A ground-glass density nodule (GGN) is a circumscribed area of increased pulmonary attenuation with preservation of the bronchial and vascular margins. In broad sense, GGN indicating sub-solid nodule, should include part solid (with solid component) and non-solid (pure GGN without solid component). The etiology of GGN include non-neoplastic (infection, inflammation, hemorrhage, etc.), neoplastic diseases (primary lung malignancy, rarely metastasis, etc.), stage 0 adenocarcinoma in situ (AIS) and preinvasive lesion such as atypical adenomatous hyperplasia (AAH). The challenges of radiologists in the diagnosis of these small GGN, mainly due to the limitation of CT resolution (above 500 µm) compared to microscopic resolution (<1µm). The current international guidelines (Lung RADS, LU-RADS, Fleischner society guideline) for lung nodule(s) are based on their size and characteristic as well as the follow up information. Based on the CT detected nodules, measurement of the mean diameter of the whole nodule, solid component and interval change of the nodule size is the most common clinically practical method. CT image-based features combined with patient-based features is considered an important prediction model in personalized medicine. Taiwan Lung Cancer Screening in Never Smoker Trial (TALENT) showed a lung cancer detection rate 2.6% (313/12011) in nonsmoker population with the prevalence of lung cancer with family history 3.2%. 96.5% of the confirmed lung cancer were stage 0-1. An accurate detection, precise measurement and characterization of lung nodules are very important for patient management and follow up strategy in LDCT lung cancer screening. We analyzed 100 surgically resected GGNs (January 2019-March 2019), including 31 men and 69 women (mean age 57.6 years), average size 7.7±2.8mm, 6 benign, 6 AAH, 21 AIS, 33 MIA, 34 invasive adenocarcinoma (IA) (5 lepidic, 20 acinar, 7 papillary, 1 no predominant subtype, 1 mucinous). Benign and preinvasive lesion account for 33%, MIA 33%, IA 34%. Modified Taiwan guideline for the risk prediction and nodule management should be performed to avoid overdiagnosis and overtreatment.