Recently, Futong Dongfang published an academic paper titled “Research on CT Image Detection of Small Bone Fractures Based on Global Attention Multi-Task Network Method” in the journal China Medical Equipment, where the company served as the first author. This research was led by the Futong Genesis AI and received support from the national key research and development program “Basic Research Conditions and Major Scientific Instrument Equipment Development.” This represents a significant breakthrough for the Genesis AI in the field of smart healthcare, showcasing Futong Dongfang’s strong research capabilities and innovative strength in this area. 

 Breakthrough Research Achievement

The paper aims to enhance the perception ability of CT image detection for small bone fractures using a global attention multi-task network, enabling fast and accurate instance-level detection of small fracture targets to assist clinicians in timely treatment. The study introduces a grouped non-local network method to compute the long-range dependencies between any positions and channels in continuous slices of CT images, integrating the multi-object detection model 3D RetinaNet single-stage detector with the medical image semantic segmentation architecture (3D U-Net) to construct an end-to-end multi-task 3D convolutional network. By employing a multi-task joint approach, the research team achieved instance-level detection of small fractures.

In testing, the detection accuracy of this multi-task 3D convolutional network method significantly surpassed that of the single-task networks FracNet, 3D RetinaNet, and 3D Retina U-Net, with average precision scores higher by 7.8% and 11.4%, respectively. Additionally, the method outperformed the two single-task network detection methods, 3D Faster R-CNN and 3D Mask R-CNN, with average precision scores higher by approximately 6.7% and 3.1%. These results indicate that the performance of small fracture detection is significantly improved by integrating different modules within the global attention multi-task network, and the introduction of the grouped non-local network method further enhances detection accuracy.

This research received significant support from the national key research and development program “Basic Research Conditions and Major Scientific Instrument Equipment Development.” The objective of this project is to promote the intelligent analysis and application of health data in China while fostering innovation in basic research conditions and major scientific instruments. As a key participant in this project, Futong Dongfang is not only actively engaged in cutting-edge technology research but is also committed to transforming research outcomes into practical applications, providing efficient services and support to medical institutions and patients.

Integration of Research and Industry: AI Empowering Smart Healthcare

Futong Dongfang emphasizes the mutual promotion of research and industrial applications, committed to advancing the deep application of cutting-edge technologies such as artificial intelligence, big data, cloud computing, and the Internet of Things in various industries. In 2019, Futong Technology established the Genesis AI, an in-house AI research center aiming to accelerate the intelligent transformation of the industry with new quality productivity.

In the field of smart healthcare, Futong Dongfang, leveraging years of technical accumulation and profound industry understanding, has developed a series of core products and solutions, including smart outpatient services, chronic disease follow-up platforms, specialized disease databases, medical knowledge graphs, medical decision support, healthcare data asset management, smart medical insurance management, and intelligent quality control management.

Relying on advanced algorithms and a self-developed automated blood relationship label management platform, the Genesis AI demonstrates strong capabilities in aligning entities and concepts in the field of smart healthcare. Furthermore, through the collaborative development of data rule experience by the AI team and medical team, the center has constructed a million-level medical knowledge graph of triples, a domain-specific medical large model, a retrieval-enhanced generation system based on a multi-label system, and dynamic knowledge graph integration and updating technology based on causal reasoning. These technologies provide Futong Dongfang with more scientific knowledge management capabilities, forming synergistic advantages in model performance, decision support, resource optimization, data management, knowledge discovery, and complex applications in the field of smart healthcare, thereby better empowering smart healthcare.

Futong Dongfang has always regarded technological innovation as its core driving force, dedicated to empowering smart healthcare through advanced artificial intelligence technology. This research achievement not only provides a new solution for the detection of small fractures but also represents a high recognition of Futong Dongfang's research strength and innovative capability in the field of smart healthcare. In the future, Futong Dongfang will continue to uphold the spirit of innovation, focus on deepening industry expertise, and leverage cutting-edge technology to unlock data potential, bringing more research breakthroughs and innovative applications to the industry