Prediction of Ultrashort Echo Time MRI-Based Radiomics for Histological Subtypes of Non-Small Cell Lung Cancer
10.3969/j.issn.1005-5185.2024.07.008
- VernacularTitle:超短回波时间MRI影像组学对非小细胞肺癌组织学亚型的预测价值
- Author:
Fan YANG
1
;
Jing ZHAO
;
Jing WANG
;
Yun LI
Author Information
1. 郑州市第七人民医院磁共振科,河南 郑州 450000
- Keywords:
Carcinoma,non-small-cell lung;
Magnetic resonance imaging;
Radiomics;
Histological subtypes;
Forecasting
- From:
Chinese Journal of Medical Imaging
2024;32(7):682-685,691
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To explore the predictive ability of an ultrashort echo time magnetic resonance imaging(UTE-MRI)based radiomic model for histological subtypes of non-small cell lung cancer.Materials and Methods The imaging data of 67 non-small cell lung cancer patients who underwent UTE-MRI at the 7th People's Hospital of Zhengzhou from February to December 2022 were retrospectively analyzed,and radiomic features were also extracted.Least absolute shrinkage and selection operator and SelectKBest were used for histological feature screening.Logistic regression analysis and receiver operating characteristic curve were used for the development of prediction model and the assessment of diagnostic performance,respectively.Bootstrap(1 000 samples)and calibration curves were used for validation of the prediction model.Results One gray-level run-length matrix feature,one neighborhood gray-tone difference matrix feature and three gray-level size-zone matrix features were screened for the development of the prediction model.The receiver operating characteristic curve showed that the model was able to discriminate between squamous cell carcinoma and adenomatous carcinoma with an area under the curve of 0.903(95%CI 0.806-0.962),the sensitivity and specificity of 88.64%and 78.26%,respectively.The model also showed high performance in the Bootstrap-based validation,with an area under the curve of 0.882(95%CI 0.858-0.896),while the calibration curve showed good agreement between the model's predictions and actual observations.Conclusion A prediction model based on the radiomic features of UTE-MRI can effectively discriminate between squamous cell carcinoma and adenomatous carcinoma of the lung,and is expected to provide a new option for non-invasive evaluation of preoperative histological subtypes in non-small cell lung cancer.
