Cross-Domain Collaboration of 'AI-Enhanced Medical Imaging+' Emerges as a New Strategic Direction for Healthcare AI Enterprises

In early 2017, medical imaging expert Tao Xiaodong joined iFLYTEK to head the Smart Healthcare Division.

This AI company, labeled as a “smart voice” player, has appointed a medical imaging expert to head its healthcare division. iFlytek’s intention is clear: to strengthen its footprint in the field of medical imaging.

Nowadays, Tao Xiaodong has been working for six months. VCBeat (WeChat ID: vcbeat) interviewed him.He stated that iFlytek’s imaging division would integrate medical imaging with clinical information to assist physicians in making diagnoses.。

Recall that not long ago, genetic AI expert Xiong Huiyuan joined Infervision.Responsible for the research and development of radiomics products(Radiomics tightly integrates imaging, pathological, and genetic information, leveraging engineering methodologies to provide a comprehensive suite of solutions for clinical decision-making and scientific research.)

At the iKang Guobin Strategic Upgrade Launch Event, the newly introduced product iKangCare+ also leveragesPrecision Medicine and Artificial Intelligence Jointly Deliver Precise Health Checkup Services to Users。

Watson Health Imaging integrates medical imaging with clinical data to assist physicians in making diagnoses...

These events and products are gradually reflecting new trends in medical artificial intelligence:Cross-disciplinary Collaboration and R&D in AI-Based Medical Imaging Combined with Clinical Information May Become a New Direction for the Development of Healthcare AI Companies。

There are numerous AI startups in China’s medical imaging sector, focusing on enabling machines to identify lesions and determine whether tumors are benign or malignant. However, most of these companies currently rely solely on imaging data for their assessments. Tao Xiaodong of iFlytek previously worked at General Electric (GE) in the United States, conducting research in medical imaging, and served as Chief Architect for Philips Healthcare’s Radiology Solutions. Perhaps due to his years of experience working at multinational corporations, his strategic approach differs from that of most AI companies.

Tao Xiaodong stated: “We are no longer relying solely on imaging; instead, we place greater emphasis on integrating imaging with other clinical information to assist clinicians. Our R&D efforts encompass traditional image segmentation, image registration, and various image-based measurements, as well as the fusion of imaging data with other structured and unstructured data to develop artificial intelligence algorithms.”

VCBeat has previously interviewed a series of doctors, who statedWhile machine-based lesion recognition indeed helps physicians reduce missed diagnoses and improve efficiency, determining the benign or malignant nature of tumors and identifying the specific type of cancer requires integration of clinical data, pathological findings, and even gene sequencing results. Radiological imaging alone can only identify the location and size of tumors.。

The approach of “medical imaging + clinical information” is not unique to iFlytek.

According to VCBeat,Watson Health’s imaging product, called Watson Imaging Clinical Assessment, is also a tool that leverages medical imaging and clinical information to assist physicians in making diagnostic judgments, albeit with some differences from iFlytek.。

Watson Imaging Clinical Assessment can create more continuous and reliable electronic health records to improve the quality of medical reports and the accuracy of the billing process. Continuous patient records encompass comprehensive patient data obtained from primary care physicians, specialists, and emergency department physicians. Physicians leverage this longitudinal patient data along with medical images to make final clinical judgments.

Currently,This product targets aortic stenosis,If a cardiologist provides a diagnosis of aortic stenosis, Watson Imaging Clinical Assessment will verify whether the diagnosis has been uploaded to the electronic medical record and whether the uploaded content includes the patient’s problem list and diagnostic records.

IBM also establishedWatson Health Medical Imaging Consortium，This collaborative consortium comprises 15 leading healthcare institutions, academic medical centers, mobile radiology providers, and imaging technology companies.. To date, the diseases studied by the organization have covered breast cancer, lung cancer, diabetes, eye diseases, brain disorders, stroke, and heart disease.

Recently, Infervision has recruited Xiong Huiyuan, an expert in the field of genomics and artificial intelligence, to lead the research and development of its Radiomics-related products, integrating imaging, pathology, and genomic data to support clinical decision-making.。

Radiomics is an emerging frontier discipline based on quantitative imaging, feature calculation, image analysis, and model construction. It leverages various imaging features to intuitively and quantitatively describe the morphology and pathology of tumors in clinical practice, thereby providing robust imaging support for clinical decision-making.

