Radiomics Nomogram Analyses for Differentiating Pneumonia and Acute Paraquat Lung Injury

• Published in: Scientific reports Vol. 9; no. 1; pp. 15029 - 9
• Main Authors: Yanling, Wang, Duo, Gao, Zuojun, Geng, Zhongqiang, Shi, Yankai, Wu, Shan, Lu, Hongying, Cui
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 21.10.2019; Nature Publishing Group
• Subjects: 639/705; 692/499; Acute Disease; Adult; Algorithms; Aspartate transaminase; Body temperature; Calibration; Clinical Decision-Making; Cohort Studies; Diagnosis, Differential; Female; Humanities and Social Sciences; Humans; Laboratories; Laboratory tests; Lungs; Male; multidisciplinary; Nausea; Nomograms; Paraquat; Patients; Pneumonia; Pneumonia - diagnosis; Pneumonia - diagnostic imaging; Poisoning; Prediction models; Public health; Pulmonary Edema - diagnosis; Pulmonary Edema - diagnostic imaging; Radiomics; Reproducibility of Results; Risk factors; ROC Curve; Science; Science (multidisciplinary); Suicide; Tomography, X-Ray Computed; Transaminase; Vomiting
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-019-50886-7

Paraquat poisoning has become a serious public health problem in some Asian countries because of misuse or suicide. We sought to develop and validate a radiomics nomogram incorporating radiomics signature and laboratory bio-markers, for differentiating bacterial pneumonia and acute paraquat lung injury. 180 patients with pneumonia and acute paraquat who underwent CT examinations between December 2014 and October 2017 were retrospectively evaluated for testing and validation. Clinical information including demographic data, clinical symptoms and laboratory test were also recorded. A prediction model was built by using backward logistic regression and presented on a nomogram. The radiomics-based features yielded areas under the receiver operating characteristic curve of 0.870 (95% CI 0.757–0.894), sensitivity of 0.857, specificity of 0.804, positive predictive value of 83.3%, negative predictive value of 0.818 in the primary cohort, while in the validation cohort the model showed similar results (0.865 (95% CI 0.686–0.907), 0.833, 0.792, 81.5%, respectively). The individualized nomogram included radiomics signature, body temperature, nausea and vomiting, and aspartate transaminase. We have developed a radiomics nomogram that combination of the radiomics features and clinical risk factors to differentiate paraquat lung injury and pneumonia for patients with an unclear medical history of exposure to paraquat poisoning, providing appropriate therapy decision support.