Artificial doctor's assistant

Another neural network was taught to diagnose a problem by X-ray

Maxim Agadzhanov, Geektimes

Google, IBM and others have been working for a long time to create an AI (its weak form) that could analyze X-rays. What for? The problem is that radiology specialists, and not only them, have to spend a lot of time analyzing medical images. There are a lot of such pictures, and you need to view and give your answer for each one for a certain time. 

The specialist has very little time left to analyze the same X-ray image. And it's good if the doctor is fresh and cheerful when viewing the image. And what if he works already at the end of the working day, after viewing a couple hundred of the same images? The human factor is very strong here, and the probability of error increases many times. In order to make the task easier for a specialist, scientists are trying to use the capabilities of artificial intelligence. 

Another problem of doctors who regularly view medical images (not necessarily radiographs) is this is a "search satisfaction" error. It lies in the fact that the doctor who looks at the picture, having discovered the problem, may not search for others, deciding that his assumption is correct, and immediately diagnose. The consequences can be quite severe for the patient, given that the identified problem is not always a manifestation of the underlying disease. 

Now a team of scientists led by Andrew Eun is engaged in the development of a neural network that would search for manifestations of various diseases in medical images. Specialists created a neural network, which was trained on the example of a database consisting of several tens of thousands of images (almost 50 thousand) received from more than 14 thousand medical institutions. At the same time, each of the images was previously analyzed by doctors who diagnosed and marked the radiograph as normal or pathological. 

The results of the work of the neural network and three radiologists-doctors

The effectiveness of the neural network after training was compared with the work of three radiologists-doctors. As it turned out, in two cases the neural network almost did not lag behind the person, and in one it surpassed him. In general, the computer correctly identified the damage in 74.9% of cases. It is worth noting that scientists have revealed the results and materials of their research to the world. So, the database on which the neural network was trained is publicly available and is available on the Stanford website. It is ready so that it can be used to train other neural networks. 

Neural networks also work with other types of medical images. For example, a deep neural network learns to recognize traces of the disease in the images of positron emission tomography of the brain (PET). We are talking about Alzheimer's disease, which is characterized by the appearance of amyloid plaques with a slowdown in brain metabolism. 

Previously, scientists have found that some types of PET scans are able to detect signs of these negative conditions. Consequently, the technology can work to identify moderate cognitive impairments in humans, impairments that will later lead to the onset of Alzheimer's disease. 

However, it is quite difficult for human scientists to interpret the resulting images. But the neural network can quite cope with this thanks to one or two markers. To train the computer system, specialists used brain images of 182 people aged 70 with a healthy brain and 139 brain images of people of about the same age with a diagnosis of Alzheimer's disease. As a result, the AI was able to recognize the difference between a healthy and a sick brain, and did it with a high degree of accuracy – above 90%. 

As for Andrew Eun and his team, they are trying to use the capabilities of the neural network for another project. We are talking about patients with very serious diseases and palliative care. The neural network tries to predict how serious the patient's condition is (mainly, we are talking about very elderly people). If we are talking about a progressive disease that takes the patient no more than a year of life, then a team of palliative therapists enters the work, who try to remove the negative manifestations of the disease (pain, psychological state, etc.) to some extent. The problem is that the team has to start working at a certain time so that the effect is maximized. And here the neural network also shows significant success.

In general, AI (its weak form) is now considered by scientists as a doctor's assistant, and not an alternative, so to speak. Neural networks help a specialist to identify various kinds of problems, and already a human doctor makes an accurate diagnosis using the help of his digital assistants. As a result, time is saved and diagnostic accuracy is increased. Over time, neural networks will become reliable assistants to doctors – today this practice is experienced, but the results obtained inspire healthy optimism in the possibilities of computer technology in such an area as healthcare.

Portal "Eternal youth" http://vechnayamolodost.ru