. Only two benign nodules featuring as ground glass nodules were not detected on the VS-LDCT images by either of the two observers. The reading time was significantly shorter by using the AI-based method than the traditional method (mean ± SD: 115.15 ± 73.50 vs. 301.90 ± 138.47 seconds, p ＜ 0.001). CONCLUSION. The AI system, which was composed of a VS-CAD analyzer, allows easier and significantly faster nodule detection via VS-LDCT images. AI is beneficial to radiologists in detecting lung nodules on LDCT images.
延伸閱讀
- SWEETLIN, D., NEHEMIAH, K., & KANNAN (2016). Patient-Specific Model Based Segmentation of Lung Computed Tomographic Images. Journal of Information Science and Engineering, 32(5), 1373-1394. https://www.airitilibrary.com/Article/Detail?DocID=10162364-201609-201703280016-201703280016-1373-1394
- Tsai, Y. S. (2011). Implementation of Lung Extraction Based on the Structure of Lung Nodule Computer Aided Diagnosis System [master's thesis, Yuan Ze University]. Airiti Library. https://www.airitilibrary.com/Article/Detail?DocID=U0009-2801201414592403
- 施穎銘、林聖皓（2014）。Screening for Lung Cancer with Low-Dose Computed Tomography。內科學誌，25(6)，403-409。https://doi.org/10.6314/JIMT.2014.25(6).03
- 白宗庭、王昱筌、彭鈺瓊、林招膨、黃慶順（2019）。To Compare Interpretation of Personal and Computer-Aided Detection (CAD) Detect of Lung Nodule in Low-Dose Computed Tomography Scan。台灣應用輻射與同位素雜誌，15(1)，1683-1689。https://www.airitilibrary.com/Article/Detail?DocID=18156290-201903-201905090016-201905090016-1683-1689
- 陳逸雯（2002）。Computer-aided Diagnosis System for Lung Nodule Detection〔碩士論文，中原大學〕。華藝線上圖書館。https://doi.org/10.6840/cycu200200133