Deep Learning-based Automatic Detection Algorithm for Reducing Overlooked Lung Cancers on Chest Radiographs

被引:53
作者
Jang, Sowon [1 ]
Song, Hwayoung [1 ]
Shin, Yoon Joo [2 ]
Kim, Junghoon [1 ]
Kim, Jihang [1 ]
Lee, Kyung Won [1 ,3 ]
Lee, Sung Soo [1 ]
Lee, Woojoo [4 ]
Lee, Seungjae [5 ]
Lee, Kyung Hee [1 ]
机构
[1] Seoul Natl Univ, Bundang Hosp, Dept Radiol, 300 Gumi Dong, Seongnam Si 13620, Gyeonggi Do, South Korea
[2] Konkuk Univ, Med Ctr, Dept Radiol, Seoul, South Korea
[3] Seoul Natl Univ, Coll Med, Inst Radiat Med, Med Res Ctr, Seoul, South Korea
[4] Seoul Natl Univ, Grad Sch Publ Hlth, Dept Publ Hlth Sci, Seoul, South Korea
[5] Seoul Natl Univ, Grad Sch Convergence Sci & Technol, Dept Transdisciplinary Studies, Program Biomed Radiat Sci, Seoul, South Korea
关键词
COMPUTER-AIDED DETECTION;
D O I
10.1148/radiol.2020200165
中图分类号
R8 [特种医学]; R445 [影像诊断学];
学科分类号
1002 ; 100207 ; 1009 ;
摘要
Background: It is uncertain whether a deep learning-based automatic detection algorithm (DLAD) for identifying malignant nodules on chest radiographs will help diagnose lung cancers. Purpose: To evaluate the efficacy of using a DLAD in observer performance for the detection of lung cancers on chest radiographs. Materials and Methods: Among patients diagnosed with lung cancers between January 2010 and December 2014, 117 patients (median age, 69 years; interquartile range [IQR], 64-74 years; 57 women) were retrospectively identified in whom lung cancers were visible on previous chest radiographs. For the healthy control group, 234 patients (median age, 58 years; IQR, 48-68 years; 123 women) with normal chest radiographs were randomly selected. Nine observers reviewed each chest radiograph, with and without a DLAD. They detected potential lung cancers and determined whether they would recommend chest CT for follow-up. Observer performance was compared with use of the area under the alternative free-response receiver operating characteristic curve (AUC),sensitivity, and rates of chest CT recommendation. Results: In total, 105 of the 117 patients had lung cancers that were overlooked on their original radiographs. The average AUC for all observers significantly rose from 0.67 (95% confidence interval [CI]: 0.62, 0.72) without a DLAD to 0.76 (95% CI: 0.71, 0.81)with a DLAD (P<001). With a DLAD, observers detected more overlooked lung cancers (average sensitivity, 53% [56 of 105 patients] with a DLAD vs 40% [42 of 105 patients] without a DLAD) (P<.001) and recommended chest CT for more patients(62% [66 of 105 patients] with a DLAD vs 47% [49 of 105 patients] without a DLAD) (P<001). In the healthy control group,no difference existed in the rate of chest CT recommendation (10% [23 of 234 patients] without a DLAD and 8% [20 of 234 patients]with a DLAD) (P =.13). Conclusion: Using a deep learning-based automatic detection algorithm may help observers reduce the number of overlooked lung cancers on chest radiographs, without a proportional increase in the number of follow-up chest CT examinations. (C) RSNA, 2020
引用
收藏
页码:652 / 661
页数:10
相关论文
共 20 条
[1]   updated statistics of lung and bronchus cancer in United States (2018) (vol 10, pg 1158, 2018) [J].
Boloker, Gabrielle ;
Wang, Cong ;
Zhang, Jianrong .
JOURNAL OF THORACIC DISEASE, 2019, 11 (03) :E63-E63
[2]   A status report on free-response analysis [J].
Chakraborty, D. P. .
RADIATION PROTECTION DOSIMETRY, 2010, 139 (1-3) :20-25
[3]   Computer-aided Detection of Lung Cancer on Chest Radiographs: Effect on Observer Performance [J].
de Hoop, Bartjan ;
De Boo, Diederik W. ;
Gietema, Hester A. ;
van Hoorn, Frans ;
Mearadji, Banafsche ;
Schijf, Laura ;
van Ginneken, Bram ;
Prokop, Mathias ;
Schaefer-Prokop, Cornelia .
RADIOLOGY, 2010, 257 (02) :532-540
[4]   Missed lung cancer: when, where, and why? [J].
del Ciello, Annemilia ;
Franchi, Paola ;
Contegiacomo, Andrea ;
Cicchetti, Giuseppe ;
Bonomo, Lorenzo ;
Larici, Anna Rita .
DIAGNOSTIC AND INTERVENTIONAL RADIOLOGY, 2017, 23 (02) :118-126
[5]   RECEIVER OPERATING CHARACTERISTIC RATING ANALYSIS - GENERALIZATION TO THE POPULATION OF READERS AND PATIENTS WITH THE JACKKNIFE METHOD [J].
DORFMAN, DD ;
BERBAUM, KS ;
METZ, CE .
INVESTIGATIVE RADIOLOGY, 1992, 27 (09) :723-731
[6]  
Expert Panel on Thoracic Imaging:, 2018, J Am Coll Radiol, V15, pS341, DOI 10.1016/j.jacr.2018.09.025
[7]  
Fitzmaurice C, 2017, JAMA ONCOL, V3, P524, DOI [10.1001/jamaoncol.2016.5688, 10.1001/jamaoncol.2018.2706]
[8]   A comparison of denominator degrees of freedom methods for multiple observer ROC analysis [J].
Hillis, Stephen L. .
STATISTICS IN MEDICINE, 2007, 26 (03) :596-619
[9]   Development and Validation of a Deep Learning-Based Automated Detection Algorithm for Major Thoracic Diseases on Chest Radiographs (vol 2, e191095, 2019) [J].
Hwang, E. J. ;
Park, S. ;
Jin, K-N .
JAMA NETWORK OPEN, 2019, 2 (04)
[10]   The Effect of Computer-aided Detection on Radiologist Performance in the Detection of Lung Cancers Previously Missed on a Chest Radiograph [J].
Kligerman, Seth ;
Cai, Ling ;
White, Charles S. .
JOURNAL OF THORACIC IMAGING, 2013, 28 (04) :244-252