An Artificial Intelligence Multiprocessing Scheme for the Diagnosis of Osteosarcoma MRI Images

• DOI: 10.1109/jbhi.2022.3184930
• Publication Date: 2022-06-21T19:36:39Z
• Authors: Jia Wu; Pei Xiao; Haojie Huang; Fangfang Gou; Zhixun Zhou; Zhehao Dai
• Keywords: Robustness

Osteosarcoma is the most common malignant osteosarcoma, and developing countries face great challenges in diagnosis due to lack of medical resources. Magnetic resonance imaging (MRI) has always been an important tool for detection but it a time-consuming labor-intensive task doctors manually identify MRI images. It highly subjective prone misdiagnosis. Existing computer-aided methods osteosarcoma images focus only on accuracy, ignoring computing resources countries. In addition, large amount redundant noisy data generated during should also be considered. To alleviate inefficiency faced by countries, this paper proposed artificial intelligence multiprocessing scheme pre-screening, noise reduction, segmentation For we propose Slide Block Filter remove useless Next, introduced fast non-local means algorithm using integral denoise We then segmented filtered denoised U-shaped network (ETUNet) embedded with transformer layer, which enhances functionality robustness traditional architecture. Finally, further optimized tumor boundaries conditional random fields. This conducted experiments more than 70,000 from three hospitals China. The experimental results show that our have good better performance segmentation.