Medical & Biological Engineering & Computing

, Volume 54, Issue 6, pp 953–965

Quantifying Parkinson’s disease finger-tapping severity by extracting and synthesizing finger motion properties

  • Yuko Sano
  • Akihiko Kandori
  • Keisuke Shima
  • Yuki Yamaguchi
  • Toshio Tsuji
  • Masafumi Noda
  • Fumiko Higashikawa
  • Masaru Yokoe
  • Saburo Sakoda
Original Article

DOI: 10.1007/s11517-016-1467-z

Cite this article as:
Sano, Y., Kandori, A., Shima, K. et al. Med Biol Eng Comput (2016) 54: 953. doi:10.1007/s11517-016-1467-z
  • 194 Downloads

Abstract

We propose a novel index of Parkinson’s disease (PD) finger-tapping severity, called “PDFTsi,” for quantifying the severity of symptoms related to the finger tapping of PD patients with high accuracy. To validate the efficacy of PDFTsi, the finger-tapping movements of normal controls and PD patients were measured by using magnetic sensors, and 21 characteristics were extracted from the finger-tapping waveforms. To distinguish motor deterioration due to PD from that due to aging, the aging effect on finger tapping was removed from these characteristics. Principal component analysis (PCA) was applied to the age-normalized characteristics, and principal components that represented the motion properties of finger tapping were calculated. Multiple linear regression (MLR) with stepwise variable selection was applied to the principal components, and PDFTsi was calculated. The calculated PDFTsi indicates that PDFTsi has a high estimation ability, namely a mean square error of 0.45. The estimation ability of PDFTsi is higher than that of the alternative method, MLR with stepwise regression selection without PCA, namely a mean square error of 1.30. This result suggests that PDFTsi can quantify PD finger-tapping severity accurately. Furthermore, the result of interpreting a model for calculating PDFTsi indicated that motion wideness and rhythm disorder are important for estimating PD finger-tapping severity.

Keywords

Parkinson’s disease Finger tapping Magnetic sensor Principal component analysis Multiple linear regression 

Copyright information

© International Federation for Medical and Biological Engineering 2016

Authors and Affiliations

  • Yuko Sano
    • 1
  • Akihiko Kandori
    • 1
  • Keisuke Shima
    • 2
  • Yuki Yamaguchi
    • 2
  • Toshio Tsuji
    • 2
  • Masafumi Noda
    • 3
  • Fumiko Higashikawa
    • 3
  • Masaru Yokoe
    • 4
  • Saburo Sakoda
    • 5
  1. 1.Research & Development Group, Center for Technology Innovation - HealthcareHitachi Ltd.KokubunjiJapan
  2. 2.Graduate School of EngineeringHiroshima UniversityHigashi-HiroshimaJapan
  3. 3.Graduate School of Biomedical SciencesHiroshima UniversityHiroshimaJapan
  4. 4.Graduate School of MedicineOsaka UniversitySuitaJapan
  5. 5.National Hospital OrganizationToneyama HospitalToyonakaJapan