September 16th, at the 2025 Tencent Global Digital Ecosystem Conference, Deshi Biotech founder Song Ning announced a deep cooperation with Tencent to jointly advance medical AI infrastructure development and shared the company's breakthrough achievements in medical imaging artificial intelligence.

According to reports, through this collaboration, Tencent Cloud will provide high-performance GPU computing power, elastic resource scheduling support, and AI pre-annotation models, while Deshi Biotech will provide the iMedImage™ universal medical foundation model. This partnership aims to help medical institutions deploy AI with zero coding requirements, enhance research and clinical efficiency, and improve the intelligence level of medical imaging detection.

In recent years, the contradiction between medical imaging "information overload" and shortage of medical professionals has become increasingly prominent. China's annual medical imaging service demand reaches billions of cases, yet there are only 1.43 imaging doctors per 100,000 people. Under high-intensity workload, diagnostic differences among different doctors for the same imaging can reach as high as 20%.

Using AI to empower medical imaging detection has become one of the fastest-growing tracks in AI healthcare development. It is understood that China has a total of 3,285 medical imaging detection categories. Currently, there are only 35 approved Class III AI-assisted diagnostic medical devices and projects developed by Deshi Biotech combined, accounting for merely 2.8% of all medical imaging detection categories.

Deshi Biotech's independently developed iMedImage™ universal medical imaging foundation model features hundreds of billions of parameters and completed pre-training based on 80 million data points. It supports multi-modal image processing including CT, MRI, ultrasound, and pathology, with strong universal capabilities and high scalability. iMedImage™ requires only hundreds of imaging data samples and a training cycle as short as several days to complete specialized disease model construction. Its data dependency is far lower than traditional development models, reducing development costs by over 90%, and can quickly help China's 3,285 medical detection categories achieve intelligence.