The "AI Revolution" in Clinical and Radiology Departments: Opening a New Era of Smart Healthcare

Author: Zhang Hongxia China Rehabilitation Research Center Beijing Boai Hospital

Review Chen Zhenbo China Rehabilitation Research Center Beijing Boai Hospital Deputy Chief Physician

The "new generation of technological revolution" is the emerging fourth technological revolution marked by big data, cloud computing, the Internet of Things, and artificial intelligence (AI) technology, also known as the AI ​​revolution, or "Industry 4.0." The core characteristics of the new technological revolution are: digitalization, networking, and intelligence.

Figure 1 Copyright image, no permission to reprint

Medical research has entered an era of parallel integration of big data and precision, and our reliance on theories and technologies such as digital models, informatization, and materials science has increased significantly. Scientific progress and technological innovation have laid the foundation for a new era of data-driven medicine. Among the key tasks of the 14th Five-Year Plan, AI, life health, brain science, etc. are targeted as key directions for strengthening the country's strategic scientific and technological strength.

1. What is AI?

AI is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is changing our cognitive patterns and the paradigm of medical research at an unprecedented speed and breadth. AI's powerful computing power and advanced algorithms are driving biomedicine towards data, computation and quantification. AI will become one of the important tools for humans to expand the boundaries of scientific knowledge, helping people to jump out of the cognitive scope and more deeply understand and process the complex information of life.

Figure 2 Copyright image, no permission to reprint

Medical AI systems cover multiple clinical links such as screening, diagnosis, treatment, prognosis and management. They can use a variety of heterogeneous data such as laboratory test results, images, texts and audios to complete systematic and complex tasks.

2. What are the clinical applications of AI?

AI is rapidly becoming a transformative force in the medical industry in terms of clinical applications. From improving diagnostic accuracy to developing personalized treatment plans, AI is helping to improve patient treatment outcomes and the efficiency of medical services. Currently, AI’s clinical applications include the following aspects.

(1) Diagnostic support: AI systems assist in diagnosis by analyzing medical images and patients’ clinical data. For example, AI can quickly identify cancer cells in pathological images or predict the risk of heart disease by analyzing electrocardiograms.

(2) Personalized medicine: AI can recommend the most appropriate treatment plan based on the patient’s condition, including genetic information, lifestyle habits, and medical history. This approach is expected to improve the effectiveness of treatment while reducing unnecessary side effects.

(3) Drug R&D: AI plays an increasingly important role in the drug discovery and development process, identifying new drug candidate molecules by analyzing complex biological data, accelerating the R&D cycle of new drugs, and reducing costs.

(4) Clinical decision support system: Using AI technology, we can develop advanced decision support systems to help physicians quickly find key information in massive amounts of clinical data, provide treatment recommendations, and support the clinical decision-making process.

(5) Patient monitoring and remote care: AI technology can be used to continuously monitor the health status of patients, collect data through smart wearable devices, and monitor patients' vital signs and health indicators in real time, thereby achieving early warning and intervention. I believe that with the advancement of technology, you will be able to get expert diagnosis and treatment guidance at home in the future.

(6) Epidemiology and public health: The application of AI technology in epidemiological research can help predict disease transmission trends and evaluate the effectiveness of public health strategies, thereby better responding to public health crises.

(7) Optimization of medical services: AI can optimize the operation and management of hospitals. For example, it can improve the utilization efficiency of medical resources through intelligent scheduling systems, reduce patient waiting time, and improve the quality and efficiency of hospital services.

Figure 3 Copyright image, no permission to reprint

3. What are the applications of AI in the Department of Radiology?

The application of AI in imaging is a revolutionary field. It is gradually changing the face of medical imaging by enhancing the diagnostic accuracy and efficiency of imaging. The combination of artificial intelligence and images has greatly improved the accuracy and efficiency of disease diagnosis. The following will introduce some AI technologies currently in clinical application.

Tumors: prostate cancer, breast cancer, etc. For example, in the case of prostate cancer, AI can automatically identify lesions suspected of being cancer and give a probabilistic diagnosis of the possibility of cancer.

Neurological aspects: mainly include brain structure analysis, small vessel disease, stroke, cerebral perfusion, etc. For example, cerebral perfusion can automatically give a pseudo-color image of the perfusion, identify low perfusion areas, and give an estimated volume of the ischemic area.

Blood vessels: mainly pulmonary blood vessels, neck blood vessels, head blood vessels, coronary arteries, aorta, lower limb blood vessels, etc. For example, in the case of pulmonary embolism, AI will be used to identify the area of ​​pulmonary embolism and give a VR image, which is very intuitive. You can directly see the embolus of pulmonary embolism and find blocked or missing blood vessels. The coronary artery can not only display the VR image of the heart volume and coronary artery, but also straighten the blood vessels, observe the specific degree of stenosis of the blood vessels, and predict the nature of the embolus, providing a basis for clinical diagnosis and treatment.

Chest: lung nodules, lymph nodes, bones, etc. AI software for lung nodules is widely used in clinical practice. It can quickly and accurately identify nodules, and give the size and nature of nodules (such as ground glass or solid), evaluate the possibility of benign or malignant nodules, and can also compare before and after follow-up to compare the changes in the size of nodules. The application of AI for lung nodules has greatly reduced the workload of radiologists, because if there are multiple nodules in the lungs, radiologists are likely to miss the diagnosis, but AI will not, and all nodules can be identified with one click. This automated process not only improves the speed of analysis, but also improves the consistency and accuracy of image interpretation. Of course, AI also has shortcomings, that is, the judgment of the nature of nodules is sometimes inaccurate, and the identification of nodules is too sensitive. Some inflammatory flake densities are identified as nodules or some blood vessels are identified as nodules. This will be an area where the software needs to be improved in the future.

Heart: Cardiac MRI heart function. The heart function AI can automatically calculate the myocardial thickness and heart chamber volume and evaluate the heart function without medication.

Orthopedics: fractures. AI is also widely used in fractures, which is a great boon for emergency physicians. Before the imaging physicians issue a report, emergency physicians can use AI to identify whether there is a fracture. Especially for some hidden fractures, AI's observation effect is even better than that of junior residents.

Education and training: AI can also be used in medical education and professional training, helping medical students and young physicians improve their diagnostic skills by providing virtual imaging diagnostic training.

Figure 4 Copyright image, no permission to reprint

Of course, the application of AI in the Department of Radiology also has some shortcomings, such as inaccurate recognition, high sensitivity and relatively low specificity; when dealing with lesions with poor image quality, motion artifacts or other types of artifacts, AI's recognition accuracy will decrease and errors are prone to occur. These are the directions that AI needs to be optimized in the future. At the same time, the widespread application of AI in clinical practice is also likely to cause a series of problems and challenges, such as data privacy protection and the division of responsibilities for AI misjudgments. Therefore, future development needs to seek an appropriate balance between technological innovation and the ethical and legal framework.