Current clinical applications mainly include the following four types:

1. Application of Efficacy Assessment and Prediction for Chemoradiotherapy;

2. Application of Tumor Grading. Tumor grading holds significant clinical importance, as it informs clinical diagnostic and therapeutic decision-making and the formulation of imaging-guided treatment plans;

3. Application in Differentiating Benign and Malignant Tumors. Radiomic features can distinguish between benign and malignant tumors in various clinical applications, thereby guiding clinical decision-making;

4. Applications of Tumor Genetics.Scientists have discovered a strong correlation between tumor pathology and tumor genetics. Research in tumor genetics can provide a biological foundation for tumor diagnosis. Radiomics serves as an effective integration of pathology and genomics, making it a valuable tool in the study of tumor genetics.。

Infervision Introduces AI-Powered Genetic ExpertsIt has also become evident that the diagnosis and treatment of diseases cannot be addressed by imaging alone; rather, they require the integration of multi-dimensional information, including genomics, medical imaging, and pathology. The addition of Xiong Huiyuan will accelerate Infervision’s expansion plans for developing deep learning models in genomics and other fields.

Historically, tumor screening has relied on CT imaging or pathological examination. While CT scans can localize lesions, they are unable to determine the benign or malignant nature of nodules, resulting in high rates of missed diagnoses and misdiagnoses. Although pathological assessment is considered the gold standard for definitive diagnosis, obtaining tissue samples is often excessively painful for patients. Therefore, integrating medical imaging with liquid biopsy for tumor evaluation is emerging as a significant trend.

iKang Guobin is not a technology R&D company; it is an end-user of technological products., introduced its precision medicine health screening services at a recent product upgrade launch event. Users can first determine whether they have cancer or assess their risk of developing cancer through liquid biopsy or genetic testing.

If liquid biopsy detects fragments of cancer cells in the blood, CT imaging can then be used to determine the tumor’s location. This imaging interpretation may be performed by experts or, in the future, by approved, high-performance artificial intelligence systems.

This approach achieves both “characterization” and “localization” of the tumor, enabling physicians to directly formulate a treatment plan based on this information.。

Large companies have the capability to independently develop new fields. Startups also have their own approaches:Establish a Medical AI Alliance. For instance, companies engaged in diabetic retinopathy screening via fundus imaging, intelligent voice entry, pathological image recognition, radiological image recognition, and research on clinical decision support systems can form an alliance once their products are mature. On the basis of equality and mutual benefit, members can collaborate to facilitate hospital implementation and share resources and data, thereby forming a united front with stronger competitiveness and influence.。

In summary,There are two new trends in the collaborative research and development of medical artificial intelligence: one is independent R&D, and the other is forming alliances.. Their starting point was not to build a comprehensive platform like Alibaba or Baidu, but rather to naturally form a business or R&D model based on actual clinical needs, hospital requirements, and corporate demands.

The concept of “AI+ Medical Imaging” is akin to “Internet+.” Although current collaborations focus on clinical information, genetic testing, and pathological data, it is poised to integrate with a broader range of medical AI domains in the near future.Moreover, future collaborations will not be limited to those centered on artificial intelligence in medical imaging but may also involve cross-disciplinary cooperation across various fields of healthcare AI.。

This approach can facilitate the implementation of medical AI products, bringing them closer to the real-world clinical scenarios of hospitals and physicians, thereby serving as effective assistants to doctors.

To learn more about medical imaging, you can refer to “Global Innovation Report on Medical Imaging（2015Edition）》for an in-depth interpretation. Scan the QR code below to become an official VCBeat member and gain access to the full version of《Global Innovation Report on Medical Imaging（2015Edition）》, with more industry insights and investment and financing news all included. Furthermore, over the coming year, you will have unrestricted access to the completeIndustry Trends Report, promptly grasp the latest globalInvestment and Financing Information, equipped with comprehensive medicalEnterprise Database, andMassive Resource Matchmaking。

Scan the code to become a VCBeat member,

Beta Trial Price: CNY 365/year.