Abstract:
Dementia is a common disease of cognitive impairment that in addition to medical procedures can also be detected by neuropsychological tests. Mini-Cog test, a combination of the Clock Drawing test and Recall test, is one of the most effective procedures for early screening of dementia patients. We have automated this test through Android application using computational algorithms of image analysis and pattern recognition. The online clock images drawn by the user are binarized then the digits and hands extracted from it. Digits are recognized using nearest neighbor classification (with 1000 digits in the training set) and the position and order of digits with respect to the intended position is computed. The identified arms of the clock are checked against the correctness of the time setting 11:10. Based on the position and order of digits and the time setting produced by a subject, a decision is made on the correctness (or otherwise) of the clock. The clock draw test is combined with the recall test where the subject is required to recall the three words shown prior to drawing the clock. A final decision about the mental health of the subject is then taken based on the results of the two tests. The developed application is likely to be helpful for the practitioners in an efficient screening of dementia patients. Non-medical personnel can be trained to use the system and provide details of scoring to the psychologists for further assessment enabling them to keep their focus on the patients and treatment instead of scoring and report formulation. Supervisor: Dr. Imran Ahmed Siddiqi.
