A recent investigation by the Cloud division of Google concludes that applying artificial intelligence (AI) to medicine is not only not easy, but it seems very complicated. Many experts have pointed out for years that AI is going to revolutionize health care, but the studies led by Google scientist Jia Li, on the application of AI to radiology imaging, and which were exhibited at a Conference in last March in San Francisco, they generate many doubts about it.
Jia Li’s work sought to determine how machine learning could help identify a disease in a real clinical setting, where data is sparse and where physicians require justification for their diagnosis. Some artificial intelligence experts suggest that entire areas of medical work, such as the analysis of radiological images, could become fully automated thanks to AI.
Li and his colleagues used deep learning to identify abnormalities on chest X-rays. Because they only had a small data set to work with, they used another data set to start the learning process. They also made sure that their method highlighted the area of an image that was critical for diagnosing an abnormality. This is important because deep learning is mathematically very complex, which makes it extremely opaque.
The results of the Google scientist confirmed that artificial intelligence can automate only a small part of the work of radiologists. Li believes that “doctors will not be replaced by AI in the near future. We can help doctors make better judgments and streamline the process. “
But not only Google is working to demonstrate the applications of AI in the field of Health and Medicine, and specifically in the field of Radiology. The Healthcare division of the former General Electric, now GE, recently tested Intel’s processor architecture on a system for classifying radiological images using deep learning. The result was performance that nearly six times exceeds the systems in use today.
As the amount of medical imaging data increases, so does the workload of radiologists. They are being asked to become more involved in patient care, while handling more and more images. What if AI technology could help radiologists get their work done faster and more accurately?
Despite Jia Li’s studies, the truth is that deep learning and other forms of artificial intelligence have the potential to streamline medical imaging workflows, improve image acquisition techniques to reduce X-ray exposure. and increase the value of image data research. This is what Intel and GE Healthcare say, explaining how “we are seeing promising demonstrations of the application of artificial intelligence to medical images.” And this is only the beginning of a spiral that many technologists and scientists postulate for the Health environment with the use of AI, always for the benefit of the patient.
