Deepwise Medical Launches Multi-Category AI Products, Signaling Global Commitment to Intelligent Healthcare Transformation

As a first-tier player in AI for medical imaging, DeepWise has been making rapid strides recently.

After being selected as one of Deloitte’s “China Technology Fast 50 Rising Stars” in 2018, Professor Yizhou Yu, Co-founder and Chief Scientist of DeepWise, was successfully elected as a 2018 ACM Distinguished Scientist and named an IEEE Fellow in 2019.

Meanwhile, leveraging the momentum from CCR and RSNA in November, DeepWise successively launched seven AI products in collaboration with partners both domestically and internationally, a pace of development that has been astonishing.

So, what factors have driven the rapid growth of this AI startup?

“After developing so many products, we have found that to maintain DeepWise’s core competitiveness, our team must continue to innovate. This innovation stems not only from the development of new products but also from the optimization and upgrading of existing ones,” said Li Yiming, CTO of DeepWise, in an interview.

For existing pulmonary nodule products, the DeepWise team will continue to deeply explore the potential of artificial intelligence, focusing not only on improving diagnostic accuracy and reducing false-positive rates but also collaborating with physicians to design user interfaces tailored to the daily workflows of radiologists. The aim is to enhance both work efficiency and operational proficiency, thereby providing physicians with an optimal user experience.

Meanwhile, cross-departmental needs will remain a constant focus for AI development; therefore, exploring the correlation between imaging and pathology is also a key R&D objective for DeepWise. Taking its pulmonary nodule product as an example, DeepWise has already achieved a high level of accuracy in differentiating between benign and malignant nodules, but it will continue to further explore the link between imaging findings and pathological results.

Next, at the business expansion level, product delivery relies on comprehensive strategic layout and support from auxiliary platforms such as intelligent cloud services. Therefore, DeepWise actively develops AI products, builds cloud platforms, and delves deeply into radiomics, striving for all-around coordinated development. Li Yiming believes that “companies in this field each have their own characteristics. Whether they are giants crossing over from non-medical sectors, traditional medical equipment manufacturers extending their reach, or startups building from scratch, sustained engagement in the medical imaging AI sector continuously tests a company’s technological, product, and commercial capabilities. Shortcomings in any of these areas can severely constrain corporate growth. This is the core rationale behind DeepWise’s multi-pronged strategy.”

At the CCR Conference in November, DeepWise officially launched four categories comprising six products under its Dr.Wise™ brand: the Dr.Wise™ AI System for Early Cancer Screening, which includes the latest-generation AI-assisted screening and diagnosis system for pulmonary nodules and an AI-assisted mammography screening system; the Dr.Wise™ AI Detection and Analysis System for Stroke, covering both hemorrhagic and ischemic stroke; the DeepWise Intelligent Imaging Cloud (Dr.Wise™ Cloud); and the Dr.Wise™ Multimodal Research Platform.

DeepWise has consistently adhered to a scenario-driven approach in its product research and development strategy, striving to maximize the liberation of physicians with limited resources and enable more patients to benefit from technological advancements. The DeepWise Dr.Wise™ AI System for Early Cancer Screening comprises two products: pulmonary nodule screening and mammography screening, which target lung cancer and breast cancer, respectively—the two most prevalent cancers globally. Lung cancer, in particular, ranks as the leading cause of both incidence and mortality among cancers worldwide and in China. Early diagnosis and treatment can significantly reduce mortality rates, holding immeasurable significance for patients, their families, and society at large.

In addition to the two most common types of cancer, DeepWise also applies artificial intelligence to complex brain conditions. The Dr.Wise™ AI Stroke Detection and Analysis System provides specialized detection for both hemorrhagic and ischemic strokes.

To enable the aforementioned applications to integrate effectively with PACS systems, streamline data transmission and analysis, and achieve digitalization of diagnosis and treatment, DeepWise has launched Dr.Wise™ Cloud. This platform provides a rapid data integration hub for AI products. By leveraging advanced artificial intelligence technologies and the connectivity advantages of the internet, it allows physicians to efficiently access medical imaging data and AI processing results from various locations, offering precise diagnostic recommendations. This enhances the flexibility and efficiency of image interpretation, expands overall diagnostic resources, and facilitates the distribution of high-quality medical resources to primary care institutions.

Dr.Wise™ Multimodal Research Platform, based on data analysis of imaging, pathology, and genomics, extracts a large number of high-dimensional qualitative features to provide valuable guidance for disease diagnosis, treatment planning, and prognosis assessment. Leveraging advanced computer deep learning technologies, the platform simplifies complex research workflows, enhances research efficiency, and assists physicians in generating high-value research outcomes and publications.

DeepWise’s North American trip continued its expansion in the field of AI medical imaging. On November 27, at the RSNA Annual Meeting held in Chicago, DeepWise unveiled two new products. Focusing on stroke and bone age assessment respectively, these solutions further enhance cardiovascular and cerebrovascular detection capabilities while expanding into new application scenarios for pediatric care.

