Histological Classification and Invasion Prediction of Thymoma by Machine Learning-Based Computed Tomography Imaging. (5th August 2022)

• Record Type: Journal Article
• Authors: Wang, Danfeng; Zhang, Yiwei; Li, Bingli; Zhuang, Qiaowei; Zhang, Xiaoqin; Lin, Daiying
• Other Names: Chumnanvej Sorayouth Academic Editor.
• Is Part Of: Contrast media & molecular imaging. Volume 2022(2022)
• Publication Date: 2022-08-05
• Subjects: Diagnostic imaging -- Periodicals; Magnetic resonance imaging -- Periodicals; Contrast media (Diagnostic imaging) -- Periodicals; Contrast Media -- Periodicals; Diagnostic Imaging -- Periodicals; Substances de contraste -- Périodiques; Diagnostics moléculaires -- Périodiques; Imagerie médicale; Substance de contraste; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.0754
• DOI: 10.1155/2022/4594757
• Languages: English
• ISSNs: 1555-4309

Abstract

Abstract : Purpose . The values of machine learning-based computed tomography (CT) imaging in histological classification and invasion prediction of thymoma were investigated. Methods . 181 patients diagnosed with thymoma by surgery or biopsy in Shantou Central Hospital between February 2017 and March 2022 were selected. According to the concept of simplified histological classification and the latest histological classification by the WHO, thymoma was divided into two groups, including low-risk (types A, AB, B1, and metaplastic type) and high-risk groups (types B2 and B3). CT images were reconstructed by filtering back projection (FBP) algorithm. CT image features were collected for statistical analysis. Results . The main symptoms of patients diagnosed with thymoma included respiratory tract infection, chest distress and shortness of breath, and chest pain. 35.91% of them suffered from complicated myasthenia gravis. Tumor size and position in low-risk and high-risk groups showed no statistical significance (P > 0.05 ). Tumor morphology and boundary between the two groups suggested statistical difference (P < 0.05 ). Whether tumor invaded adjacent tissues was apparently correlated with simplified histological classification (P < 0.01 ). The sensitivity and specificity of CT images for the invasion of mediastinal pleura or pericardium were around 90% and negative predictive values both reached above 95%. Those of the CT images for lung invasion were over 80%. The negative and positive predictive values were 93.54% and 63.82%, respectively. Those of the CT images for blood vessel invasion were 67.32% and 97.93%. The negative and positive predictive values were 98.21% and 83%, respectively. Conclusion . The machine learning-based CT image had significant values in the prediction of different histological classification and even invasion level.