Related Experiment Video
Updated: Jan 17, 2026
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Development and Validation of Multiparametric MRI-based Clini-radiomic Model for Preoperative Prediction of
Yanghua Fan1, Wentai Zhang2, Anna Mou3
1Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing, China (Y.F., S.G.); Department of Neurosurgery, Beijing Neurosurgical Institute, Beijing, China (Y.F.).
Rationale And Objectives:
Preoperative differentiation molecular subtype is critical for tailoring management of somatotroph adenomas, as these subtypes differ markedly in invasiveness, treatment response, and recurrence risk. Current reliance on postoperative histopathology delays optimal therapeutic decision-making. This study aimed to develop a non-invasive clini-radiomic model using multiparametric MRI to preoperatively distinguish between sparsely granulated somatotroph adenomas (SGSA) and densely granulated somatotroph adenomas (DGSA).
Methods:
A total of 154 patients with pathologically confirmed somatotroph were retrospectively enrolled and randomized into a training set (n=102) and internal validation set (n=52). Radiomic features were extracted from multiparametric MRI and screened using the elastic net and recursive feature elimination algorithms to construct a fusion radiomic model. A clini-radiomic model was further developed by integrating the radiomic signature with significant clinical features. Model performance was evaluated using receiver operating characteristic curves, calibration curves, and decision curve analysis (DCA) with external validation in 58 multicenter patients.
Results:
The fusion radiomic model, based on 4 selected radiomic features, achieved areas under the curve (AUC) of 0.940 (95% CI: 0.923-0.952) and 0.880 (95% CI: 0.864-0.905) in the training and internal validation sets, respectively. The clini-radiomic model, incorporating radiomic features and random GH levels, showed improved performance with AUCs of 0.950 (95% CI: 0.939-0.964) in the training set, 0.900 (95% CI: 0.885-0.924) in the internal validation set, and 0.91 in external multicenter validation. Calibration curves and DCA confirmed good agreement between predictions and actual outcomes, with substantial clinical utility.
Conclusion:
The clini-radiomic model enables non-invasive preoperative differentiation of SGSA and DGSA, providing a reliable tool to guide personalized treatment strategies for somatotroph adenomas.
More Related Videos
12:23Multi-modal Imaging of Angiogenesis in a Nude Rat Model of Breast Cancer Bone Metastasis Using Magnetic Resonance Imaging, Volumetric Computed Tomography and Ultrasound
Published on: August 14, 2012
06:24Dynamic Contrast Enhanced Magnetic Resonance Imaging of an Orthotopic Pancreatic Cancer Mouse Model
Published on: April 18, 2015