The rise of artificial intelligence ( AI) has become one of the new driving forces for socio-economic development. It plays a crucial role in improving social productivity and achieving social development and economic transformation. As a core force leading the next generation of industrial transformation, AI has demonstrated new applications in the medical field and spawned new business models through deep integration.
In fact, compared to sectors like manufacturing, media, retail, and education, artificial intelligence (AI) is still in its early stages in healthcare, with relatively low commercialization and industry penetration. This is inevitably closely related to the nursing and conservative nature of the healthcare industry. However, it is undeniable that the integration of AI into healthcare has already addressed many of the difficulties of traditional medicine, with broad market demand, diverse business trends, and vast development potential.
The COVID-19 pandemic has propelled artificial intelligence (AI) to play a crucial role in the fight against the virus, enhancing overall efficiency. The pandemic has served as a litmus test for AI in the medical field, demonstrating its strength and value. In terms of application scenarios, AI in healthcare is still in its early stages, with image recognition, remote inquiry, and health management currently ranking in the first tier.
Image recognition, as a subfield of assisted diagnosis, is the most widely used application of artificial intelligence in the medical field.
The concept of imaging diagnosis and treatment originated in the field of oncology and has since been extended to the entire field of medical imaging. Understanding medical images and extracting key information valuable for diagnostic and treatment decisions is a crucial part of the diagnostic and treatment process.
Previously, medical image preprocessing and diagnosis required the involvement of 4-5 doctors. However, with AI-based image diagnosis, which trains computers to analyze medical images, only one doctor is needed for quality control and verification, significantly improving the efficiency of medical procedures.
Artificial intelligence first exploded and landed in medical imaging, mainly because image data is relatively easy to access and process. Compared to medical records and other data accumulated over three to five years, imaging data only requires a single shot and can be obtained in a few seconds. A single imaging film can reflect most of a patient's condition, becoming a direct basis for doctors to determine treatment plans.
The vast and relatively standardized database of medical images, along with the continuous advancements in intelligent image recognition algorithms, provides a solid foundation for the application of artificial intelligence in medicine.
From a technical perspective, medical image diagnosis primarily relies on image recognition and deep learning. Following the clinical diagnostic pathway, image recognition technology is first applied to the perception stage to analyze and process unstructured image data and extract useful information.
Secondly, deep learning technology is used to input a large amount of clinical imaging data and diagnostic experience into the artificial intelligence model, enabling the neural network to undergo deep learning training. Finally, based on the continuously validated and refined algorithm model, intelligent reasoning for image diagnosis is performed, outputting personalized diagnostic judgment results.
The combination of artificial intelligence based on image recognition and deep learning with medical imaging can address at least three needs. First, focus identification and annotation, which involves using AI in medical imaging products for image segmentation, feature extraction, quantitative analysis, and comparative analysis. Meeting this need, automatic recognition and labeling systems for medical images such as X- rays, CT scans , and MRIs , can significantly improve the diagnostic efficiency of radiologists. Currently, AI medical imaging systems can process over 100,000 images in seconds , improving diagnostic accuracy, particularly reducing the probability of false negatives.
Second, automatic target delineation and adaptive radiotherapy. Automatic target mapping and adaptive radiotherapy products can help radiation oncologists automatically delineate 200 to 450 CT scans, significantly reducing the time required to 30 minutes. Furthermore, during the patient's 15-20 irradiation sessions, the product continuously identifies lesion locations to achieve adaptive radiotherapy, effectively reducing radiation damage to the patient's healthy tissues.
Third, 3D image reconstruction. Registration algorithms based on grayscale statistics and feature points can solve the problem of fault image registration, save registration time, and play a role in lesion location, lesion range, identification of benign and malignant lesions, and surgical plan design.
In terms of application, China's current AL medical imaging products are mainly focused on major areas such as the chest, head, pelvis, and limb joints, and are primarily concentrated in leading cities for cancer and chronic disease screening.
In the early stages of the development and application of artificial intelligence (AI) in medical imaging, lung nodules and fundus screening were popular areas. With the maturation and iteration of the technology over the past two years, major AI medical imaging companies are expanding their business scope, with breast cancer, stroke, and bone age testing around the joints becoming key areas of focus for market players. AI medical imaging is also involved in the quantitative analysis and evaluation of COVID-19 treatment efficacy, becoming a key force in improving diagnostic efficiency and quality.
Dual entry of policy and capital
If the relative accessibility and processability of image data are the main reasons for the initial explosion and implementation of artificial intelligence in medical imaging, then the support of national policies and the large-scale access of capital have given artificial intelligence the power to continuously update its application in medical imaging.
In terms of new policies, from 2013 to 2017 , various government departments introduced a number of policies to continuously increase support for domestic medical imaging equipment, third-party independent medical imaging diagnostic centers, telemedicine and other fields.
At the end of 2016 , the State Council released the "13th Five-Year Plan" for the development of national strategic emerging industries, which repeatedly mentioned medical imaging, pointing out the need to "develop high-quality medical imaging equipment" and "support enterprises, medical institutions, research institutions, and others to jointly build third-party imaging centers." In January 2017 , the National Development and Reform Commission included medical imaging equipment and services in its plan.
