Biomedicines 2022, 10, 2839. https://doi.org/10.3390/biomedicines10112839 www.mdpi.com/journal/biomedicines Article Artificial Intelligence for Early Detection of Chest Nodules in X-ray Images Hwa-Yen Chiu 1,2,3,4 , Rita Huan-Ting Peng 2 , Yi-Chian Lin 2 , Ting-Wei Wang 2 , Ya-Xuan Yang 2 , Ying-Ying Chen 1,5 , Mei-Han Wu 4,6,7 , Tsu-Hui Shiao 1,4 , Heng-Sheng Chao 1,8 , Yuh-Min Chen 1,4 and Yu-Te Wu 2,9, * 1 Department of Chest Medicine, Taipei Veterans General Hospital, Taipei 112, Taiwan 2 Institute of Biophotonics, National Yang Ming Chiao Tung University, Taipei 112, Taiwan 3 Division of Internal Medicine, Hsinchu Branch, Taipei Veterans General Hospital, Hsinchu 310, Taiwan 4 School of Medicine, National Yang Ming Chiao Tung University, Taipei 112, Taiwan 5 Department of Critical Care Medicine, Taiwan Adventist Hospital, Taipei 105, Taiwan 6 Department of Medical Imaging, Cheng Hsin General Hospital, Taipei 112, Taiwan 7 Department of Radiology, Taipei Veterans General Hospital, Taipei 112, Taiwan 8 Institute of Biomedical Informatics, National Yang Ming Chiao Tung University, Taipei 112, Taiwan 9 Brain Research Center, National Yang Ming Chiao Tung University, Taipei 112, Taiwan * Correspondence: ytwu@nycu.edu.tw; Tel.: +886-2-2826-7000 Abstract: Early detection increases overall survival among patients with lung cancer. This study formulated a machine learning method that processes chest X-rays (CXRs) to detect lung cancer early. After we preprocessed our dataset using monochrome and brightness correction, we used different kinds of preprocessing methods to enhance image contrast and then used U-net to perform lung segmentation. We used 559 CXRs with a single lung nodule labeled by experts to train a You Only Look Once version 4 (YOLOv4) deep-learning architecture to detect lung nodules. In a testing dataset of 100 CXRs from patients at Taipei Veterans General Hospital and 154 CXRs from the Jap- anese Society of Radiological Technology dataset, the sensitivity of the AI model using a combina- tion of different preprocessing methods performed the best at 79%, with 3.04 false positives per image. We then tested the AI by using 383 sets of CXRs obtained in the past 5 years prior to lung cancer diagnoses. The median time from detection to diagnosis for radiologists assisted with AI was 46 (3–523) days, longer than that for radiologists (8 (0–263) days). The AI model can assist radiolo- gists in the early detection of lung nodules. Keywords: artificial intelligence; AI; detection; lung cancer; machine learning 1. Introduction Lung cancer is the leading cause of death in patients with neoplasm in Taiwan. Pa- tients with lung cancer are usually asymptomatic or have nonspecific complaints; how- ever, more than half of initial lung cancer diagnoses are stage IIIB or higher, meaning the tumors are unresectable [1]. One method of improving lung cancer survival rate is early screening. Both the National Lung Screening Trial (NLST) and the Dutch–Belgian Ran- domized Lung Cancer Screening (NELSON) trial revealed that early detection of lung cancer resulted in a 20% improvement in overall survival [2,3]. Therefore, effective screen- ing tools for early diagnosis of lung cancer warrant investigation. The standard procedures for lung cancer screening are chest X-rays (CXRs) and low- dose chest computed tomography (LDCT). LDCT is currently the most powerful diagnos- tic tool, with a resolution as fine as 1 mm, whereas nodules must be larger than 1 cm to be detectable on a CXR [4]. Several studies have investigated nodule detection in chest computed tomography (CT) images and found that artificial intelligence (AI) outperforms humans in this field [5]. The incidence of lung cancer was approximately 6 cases per 1000 person-years in both the NLST and the NELSON trials [2,3]. Active LDCT screening is Citation: Chiu, H.-Y.; Peng, R.H.-T.; Lin, Y.-C.; Wang, T.-W.; Yang, Y.-X.; Chen, Y.-Y.; Wu, M.-H.; Shiao, T.-H.; Chao, H.-S.; Chen, Y.-M.; et al. Artificial Intelligence for Early Detection of Chest Nodules in X-ray Images. Biomedicines 2022, 10, 2839. https://doi.org/10.3390/ biomedicines10112839 Academic Editor: Ryota Niikura Received: 17 October 2022 Accepted: 4 November 2022 Published: 7 November 2022 Publisher’s Note: MDPI stays neu- tral with regard to jurisdictional claims in published maps and institu- tional affiliations. Copyright: © 2022 by the authors. Li- censee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and con- ditions of the Creative Commons At- tribution (CC BY) license (https://cre- ativecommons.org/licenses/by/4.0/).