Alzheimer''s disease is the primary cause of dementia worldwide. Although there is no cure, it is considered crucial to be able to develop effective therapies that act before their development is irreversible.
The study by the University of Catalunya (UOC) has proven to indistinguish people who have become resistant to disease due to certain artificial intelligence techniques. This technique, which uses special artificial intelligence techniques to compare magnetic resonance images, is more effective than the other methods currently in use.
Fine-tuning the diagnosis
Alzheimer''s disease affects more than 50 million people worldwide, and the ageing of the population means that there may be many more patients in the future decades. However, it may be preceded by a mild cognitive impairment, which is substantially less severe than those experienced by people with Alzheimer''s, and is also more severe than expected. "These patients may progress and worsen or remain in the same condition as time passes," said Mona Ashtari-Majlan, an associate professor in the Digital Information Technology division.
Identifying these cases correct might help to improve the quality of clinical trials used to develop therapies, which increasingly seek to address the initial phases of the disease. In order to do this, the researchers used a multi-stream convolutional neural network, which is a technique that is extremely beneficial for image recognition and classification.
"We first compared MRIs from Alzheimer''s patients and healthy people to identify key landmarks," said Ashtari-Majlan. After testing the system, they fine-tuned the proposed architecture with resonance images from people who had already been diagnosed with severe cognitive impairment with significantly reduced differences. In total, almost 700 images from publicly available datasets were used.
According to Ashtari-Majlan, the learning process "overcomes the complexity of learning caused by the subtle structural changes between the two forms of mild cognitive impairment, which are significantly reduced than those between a normal brain and a brain affected by the disease. Moreover, the proposed technique might address the small sample size issue, where the number of MRIs for mild cognitive impairment cases is lower than for Alzheimer''s."
The new technique allows the two types of mild cognitive impairment to be distinguished and classified with an accuracy rate of over 85%. "The evaluation criteria indicate that our proposed technique outperforms existing ones," she said, adding that even after they are combined with biomarkers such as age and cognitive tests. "We can share our experience with anyone who wants to reproduce the results and compare their methods with ours."