Using a common brain scan, researchers used a machine-learning algorithm to diagnose Alzheimer’s in its early stage, about six months prior to a clinical diagnosis. This allows doctors to offer treatment.
Alzheimer’s has no known cure, but some drugs have been used in the recent years to help control the progression of this condition. However, to be effective these treatments should be given in Alzheimer’s early stages. This has motivated scientists to come up with ways of diagnosing the condition early.
- A recent study by Jae Ho Sohn, UC San Francisco’s resident in the Department of Radiology and Biomedical Imaging, combined machine-learning with neuroimaging to predict whether a patient would get Alzheimer’s disease when they faced a memory impairment for the first time.
- Sohn used a machine-learning algorithm with PET scans to diagnose Alzheimer’s disease in its early stage.
- Sohn fed the algorithm with images from the Alzheimer’s Disease Neuroimaging Initiative (ADNI)-a huge public database containing PET scans of patients diagnosed with no disorder, mild cognitive impairment or Alzheimer’s disease. Later on, the algorithm started learning on its own regarding the features that can predict Alzheimer’s disease and those that can’t.
- After training the algorithm on 1921 scans, the researchers tested it on two other datasets to assess its performance. First, they used 188 images obtained from the same ADNI database, but were not introduced to the algorithm yet. Second, they took scans from 40 patients who had a possible cognitive impairment.
- The results were excellent. The algorithm correctly identified 92% of patients who had Alzheimer’s disease in the first dataset and 98% in the second one.
- The algorithm made correct predictions about 75.6 months (6.3 years) before patients received their diagnosis.
Sohn now wants to calibrate and set the algorithm on bigger and more diverse datasets from various countries and hospitals.
If it can withstand the tests, neurologists can use it as diagnostic and predictive equipment for Alzheimer’s disease. This can provide treatments to the patients sooner.