New algorithm for accurate diagnosis of Alzheimer's disease

Text |

In recent years, with the development of artificial intelligence technology, "AI + medical treatment" has continuously achieved new breakthroughs. Researchers have developed a new machine-learning system that can accurately diagnose Alzheimer's disease from a single MRI scan and distinguish between early and late stages of the disease.

Alzheimer's disease (AD) is a neurodegenerative disorder in which patients experience memory impairment, aphasia, apraxia, agnosia, executive dysfunction, and changes in personality and behavior. Those with onset before the age of 65 are called Alzheimer's disease; those with onset after the age of 65 are called senile dementia. This disease is like an eraser for memory, it will erase the good memories between patients, family members and friends little by little.

As society continues to develop and progress, human life expectancy increases, and society ages, the burden of increased morbidity for Alzheimer's disease may exceed the current ability to diagnose and manage the disease. Currently, there are about 10 million Alzheimer's patients in China, and it is expected that by 2050, this number will exceed 40 million .

Alzheimer's disease is generally divided into three stages: asymptomatic (preclinical AD), pre-dementia (mild cognitive impairment caused by AD), and dementia (dementia caused by AD). While some drugs developed in recent years can help patients prevent disease progression, they must be done early to have a full effect.

Therefore, early and accurate diagnosis of neurodegenerative diseases such as Alzheimer's disease will play a very important role in effectively preventing the onset of Alzheimer's disease patients and improving their lives. Here, the researchers turned to machine learning techniques.

For the study, the scientists adapted an algorithm for classifying cancer tumors and applied it to the brain. They divided the brain into 115 regions and assigned 660 features of different sizes, shapes and textures to assess each region, and finally trained the algorithm to identify changes in these features to predict Alzheimer's disease exist.

Using data from the Alzheimer's Disease Neuroimaging Program, the team tested their method on brain scans of more than 400 patients with early and advanced Alzheimer's disease, healthy controls and patients with other neurological disorders. It was found that the algorithm was able to detect Alzheimer's patients from healthy subjects with 98 percent accuracy.

Notably, the algorithm can also distinguish early Alzheimer's brains from late scans with 79 percent accuracy. In the future, the algorithm may provide new avenues for the prevention and treatment of Alzheimer's disease.