Quantitative Assessment of Limb Stability from Handwriting Images
ZhiFei He, Lu-Chuan Zeng, Ruijiang Xie, Keyi Li, Hai Lin Su, Jie Lei, Li Lan, Linglin Xia, Deyu Lin · 2021
Limb stability refers to the ability of limbs to maintain posture during physical movements. Observing limb movements and quantizing limb stability is crucial for tremor detection. In this paper, we propose a quantitative metric to quantify tremor of limbs. We applied the Otsu threshold segmentation method to quantify features of handwriting images that were digitized and identified. Then we employed the eight-connected components to quantify the deviation degree, obtained two parameters that can quantitatively evaluate the patient's left and right hand's stability. Further we studied the correlation between the patient's living habits and limb stability to predict latent tremor. The tremor pre-diagnosis based on the C4.5 algorithm reached a classification accuracy of 97.10%. Experimental results demonstrated that the parameters obtained by our method reach higher accuracy and reliability when quantitatively evaluating the stability of limbs. This method is beneficial for tremor detection of the patient's handwriting. Further research includes efforts on creating metrics that can be assessed in absolute scales, using a broader set of samples, and integrate the metrics into clinical practice for tremor diagnosis.