Related Experiment Video
Updated: Nov 2, 2025
07:54
Implantation and Monitoring by PET/CT of an Orthotopic Model of Human Pleural Mesothelioma in Athymic Mice
Published on: December 21, 2019
7.1K
Deep learning with deep convolutional neural network using FDG-PET/CT for malignant pleural mesothelioma diagnosis
Kazuhiro Kitajima1, Hidetoshi Matsuo2, Atsushi Kono2
1Department of Radiology, Hyogo College of Medicine, Nishinomiya, Hyogo, Japan.
Oncotarget
|June 17, 2021
Summary
Artificial intelligence (AI) deep learning with 3D DCNN shows promise in differentiating malignant pleural mesothelioma (MPM) from benign conditions using FDG-PET/CT scans. Protocol D, combining AI, clinical data, and imaging, achieved the highest diagnostic accuracy.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Oncology
Background:
- Malignant pleural mesothelioma (MPM) diagnosis can be challenging.
- Differentiating MPM from benign pleural diseases requires accurate imaging analysis.
Purpose of the Study:
- To evaluate the diagnostic accuracy of an artificial intelligence (AI) deep learning method using a 3D deep convolutional neural network (3D DCNN).
- To differentiate malignant pleural mesothelioma (MPM) from benign pleural disease using FDG-PET/CT results.
Main Methods:
- Retrospective analysis of 875 patients with suspected MPM undergoing FDG-PET/CT.
- Four protocols were compared: AI with PET/CT alone (A), human visual reading (B), quantitative SUVmax (C), and a combination of AI, SUVmax, gender, and age (D).
- Receiver Operating Characteristic (ROC) curve analyses were performed.
Main Results:
- Protocol D demonstrated the highest diagnostic performance with an AUC of 0.896, sensitivity of 88.5%, specificity of 73.6%, and accuracy of 82.4%.
- Protocol D showed significantly better diagnostic performance compared to protocols A, B, and C (p < 0.05).
Conclusions:
- Deep learning with 3D DCNN, integrated with FDG-PET/CT and clinical features, offers a flexible and potentially powerful tool for the differential diagnosis of MPM.
- This AI-driven approach enhances diagnostic accuracy for distinguishing malignant from benign pleural conditions.
