Please enable JavaScript.

Coggle requires JavaScript to display documents.

Analysis (Functional Data Analysis (Registration (Types of Registration…

- - - - divide by body mass
        
        Moisio et al. 2003: once weight was accounted for, height no longer had significant effect on variance in the data set
        
        gravitational forces will be different - Newton's Laws
    - - gender differences would suggest normalizing by height to account for variance
      - This could be important for moments as a product of force and length - but Moisio et al. 2003 looked at this
      - Difference in adult height are <10%: not important to normalise
    - - leg length
        
        Hara et al. 2016: VICON leg length is best predictor for HJC
        
        VICON uses mean leg length and InterASIS distance in HJC calculation
    - - ensures variation is due to walking mechanics
- - - - no time phase shifting of the signal
    - - higher frequencies are attenuated
    - - amplitudes of the frequencies are relatively unaffected by filter
    - - number of times signal passed through filter
  - - - : folder titled processing and filtering (methodologies)
    - - Filter at same as kinematic
        
        Pro: noise is completely gone
        
        Con: may filter out real forces
      - Filter different than kinematics WILL USE
        
        Pro: keep physiologically meaningful data (possibly)
        
        Con: time phase shift
        
        could be dealt with by correct filter
    - - Harmonic Analysis
        
        power of each harmonic
        
        Con: researcher must decide how much power to accept
      - Residual Analysis WILL USE
        
        most common technique
        
        difference between filtered and unfiltered signals
        
        Con: only displacement
        
        Yu et al. (1999) - contains velocities & accelerations
        
        will do the extra steps influence outcome significantly?
    - - Many articles base cut-off on previous research
- - - - Shift Registration
        
        Pro: simply shift data without dealing with stretching or compressing time
        
        Con: only works for one peak data
      - Landmark registration
        
        Pros
        
        Creates a time-warping function
        
        works with multiple peaks in data
        
        Con: Does this extra work actually lower variability to a significant extent?
  - - - Aligns peaks in data
      - all data the same length
        
        con: lose some of time aspect
  - - - As we focus on difference of amplitude in biomechanics - wont be fixing
    - - fixed through time normalization and landmark registration
- - - - symmetrical axis of rotation approach
      - includes rotation and transformation in calculation for two segments independently
      - no preliminary transformation is required to different coordinate system
      - similar ROM to SCoRE accuracy
    - - larger ROM (>20) -> more accurate
      - works with two moving segments (calculated as such)
      - Marker artefacts are cummulative - still accurate predicition of HJC
      - Fast computation and easy to implement
      - symmetrical centre of rotation estimation
  - - - miscalculation of hip joint centre
        
        fixed by SCoRE
    - - uses 3 points
        
        calculated joint centre
        
        real marker
        
        perpendicular distance (joint centre offset)