The process of detecting a tumor is already a challenge. The removal of a tumor is another slow and difficult procedure. It’s normal for doctors to study more than 100 images, each displaying a thin slice of the tumor and the surrounding organs. Then, the painstakingly slow process of marking the border begins.
Medical professionals will typically go through each image and place markers in order to map out the tumor. It’s critical that the marking process is very accurate. Otherwise, radiotherapy beams may start frying the tumor and the surrounding healthy tissues.
It’s easy for the marking process to take hours. Now, with the help of AI, the process can be done in mere minutes.
One of the popular AI systems that are helping the NHS is Microsoft’s InnerEye. For now, patients with prostate cancer are the prime beneficiary of the InnerEye technology.
The process still starts by creating a scan of the tumor and the surrounding organs. After producing the images, the process of marking can begin. It’s in this part of the process in which InnerEye takes over.
InnerEye then assimilates the data, creates a 3D representation and automatically marks the border. The entire process only takes a couple of minutes.
InnerEye learned how to do the marking process by training on thousands of images from previous patients and with the guidance of experts. The system has already saved the lives of countless patients with prostate cancer. InnerEye is already in the process of learning how to mark brain tumors accurately.
In the future, InnerEye could potentially revolutionize the treatment of tumors. Currently, because of the costs and time-consuming process of marking and mapping, it’s typically done once. Once InnerEye develops to the point in which it can map and mark tumors in seconds, it can pave a way to “adaptive radiotherapy.”
This means that instead of marking and mapping a tumor one time, InnerEye could continuously repeat the process while the radiotherapy treatment is in session. The simultaneous operation can significantly increase the accuracy in which the tumor is irradiated.
Other applications of AI in NHS are still in the works. However, the sky’s the limit. AI could help in the identification of skin cancer from images of moles. It may assist in detecting disorders in the eye through retinal scan images. It may even identify early heart diseases from echocardiograms. Some AI projects are very ambitious like assessing which among the stroke patients need more urgent care.
While AI technology in health care may sound like it’s coming straight from a futuristic sci-fi movie, it’s slowly becoming a reality. However, AI technology will not revolutionize the NHS in a single night.
Just like any medical innovation, AI systems must be carefully evaluated, tested and validated before they can be safely approved. Also, there’s a need for a method to “override” the AI’s diagnosis. A scenario in which the AI is telling the situation is low-risk while the lab tests indicate a high-risk situation is possible. Hence, a system should be in place to minimize confusion.