AUTHOR=Hu Yuntao , Liu Nian , Tang Lingling , Liu Qianqian , Pan Ke , Lei Lixing , Huang Xiaohua 

TITLE=Three-Dimensional Radiomics Features of Magnetic Resonance T2-Weighted Imaging Combined With Clinical Characteristics to Predict the Recurrence of Acute Pancreatitis

JOURNAL=Frontiers in Medicine

VOLUME=Volume 9 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2022.777368

DOI=10.3389/fmed.2022.777368

ISSN=2296-858X

ABSTRACT=Objective: To explore the diagnostic value of machine learning model based on magnetic resonance T2-weighted imaging for predicting the recurrence of acute pancreatitis.
Methods: We retrospectively collected 190 patients with acute pancreatitis (AP), including 122 patients with initial acute pancreatitis and 68 patients with recurrent acute pancreatitis (RAP). At the same time, the clinical characteristics of the two groups were collected. They were randomly divided into training group and validation group in the ratio of 7:3. 134 cases in the training group, including 86 cases of AP and 48 cases of RAP. There were 56 cases in the validation group, including 36 cases of AP and 20 cases of RAP. Least absolute shrinkage and selection operator (LASSO) were used for feature screening. Logistic regression was used to establish the radiomics model, clinical model and combined model for predicting AP recurrence. The predictive ability of the three models was evaluated by the area under the curve (AUC). The recurrence risk in patients with AP was assessed using the nomogram.
Results: The AUCs of radiomics model in training group and validation group were 0.804 and 0.788, respectively. The AUCs of the combined model in the training group and the validation group were 0.833 and 0.799, respectively. The AUCs of the clinical model in training group and validation group were 0.677 and 0.572, respectively. The sensitivities of the radiomics model, combined model, and clinical model were 0.646, 0.691 and 0.765, respectively. The specificities of the radiomics model, combined model, and clinical model were 0.791, 0.828 and 0.590, respectively. There was no significant difference in AUC between the radiomics model and the combined model for predicting RAP (p =0.067). The AUCs of the radiomics model and combined model were greater than those of the clinical model (p =0.008 and p =0.007, respectively.).
Conclusions: Radiomics features based on magnetic resonance T2WI could be used as biomarkers to predict the recurrence of AP, and radiomics model and combined model can provide new directions for predicting recurrence of acute pancreatitis.