The second-generation Dr. Wise™ AI-assisted medical diagnostic system for stroke, based on ASPECTS scoring of MRI diffusion sequences, can automatically detect infarct lesions in ischemic cerebrovascular disease and accurately measure their anatomical locations and lesion volumes, thereby guiding subsequent clinical treatment. The system demonstrates excellent lesion detection rates, and its scoring results show a very high correlation with physicians’ assessments.

In Version 2.0, the automated detection system for hemorrhagic cerebrovascular lesions based on non-contrast CT scans has improved its detection rate over the previous version. It now includes detection and classification of subdural hemorrhage, epidural hemorrhage, subarachnoid hemorrhage, and intraventricular hematomas, enabling precise segmentation and attribute recognition for each type of hemorrhage. This assists radiologists and emergency physicians in promptly identifying lesions and provides volumetric measurements for different lesion categories. For complex intracranial hemorrhages, the system can detect and analyze intricate lesions.

DeepWise has launched the Dr. Wise™ AI-Assisted Bone Age Assessment System for pediatric applications. This system is built on optimized deep learning algorithms and a dataset comprising tens of thousands of imaging cases from children of diverse ages, genders, and ethnicities. While conventional bone age assessment by physicians typically takes around 30 minutes, the Dr. Wise™ system completes the evaluation in just a few seconds. Furthermore, it leverages artificial intelligence to automatically detect and grade epiphyses, perform automated calculations, and generate comprehensive growth and development reports, thereby enhancing both diagnostic efficiency and accuracy.

The healthcare industry is one with profound depth and a critical need for accumulated expertise, while artificial intelligence stands as one of the most rapidly evolving technologies today. Therefore, integrating these two vastly different domains and ensuring their efficient operation requires exceptional cross-disciplinary talent to master.

DeepWise’s team of intelligent professional model algorithms comprises PhDs and senior researchers from Peking University, the Chinese Academy of Sciences, and Stanford University. Seventy percent of its members have published papers at top-tier conferences such as ICCV, AAAI, and CVPR. The team includes many seasoned imaging experts and management specialists with years of experience at renowned multinational corporations in the healthcare industry. DeepWise’s CEO, Qiao Xin, previously worked in Siemens Healthineers’ Medical Imaging Division and brings nearly 15 years of experience in the healthcare sector, offering profound and clear insights into the medical imaging industry.

Recently, Dr. Yizhou Yu, Chief Scientist at DeepWise, was successfully elected as an ACM Distinguished Scientist (2018) and named an IEEE Fellow (Class of 2019). As a leading authority in artificial intelligence, deep learning, computer vision, and graphics processing, Dr. Yu has authored more than 130 academic papers to date, with nearly 70 presented at top-tier industry conferences. Despite these accolades, Dr. Yu remains deeply engaged at the research institute, dedicating himself to thoughtful product development.

As the key figure behind DeepWise’s research, Yizhou Yu has consolidated the company’s core technical expertise to explore cutting-edge technologies. This effort aims to provide leading technological solutions for medical AI, deliver high-quality solutions addressing critical clinical pain points for healthcare institutions, and cultivate outstanding talent in medical AI technology. Currently, DeepWise has established long-term academic and research collaborations with more than 10 top-tier domestic and international academic institutions, as well as long-term clinical research partnerships with over 20 leading Grade A tertiary hospitals. Multiple clinical products, represented by early screening solutions for lung cancer and breast cancer, have been successfully deployed in nearly 200 hospitals. Throughout this implementation process, DeepWise continuously incorporates clinical feedback to refine its products, ensuring superior practicality, user experience, and other aspects compared to similar offerings. Rigorous scientific inquiry and relentless product refinement have become the hallmark characteristics of DeepWise’s products.

“Medical AI shares commonalities with AI in other fields; for instance, many recent advances in computer vision can be extended to medical imaging, yielding favorable results. However, medical AI also faces unique challenges, such as limited training data, highly imbalanced data distributions, poor consistency in data annotation, and diverse data types (including multimodal imaging and combined text-and-image data). These challenges dictate that we cannot simply apply existing general-purpose AI technologies to the healthcare sector. Instead, it is essential to ‘tailor-make’ new AI technologies and algorithms specifically for medical applications. Only in this way can we truly address certain pain points in the healthcare field,” said Yu Yizhou.

Regarding the future of DeepWise, Yu Yizhou stated, “For instance, we need to develop AI models that can be trained on small datasets yet exhibit strong generalization capabilities, insensitivity to data distribution shifts, and high tolerance for inconsistencies in annotation. We also need to develop AI technologies capable of effectively fusing multimodal and multi-type data. Therefore, the journey for medical AI is long and arduous. Our researchers must possess both the determination to overcome challenges and the patience for long-term technological accumulation. We believe that a bright future for medical AI will ultimately arrive.”