2024-05-27
Share：
Recently, a research paper co-authored by a team from the Agency for Science, Technology and Research (ASTAR) in Singapore and DHC titled "A comprehensive AI model development framework for consistent Gleason grading" was published in the Nature subjournal Communications Medicine. This paper primarily explores the application of AI technology in enhancing the accuracy, consistency, and efficiency of prostate cancer Gleason Grading (GG), a crucial factor for assessing prostate cancer aggressiveness and guiding treatment decisions. Traditional manual microscopic scoring is time-consuming and may be subjective. Digital pathology AI technology can improve the objectivity of scoring, but its performance and consistency are affected by differences between scanning devices, making it difficult to widely apply digital technologies.
The research team developed an AI-based pathology calibration system, quality control system, and auxiliary diagnostic system capable of automatically checking the quality of histological scanned images, standardizing images produced by different devices, and continuously optimizing the performance of auxiliary diagnostic models based on feedback from pathologists. This breakthrough not only addresses the scalability and generalizability issues of AI in pathology but also paves the way for the widespread adoption of AI-assisted pathology diagnostic models in healthcare systems.
In 2023, DHC and Professor Yu Weimiao from ASTAR engaged in extensive academic collaboration in the field of pathology. The collaboration focused on integrating advanced pathology technologies with DHC's multimodal big data platform to develop several innovative tools, including a pathology calibration system, quality control system, and decision support system. These tools significantly enhanced the processing capability of multimodal big data and the precision of pathology-assisted diagnosis.
By the end of the year, the joint team participated in an international competition on ovarian cancer pathology analysis hosted by Kaggle, utilizing the pathology image calibration system, quality control system, and a weakly supervised learning algorithm based on the Transformer architecture to successfully achieve automatic recognition of ovarian cancer subtypes in whole slide images (WSI) and tissue microarrays (TMA). This technological breakthrough not only achieved excellent results in the competition but has also been expanded to recognize pathology images of more than 30 different cancers, providing strong support for the construction of DHC's large pathology base model and further advancing the development of pathology research and clinical diagnosis.
As one of the earliest established companies in medical big data and artificial intelligence in China, DHC has been focusing on the deep integration and innovative application of medical big data and artificial intelligence technology since its establishment. The company has over 300 million patient encounters in data governance experience and has successfully built a big data integration-governance-analysis closed loop centered on multi-omics and multimodality. Additionally, DHC has participated in the construction of several major national disease-specific database platforms, supporting the formulation of multiple disease-specific standard datasets, laying a solid foundation for building a "data bank."
With the advent of the large model era, DHC, relying on high-quality multimodal big data, has independently developed large models in the medical field supported by four major modality base models: text, imaging, pathology, and precision. These large models, through deep governance of clinical, imaging, pathology, and genetic data, empower various scenarios such as medical big data, clinical diagnosis and treatment, intelligent scientific research, and health management, gradually promoting the clinical application transformation of multimodal technology and achieving digital precision diagnosis and treatment.
In the future, DHC will continue to forge ahead on the path of smart healthcare, adhering to the philosophy of "the ultimate in technology, infinite health," providing not only high-quality products and solutions for the entire medical and health ecosystem but also driving medical research and academic progress with innovative technology, aiming to contribute to the realization of the "Healthy China" strategy.
TOP
