AI-Assisted Diagnosis and Decision-Making Method in Developing Countries for Osteosarcoma

Area of Science:

• Medical Imaging
• Oncology
• Artificial Intelligence

Background:

• Osteosarcoma is a high-mortality bone cancer requiring early diagnosis.
• Magnetic Resonance Imaging (MRI) is crucial for osteosarcoma detection.
• Challenges in osteosarcoma MRI interpretation include complex structures, heterogeneity, and limited resources in developing regions.

Purpose of the Study:

• To develop an efficient and accurate MRI image segmentation system for osteosarcoma.
• To reduce diagnostic time and improve the accuracy of lesion area prediction.
• To create a system with lower hardware requirements for wider accessibility.

Main Methods:

• Proposed an Attention Condenser-based MRI image segmentation system for osteosarcoma (OMSAS).
• Developed an Attention Condenser-based residual structure network (ACRNet) inspired by AttendSeg.
• Validated the model on over 4000 MRI samples from two Chinese hospitals.

Main Results:

• The OMSAS system demonstrated higher efficiency and accuracy in osteosarcoma MRI segmentation.
• ACRNet achieved accurate segmentation with reduced structural complexity and lower hardware demands.
• The model outperformed existing methods in experimental tests.

Conclusions:

• The proposed OMSAS system, powered by ACRNet, offers a promising solution for efficient and accurate osteosarcoma MRI segmentation.
• This technology can assist physicians in rapid lesion localization and segmentation, particularly in resource-limited settings.
• The system's lighter structure and high performance suggest potential for widespread clinical adoption.