Efficiency of a computer-aided diagnosis (CAD) system with deep learning in detection of pulmonary nodules on 1-mm-thick
- PDF / 1,493,242 Bytes
- 10 Pages / 595.276 x 790.866 pts Page_size
- 12 Downloads / 191 Views
ORIGINAL ARTICLE
Efficiency of a computer‑aided diagnosis (CAD) system with deep learning in detection of pulmonary nodules on 1‑mm‑thick images of computed tomography Takenori Kozuka1 · Yuko Matsukubo1 · Tomoya Kadoba1 · Teruyoshi Oda1 · Ayako Suzuki1 · Tomoko Hyodo1 · SungWoon Im1 · Hayato Kaida1 · Yukinobu Yagyu1 · Masakatsu Tsurusaki1 · Mitsuru Matsuki1 · Kazunari Ishii1 Received: 30 March 2020 / Accepted: 18 June 2020 © Japan Radiological Society 2020
Abstract Purpose To evaluate the performance of a deep learning-based computer-aided diagnosis (CAD) system at detecting pulmonary nodules on CT by comparing radiologists’ readings with and without CAD. Materials and methods A total of 120 chest CT images were randomly selected from patients with suspected lung cancer. The gold standard of nodules ≥ 3 mm was established by a panel of three expert radiologists. Two less experienced radiologists read the images without and afterward with CAD system. Their reading times were recorded. Results The radiologists’ sensitivity increased from 20.9% to 38.0% with the introduction of CAD. The positive predictive value (PPV) decreased from 70.5% to 61.8%, and the F1-score increased from 32.2% to 47.0%. The sensitivity significantly increased from 13.7% to 32.4% for small nodules (3–6 mm) and from 33.3% to 47.6% for medium nodules (6–10 mm). CAD alone showed a sensitivity of 70.3%, a PPV of 57.9%, and an F1-score of 63.5%. Reading time decreased by 11.3% with the use of CAD. Conclusion CAD improved the less experienced radiologists’ sensitivity in detecting pulmonary nodules of all sizes, especially including a significant improvement in the detection of clinically important-sized medium nodules (6–10 mm) as well as small nodules (3–6 mm) and reduced their reading time. Keywords Diagnosis · Computer assisted · Deep learning · Multiple pulmonary nodules · Multidetector computed tomography
* Takenori Kozuka [email protected]
Hayato Kaida [email protected]
Yuko Matsukubo [email protected]
Yukinobu Yagyu y‑[email protected]
Tomoya Kadoba [email protected]
Masakatsu Tsurusaki [email protected]‑net.ne.jp
Teruyoshi Oda [email protected]
Mitsuru Matsuki rad053@osaka‑med.ac.jp
Ayako Suzuki [email protected]
Kazunari Ishii [email protected]
Tomoko Hyodo [email protected]‑u.ac.jp
1
SungWoon Im [email protected]
Department of Radiology, Kindai University Faculty of Medicine, 377‑2 Ohno‑Higashi, Osaka‑Sayama, Osaka 589‑8511, Japan
13
Vol.:(0123456789)
Introduction Lung cancer is one of the most widespread diseases worldwide and the leading cause of cancer-related death. As reported in Global Cancer Statistics 2018, lung cancer remains the leading cause of cancer incidence and mortality worldwide, with 2.1 million new lung cancer cases and 1.8 million deaths predicted in 2018, representing close to 1 in 5 (18.4%) cancer deaths [1]. Early diagnosis is important in lung cancer practice to improve the effectiveness of treatment and increase patients’
Data Loading...
