Deep neural network (DNN) is one of the most widely used forms of artificial intelligence (AI), which has revolutionized computer science and has quickly expanded into a much broader range of medical imaging1. DNN has been used to develop state-of-the-art algorithms for image recognition and it has recently been shown a wide variety of applications in different imaging subspecialties including neuro, lung, abdomen, breast, musculoskeletal and cardiac imaging2-5. However, most of the algorithms are trained in one country. It is necessary to evaluate whether the...
Methods and materials
The institutional review board of the University of Miyazaki Hospital approved our retrospective, observational study and waived the requirement for informed consent. The data were retrospectively collected from Japanese patients who underwent both non-contrast and dynamic contrast-enhanced CT scans at the University of Miyazaki Hospital. We analyzed 674 kidneys of 337 patients (201 males, 136 females; mean age 58.2 years; age range, 21-83 years). They were normal (n=301) and abnormal kidneys (n=373). The abnormal kidneys included 148 RCCs confirmed by surgery, 166 cysts,...
The accuracy, sensitivity, specificity and precision for the detection of abnormal kidneys were 76.5%, 61.7%, 95.6% and 94.6%, respectively; for RCC were 89.0%, 82.0%, 95.0%, and 94.6%, respectively. When comparing this performance with that on the US cohort, we observed similar accuracy, with a slight increase in specificity and a slight decrease in sensitivity.
The AI algorithm developed with training data of US patients showed high detectability of abnormal kidneys and RCC even in Japanese patients.
Personal information and conflict of interest
M. Azuma; Miyazaki/JP - Grant Recipient at NTT DATA Corporation N. Terada; Miyazaki/JP - Grant Recipient at NTT DATA Corporation S. Mukai; Miyazaki/JP - Grant Recipient at NTT DATA Corporation D.-A. Bunu; Tokyo/JP - nothing to disclose T. Okada; Tokyo/JP - nothing to disclose T. Kamoto; Miyazaki/JP - Grant Recipient at NTT DATA Corporation K. Araki; Miyazaki/JP - nothing to disclose T. Hirai; Miyazaki/JP - Grant Recipient at NTT DATA Corporation
1, LeCun Y, Bengio Y, Hinton G. Deep learning. Nature 2015;521(7553):436–444.
2, Ueda D, Yamamoto A, Nishimori M, Shimono T, Doishita S, Shimazaki A, Katayama Y, Fukumoto S, Choppin A, Shimahara Y, Miki Y. Deep Learning for MR Angiography: Automated Detection of Cerebral Aneurysms. Radiology. 2019;290(1):187-194.
3, Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis. Med Image Anal 2017;42:60–88.
4, Shen D, Wu G, Suk H-I. Deep learning in medical image analysis. Annu Rev Biomed Eng...