Deepwise Medical Expands Capabilities with Nine AI Class III Certificates, Redefining Boundaries in Intelligent Healthcare

In the Post-Imaging AI Era, How Strong Can the Capabilities of a Single Medical AI Company Be?

A straightforward metric for evaluation is the number of Class III medical device certificates held by a company. With the recent approval, at the end of last month, of its AI-powered software for assisted evaluation of head and neck CT angiography images, DeepWise now holds nine Class III AI certificates. Its solutions provide intelligent empowerment for common DR, CT, and mammography systems, offering assisted diagnosis across high-volume applications involving the chest, cardiovascular and cerebrovascular systems, women’s health, and pediatric growth and development.

However, intelligent capabilities developed for a single device or a single disease may seem straightforward but prove ineffective in front of clinicians. In most cases, they do not answer multiple-choice questions; instead, they identify every possible finding within "open-ended" medical images.

DeepWise clearly understands the real needs of hospitals for AI. Seemingly independent AI applications have long transcended their individual capability boundaries through platform integration, thereby creating new, end-to-end capabilities for assisted disease diagnosis and treatment.

Understanding the current value of DeepWise requires a two-step approach. First, examine whether individual AI products can truly optimize physicians’ workflows by addressing specific, discrete problems. Second, assess whether the platform has the capability to integrate existing intelligent algorithms, thereby generating emergent value.

The value of a single medical AI product lies in cost reduction and efficiency improvement within specific scenarios, characterized by high specificity and minimal or no disruption to physicians' diagnostic and treatment workflows. The recently approved software for assisted evaluation of head and neck CT angiography (CTA) images utilizes DeepWise’s state-of-the-art deep learning algorithms combined with extensive expert-annotated data. It automatically performs head and neck CTA image reconstruction, intelligently detects and measures vascular stenosis and plaques, and generates structured reports for assisted diagnosis.

Compared with traditional workstations, the advantages of head and neck CT angiography image-assisted evaluation software are reflected in features such as intelligent and rapid reconstruction, comprehensive coverage of clinical workflows, and algorithm-based automatic bone removal. On one hand, deep learning-enabled AI can identify blood vessels without the need for subtraction-based bone removal, completing post-processing images of head and neck vasculature within minutes, requiring physicians only to confirm the final results. On the other hand, DeepWise utilizes AI to comprehensively cover the entire process, including image acquisition, post-processing, lesion assessment, film printing, and structured reporting, thereby further improving imaging processing efficiency and reducing physicians’ workload.

The integrated solutions, formed by the convergence of multiple AI technologies on the platform, can meet specialized needs and even empower healthcare institutions with capabilities they previously lacked.

For instance, in the management of acute stroke at primary healthcare institutions in China, except for aneurysmal subarachnoid hemorrhage (SAH) where CT angiography (CTA) is partially available, most emergency stroke cases rely solely on non-contrast computed tomography (NCCT) without further cerebral CT perfusion (CTP) imaging, leading to occasional under-treatment and over-treatment.

However, by integrating the head and neck CT angiography system with DeepWise’s self-developed head and neck CT perfusion (CTP) product to form the Dr.Wise® Head and Neck Vascular Solution, acute stroke diagnosis and treatment can be effectively implemented at the primary care level.

In simple terms, after admitting stroke patients, head and neck CTA can standardize and homogenize the post-processing reconstruction of vascular images, reducing diagnostic errors; CTP, based on quantitative values such as CBV, CBF, TTP, MTT, and Tmax, as well as high-precision quantitative analysis of brain tissue windows, accurately measures the ischemic penumbra, core infarct area, and hypoperfusion area, automatically outputs mismatch ratios, and achieves quantitative analysis of abnormal perfusion regions.

Through the synergy of these two technologies, the therapeutic time window for stroke patients has been extended to 24 hours, thereby enabling precise treatment capabilities at primary care levels and addressing the two critical challenges of “speed” and “accuracy” in emergency stroke care.

In addition to head and neck CTA and head and neck CTP, DeepWise also offers auxiliary diagnostic products for pneumonia, pulmonary nodules, and bone age. Currently, all AI and informatization solutions are integrated into DeepWise’s intelligent imaging and big data general platform, Deepwise MetAI.

As China’s first integrated innovation platform for the intelligent development of hospital radiology departments, Deepwise MetAI aims to elevate radiology workflows from the current “information networked” stage to “AI-driven digital intelligence across all processes and scenarios,” comprehensively enhancing the precision and efficiency of radiology operations.

To this end, the platform is imaging-centric and integrates multimodal data to achieve an integrated solution for “storage, transmission, computation, application, and management.” It vigorously promotes the development of comprehensive smart radiology departments and plays a positive role in health assessment, early warning, assisted diagnosis, clinical decision support, multidisciplinary consultations, research management, teaching and training, and the intelligent transformation of radiology departments.

Taking DeepWise MetAI’s Intelligent Multimodal Research Platform as an example, this sub-platform leverages a proprietary dual AI engine for “imaging + text” to create a one-stop visual analysis and processing tool for medical data under the paradigm of “AI for Medical Science.” The platform offers comprehensive capabilities, including intelligent medical database management, robust statistical analysis modules, AI-powered automatic annotation, one-click end-to-end feature extraction, feature selection, and machine learning and deep learning modules. It provides researchers with a one-stop AI solution, facilitating intelligent and precise research across multiple disease areas, such as oncology, cardiovascular diseases, neurological disorders, respiratory conditions, digestive diseases, and musculoskeletal disorders.

Overall, to empower diverse talents in the field of medical research, DeepWise has developed tailored support solutions to help them achieve greater accomplishments at various stages of their research journeys. Whether you are an ambitious “rising star” or a distinguished “leader” who has already achieved excellence in the field, the Intelligent Multimodal Research Platform will serve as an indispensable ally on your scientific journey, guiding the exploration of knowledge boundaries and pioneering a new future for medical research.

Examining the strategic layouts of leading medical AI companies, each enterprise is attempting to connect individual disease-specific points to form linear solutions, which are then woven into a “network” of smart clinical care.

To gain a competitive edge in this process, two key factors must be addressed: first, complete the development of single-point applications as quickly and efficiently as possible; second, gain deep insights into real-world clinical needs to connect these “single points” into efficient solutions that align closely with clinical practice.

To balance both aspects, DeepWise pioneered the integration of large language model capabilities into its intelligent multimodal research platform, thereby accelerating AI application development while assisting physicians in addressing real-world clinical challenges.

Today, the intelligent multimodal research platform not only retains robust capabilities in multimodal project management, AI-powered automatic annotation, multimodal statistical analysis, and AI modeling, but also innovatively introduces SAMI (Segment Any Medical Images), a universal segmentation model for medical images. SAMI accommodates diverse multimodal medical data types, including X-ray, CT, MRI, PET, ultrasound, pathological slides, endoscopy, dermoscopy, dental radiographs, and blood smears. It achieves sub-second segmentation at various levels—from organs and lesions to pathological cells, diseased tissues, and cellular components—providing a one-click AI solution for all image segmentation tasks in clinical practice.

For medical researchers, this represents a brand-new research paradigm that has revolutionized efficiency. For DeepWise, the significance of its newly developed capabilities lies in reducing AI development costs and improving application development efficiency, thereby turning previously uneconomical AI application development from theory into reality.

Taking bigger strides, they are continuously closing the gap toward the endgame of all-scenario AI.