Multi-task deep learning assists detection and diagnosis of gliomas and brain metastases.
Authors
Affiliations (12)
Affiliations (12)
- Department of Radiology, Huashan Hospital, Qidong-Fudan Innovative Institute of Medical Sciences, Fudan University, Jingan District, China.
- Department of Radiology, the Second Affiliated Hospital of Xuzhou Medical University, Xuzhou City, China.
- College of biomedical engineering, Fudan University, Yangpu District, China.
- Shanghai Engineering Research Center of Intelligent Imaging for Critical Brain Diseases, Yangpu District, China.
- Zhejiang Cancer Hospital, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences, Hangzhou, China.
- Department of Dermatology, Huashan Hospital, Fudan University, Jingan District, China. [email protected].
- Department of Neurosurgery, Huashan Hospital, Fudan University, Jingan District, China. [email protected].
- Department of Radiology, Huashan Hospital, Qidong-Fudan Innovative Institute of Medical Sciences, Fudan University, Jingan District, China. [email protected].
- Department of Radiology, Huashan Hospital, Qidong-Fudan Innovative Institute of Medical Sciences, Fudan University, Jingan District, China. [email protected].
- College of biomedical engineering, Fudan University, Yangpu District, China. [email protected].
- Shanghai Engineering Research Center of Intelligent Imaging for Critical Brain Diseases, Yangpu District, China. [email protected].
- Institute of Functional and Molecular Medical Imaging, Jingan District, China. [email protected].
Abstract
Gliomas and brain metastases (BMs) on MRI pose significant diagnostic challenges for radiologists. This study aims to develop a multi-task model and a computer-aided diagnosis (CAD) system for the detection and diagnosis of gliomas and BMs. This study enrolled 3909 participants from seven centers, and developed a brain tumor segmentation and classification network (BTSC-Net) and BTSC-CAD with visualization of tumor masks. For detection, BTSC-Net achieved a Dice coefficient of 0.888 and 0.872 on the internal and external test sets, respectively. For diagnosis, BTSC-Net achieved AUCs of 0.941 and 0.933 on the internal and external test sets, respectively. With BTSC-CAD assistance, junior radiologists achieved mean AUC improvements of 4.8% (P < 0.05) for detection and 17.3% (P < 0.001) for diagnosis, along with an average reduction of 64.75 s in reading time. BTSC-CAD significantly improved radiologists' diagnostic accuracy and efficiency.