Diagnostic classification of solitary pulmonary nodules using dual time 18F-FDG PET/CT image texture features in granuloma-endemic regions

• Published in: Scientific reports Vol. 7; no. 1; pp. 1 - 8
• Main Authors: Chen, Song, Harmon, Stephanie, Perk, Timothy, Li, Xuena, Chen, Meijie, Li, Yaming, Jeraj, Robert
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 24.08.2017; Nature Publishing Group
• Subjects: 59/78; 631/114/1305; 631/67/2321; Artificial intelligence; Benign; Classification; Granuloma; Granulomas; Heterogeneity; Humanities and Social Sciences; Learning algorithms; Lung cancer; Lung diseases; Lung nodules; multidisciplinary; Positron emission tomography; Science; Science (multidisciplinary); Survival; Tomography; Visual discrimination
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-017-08764-7

Lung cancer, the most commonly diagnosed cancer worldwide, usually presents as solid pulmonary nodules (SPNs) on early diagnostic images. Classification of malignant disease at this early timepoint is critical for improving the success of surgical resection and increasing 5-year survival rates. 18 F-fluorodeoxyglucose ( 18 F-FDG) PET/CT has demonstrated value for SPNs diagnosis with high sensitivity to detect malignant SPNs, but lower specificity in diagnosing malignant SPNs in populations with endemic infectious lung disease. This study aimed to determine whether quantitative heterogeneity derived from various texture features on dual time FDG PET/CT images (DTPI) can differentiate between malignant and benign SPNs in patients from granuloma-endemic regions. Machine learning methods were employed to find optimal discrimination between malignant and benign nodules. Machine learning models trained by texture features on DTPI images achieved significant improvements over standard clinical metrics and visual interpretation for discriminating benign from malignant SPNs, especially by texture features on delayed FDG PET/CT images.