Researchers have trained a deep learning algorithm to detect signs of Alzheimer's Disease in patients on a six-year average before diagnosing the condition by a human doctor.
California-based scientists indicated that after a training, a heavenly network can scan images of patients' brain and to find an average Alzheimer's presence 75.8 months before the actual diagnosis.
The 20-strong team based their research on a modern diagnostic method, F-FDG PET (or fluorodeoxyglucose 18 (18F) fluorine 18 (18F)) emission tomography, of which radioactive glucose flow is passed by & # 39; the brain, and to photograph.
Experts then explore and interpret these images using the naked eye for signs of Alzheimer's, predecessor known as mild cognitive impairment (MCI), or other related conditions across the spectrum.
Although it takes a lot of time, this method has resulted in faster and earlier diagnosis, and more effective treatments.
However, given this approach depending on the recognition of a pattern, researchers saw an opportunity to significantly improve their performance using a self-training AI algorithm, publishing their findings in Radiology.
"There is a wide recognition that deep learning can help in tackling the increasing complexity and volume of imaging data, as well as the expertise of a variety of trained imaging doctors," the team wrote.
"Just start to start investigating the application of machine learning technology to complex patterns of perceptions, such as those found in functional PET imaging of the brain.
"We assume that the deep learning algorithm could identify features or patterns that are not apparent on the standard clinical review of images and thereby improve individual diagnostic distribution of individuals."
They aimed to evaluate whether a deep learning algorithm could be trained to anticipate the final clinical diagnosis in patients receiving F-FDG PET, and how it succeeded in comparison to current clinical standards.
From a study of 2,109 photographs of 1,002 patients who were already diagnosed, their algorithm was found to be able to detect Alzheimer's in images taken more than six years before diagnosis.
The algorithm performed better in identifying patients who would go on to get Alzheimer's or clinicians, as well as patients who would go on to develop Alzheimer's than a MCI predecessor.
These findings are the latest in a series of studies and trials that show the potential for AI to transform healthcare and preventative diagnosis.
In September, Francis Crick Foundation revealed that AI had learned how to model and predict the rates of heart disease deaths in patients with a greater degree of accuracy than trained doctors, or models created by experts.
Meanwhile, the Google DeepMind AI project began an important milestone in the summer as its AI system was able to explore 3D images of the eye and diagnose visually impaired conditions, as well as offering advice on treatment, within seconds.
The algorithm, tested jointly with the Eye Hospital in Moorfields in London, was able to recommend the best treatment path for more than 50 eye diseases with 94% accurately.
Despite a handful of limiting factors, including a small sample size, California researchers concluded that he had developed a deep algorithm that can predict Alzheimer's "with accuracy and high strength".
That added, with access to the volume of many data and galibration opportunities for the model, the algorithm that they had developed could be integrated directly into the workload of the clinicians and acts as an essential tool of support .