Mammography Based Nomogram Integrating Radiomics and Clinical Features to Predict Benign or Malignant Regression of
Guoyan Yao1, Lijun Chen1, Tingfan Wu2
1Department of Radiology, The Second Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, Guangdong Province, 510120, People's Republic of China.
Purpose:
The study aimed to develop a nomogram based on mammography radiomic and clinical features to predict the benign and malignant progression of BI-RADS 4a lesions under follow-up.
Materials And Methods:
The retrospective study included 104 patients with over six months of follow-up, consisting of 56 malignant and 48 benign cases, totaling 202 images. Patients were randomly divided into training and validation sets at a 7:3 ratio. In total, 1316 radiomic features were extracted using AK3.30 software including morphological, first-order statistics and texture features. Spearman correlation analysis and the least absolute shrinkage and selection operator (LASSO) method were performed for feature selection. Univariate and multivariate logistic regression analyses were used to identify independent risk factors among clinical features and construct a radiomic-clinical fusion nomogram. The performance of radiomics model and radiomic-clinical fusion model were evaluated using the area under the receiver operating characteristic (ROC) curve. DeLong test was employed to compare the efficacy between the two models.
Results:
Four radiomic features were selected, combined with two clinical features (positive clinical palpation and history of breast surgery). The AUC values for the radiomics model and radiomic-clinical fusion model in the training and testing groups were 0.858 and 0.860, and 0.923 and 0.904, respectively. The DeLong test showed no significant difference between the two models with a P value > 0.05.
Conclusion:
The nomogram based on mammography radiomics and clinical features demonstrated good performance in predicting the benign and malignant progression of BI-RADS 4a lesions under follow-up, showing potential for risk stratification of BI-RADS 4a lesions.
