Preliminary Study on the Predictive Value of CT Enhanced Image Radiomics Feature Model
for Lung Squamous Cell Carcinoma and Adenocarcinoma
TANG Caiyin1, LI Tong1, DUAN Shaofeng2, ZHANG Ji1
1. Department of Imaging, Taizhou People’s Hospital, Taizhou Jiangsu 225300, China;
2. GE Healthcare Precision Health Institution, Nanjing Jiangsu 210000, China
Abstract：Objective To evaluate the ability of CT enhanced image radiomics in differentiating lung adenocarcinoma (ADC) and
lung squamous cell carcinoma (SCC). Methods 51 patients with ADC and 34 patients with SCC confirmed by pathology in Taizhou
People’s Hospital from January 2017 to December 2019 were retrospectively analyzed. The radiomics features were extracted
from the region of interest of enhanced CT images. According to the ratio of 7:3, 59 patients were selected as the training set and
26 patients as the verification set. Correlation test, one-way ANOVA or rank sum test, one-way logistic regression analysis and
multi factor random forest cross validation were used to select features and reduce dimensions, and multi-factor logistic regression
method and Bayesian prediction model were used to compare and evaluate the effectiveness of the model. The effectiveness of the
model (sensitivity, specificity, accuracy) was evaluated by the area under the curve (AUC) of receiver operating characteristic (ROC)
curve. Results 8 features were modeled by logistic regression analysis, and the ROC curve showed that the AUC of training set was
0.97, the sensitivity was 83.3%, the specificity was 97.1%, and the accuracy was 91.5%. The AUC of validation set was 0.89, the
sensitivity was 80.2%, the specificity was 73.3%, and the accuracy was 84.6%. Conclusion The application of radiomics method
based on logistic regression to predict and differentiate SCC and ADC has high value, which provides scientific basis for clinicians to
make decisions.
唐彩银1，李通1，段绍峰2，张继1. 基于CT增强图像影像组学特征模型预测肺鳞癌和腺癌价值初探[J]. 中国医疗设备, 2022, 37(3): 138-141.
TANG Caiyin1, LI Tong1, DUAN Shaofeng2, ZHANG Ji1. Preliminary Study on the Predictive Value of CT Enhanced Image Radiomics Feature Model
for Lung Squamous Cell Carcinoma and Adenocarcinoma. China Medical Devices, 2022, 37(3): 138-141.
