A single CT scan can help doctors identify multiple types of cancer, and online platforms can complete personalized medical resource docking in seconds... In recent years, artificial intelligence (AI) technology has been comprehensively transforming all aspects of tumor diagnosis and treatment. AI can run through the entire process of tumor diagnosis and treatment Li Zhicheng, Executive Director of the Medical Artificial Intelligence Research Center of the Institute of Biomedicine and Health Engineering of the Chinese Academy of Sciences, introduced to the reporter of Science and Technology Daily that "from initial image diagnosis, focus recognition, patient admission, to pathological diagnosis, visualization of surgical scheme, and even recovery tracking after discharge, AI intervention is visible and tangible to doctors and patients." Xu Zhonghuang, the director of Beijing Meimei Airui Cancer Hospital, introduced that many cancer patients are diagnosed in the middle or late stages, missing the best treatment opportunity. Early screening can help doctors find the condition at the stage of asymptomatic or precancerous lesions, and effectively reduce the incidence rate and mortality through early intervention. AI has great potential in the field of early screening of tumors. Early screening of tumors usually relies on a series of non-invasive or minimally invasive examination methods, including imaging examinations, blood marker testing, and molecular diagnosis. In this regard, breakthrough progress has been made in AI intervention. Li Zhicheng believes that with the support of image-based deep learning technology, AI's performance in certain tumor screenings has even surpassed that of human experts. In the past two years, international journals such as Nature have published multiple studies on AI assisted tumor screening. The CHIEF model developed by the Harvard Medical School team can not only diagnose 19 types of cancer, but also locate the tumor microenvironment, guide treatment strategies, and predict survival rates. PANDA, an early detection model for pancreatic cancer developed by Alibaba Dharma Institute, has an accuracy rate of 92.9% in judging the presence of lesions. These achievements indicate that AI can not only assist in diagnosis, but also play a crucial role in precision treatment. Relevant practices have demonstrated the role of AI in tumor screening. In February of this year, Alibaba's "Medical AI Multi Cancer Early Screening Public Welfare Project" was deployed in institutions such as Lishui Central Hospital in Zhejiang Province, applying innovative medical AI technology from Damo Hospital to the field of health. "The project has screened more than 50000 people in four months, and the types of diseases screened include pancreatic cancer, esophageal cancer, gastric cancer, colorectal cancer, of which 145 cancer lesions have been clinically confirmed." Lv Le, head of the medical AI team of Damo Hospital, explained that by combining a large number of historical data and complex algorithms, AI can extract information about small lesions that are difficult to detect by the naked eye from images. In tedious image analysis tasks, AI can also quickly process large amounts of data, reducing the pressure on doctors. Xu Zhonghuang said that cancer requires multidisciplinary collaboration to develop the optimal treatment plan, and AI can help solve the problems of shortage of professionals and high economic costs in this process. Taking PANDA as an example, Lv Le said that the model is equivalent to gathering the knowledge base of dozens of different professional doctors, integrating multimodal data such as imaging data, genomics information, and pathology data to achieve cross departmental data fusion. On this basis, the model can extract key lesion information and potential pathological features, and then carry out comprehensive analysis across departments. Improving cancer awareness and promoting scientific understanding in the medical field is a higher dimension of AI assisted tumor diagnosis and treatment. Li Zhicheng's team has been engaged in research on glioma for decades. When it comes to the current status of diagnosis and treatment of glioblastoma, Li Zhicheng said, "Our scientific understanding of this disease is still limited, and doctors have not fully understood the mechanisms of occurrence, development, and recurrence of glioblastoma, nor have they found practical and effective precise treatment methods." Xu Zhonghuang deeply agrees with this. The insufficient understanding of cancer limits the diagnosis and treatment methods. Faced with difficult and complicated diseases, many times in clinical practice, we can only cross the river by feeling the stones. The existing AI diagnosis and treatment models also have limitations. Li Zhicheng said that many models are trained on large-scale annotated datasets to find correlations between image features and clinical outcomes. Although this method has achieved significant results in accuracy, this "black box" operation lacks explanatory basis, making it difficult for doctors to fully trust AI diagnostic results. Therefore, it is particularly important to return to the cognitive roots of medicine. There is a lot of room for AI to play in this regard. AI can integrate multimodal data such as imaging, pathology, and genetics, provide multi-scale comprehensive analysis, and help us build more complete tumor 'profiles'. Tumors are an ecosystem composed of complex cancer cells, and the more accurately their profiles are outlined, the more they can discover previously overlooked tumor behaviors and potential therapeutic targets, providing new ideas for front-end treatment. "Li Zhicheng said that with the continuous enrichment of molecular level data such as genomics and proteomics, AI is expected to break through existing cognitive bottlenecks and help improve scientific understanding of complex cancers. Xu Zhonghuang added, "In the face of unfamiliar tumors, if AI can promote human cognitive progress, even a small step, it may fundamentally provide new methodological guidance for cancer diagnosis and treatment, truly changing the way we deal with cancer." To further empower the entire process of cancer diagnosis and treatment with AI, it is crucial to obtain high-quality, comprehensive, and massive data support. The training of AI models not only relies on doctor annotations, but also requires complete clinical cycle data. Lv Le gave an example: "During the training process of the PANDA model, doctors not only need to provide multimodal data such as pathological images, pathological reports, and CT images, but also need to manually confirm the location of the lesion and accurately outline it on enhanced CT. Then, engineers use 3D image registration technology to map the 3D outline of the lesion onto plain CT, ultimately allowing AI to learn how to identify the early pancreatic tumor in plain CT images." In this process, only doctors and AI teams work closely together to provide high-quality training data for the model. Lv Le further explained that cutting-edge medical AI algorithm teams often rely on a wide range of collaborating hospitals to provide diverse data, which is crucial for improving the model's generalization ability. The data from different hospitals provide rich pathological background for AI models, helping them to more accurately respond to various clinical situations. However, due to the large amount of data required, involvement of multiple departments, and scattered data, data acquisition has become the main bottleneck in current cancer AI research. It is not difficult to obtain a single image or pathological data, but it is very difficult to simultaneously obtain full modal data such as images, pathology, genes, etc. of the same patient, "said Li Zhicheng. This not only requires close cooperation among multiple departments, but also requires a lot of time. Current cancer research is often scattered across different disciplines, with imaging analysis being handled by imaging and engineering technicians, while genetic data is processed by molecular pathology or bioinformatics personnel. Breaking down barriers between disciplines and integrating data remains a huge challenge. Data is the fundamental 'nutrient' for whether AI can fully play its role in healthcare In Xu Zhonghuang's view, the scalability, standardization, and security of data are key considerations for hospitals when deploying medical AI. Hospitals must start from the present when planning their AI layout, ensuring standardization in data entry, archiving, and management, designing a reasonable data management framework in advance, and reserving interfaces for future data processing. The advantage of AI lies in its ability to continuously absorb new data and perform self optimization. This requires hospitals' data storage systems to have scalability to meet the growing demand for multimodal data. In terms of data security, Xu Zhonghuang believes that hospitals need to establish strict data encryption and privacy protection mechanisms to ensure that technology applications provide reliable support for clinical diagnosis and treatment work in compliance with laws, regulations, and social ethics. (New Society)