Identifying signs of Alzheimer’s disease in MR scans of patient brains is a manual process requiring experienced physicians trained for the task. To take some of the guessing out of it, scientists at VTT Technical Research Centre of Finland have developed image recognition software that automates the process, and reportedly does it with similar accuracy to that of trained doctors.
With VTT’s new method, the volume of the hippocampus can be accurately calculated automatically. Earlier automatic systems for calculating the volume of the hippocampus are not in general clinical use because of deficiencies in speed and reliability.
Using VTT’s new method, the assessment of MR images takes 3 minutes. With the fastest currently available automatic MR image assessment methods, the assessment takes 15 to 20 minutes. However, it is not uncommon for assessments to last for several hours.
The new method is part of a system which is currently being developed under the EU PredictAD project to help diagnose Alzheimer’s disease. The system will be completed in 2011. The aim of the project is to develop objective methods which are sufficiently accurate, reliable and fast for clinical use but do not require large investments in equipment.