Abstract
Parkinson's disease is a neurodegenerative disease that often comes with symptoms such as muscle stiffness, slowness of movement, and tremors at rest. Since this disease negatively influences the quality of life in patients, an early and accurate diagnosis is important for slowing the progression of the disease and providing effective treatment to patients. One of the quick and easy methods for diagnosing is the spiral drawing test and the differences between the target spiral picture and the drawing by patients can be used as an indicator of movement error. Simply, the average distance between paired samples of the target spiral and the drawing can be easily calculated and used as the level of movement error. However, finding the correct pair of samples between the target spiral and the drawing is relatively difficult, and the accurate algorithm to quantify the movement error has not been thoroughly studied. In this study, we propose algorithms applicable to the spiral drawing test, that ultimately can be used to measure the level of movement error in Parkinson's disease patients. They are equivalent inter-point distance (ED), shortest distance (SD), varying inter-point distance (VD), and equivalent angle (EA). To evaluate the performance and sensitivity of the methods, we collected data from simulation and experiments with healthy subjects and evaluated the four methods. As a result, in normal (good drawing) and severe symptom (poor drawing) conditions, the calculated errors were 3.67/5.48 from ED, 0.11/1.21 from SD, 0.38/1.46 from VD and 0.01/0.02 from EA, meaning that ED, SD, and VD measure movement error with high noise while EA is sensitive to even small symptom levels. Similarly, in the experiment data, only EA shows the linear increase of error distance to changing symptom levels from 1 to 3. In summary, we found that EA is the most effective algorithm in finding the correct pair of samples between the spiral and the drawing, and consequently yields low errors and high sensitivity to symptom levels.