Transformation and average as handy tools for pressure measurement data analysis.

Abstract

INTRODUCTION:: Pressure measurement data underneath[Table: see text] the foot represent usually a large number of data files which is equal to double product of the number of the subjects in study and of the number of the trials for each foot. Besides, 'normal' parameters are necessary to compare the results for the controls and for the subjects in study. Statistical parameters (mean values and standard deviations) are not very convenient when it is required to establish cartographical and geometrical parameters or peak pressures, maximum forces, contact areas, contact time and others for foot typing or as a standard. Sometimes it is reasonable to obtain average pressure data as a base for further data analysis. But taking into account that pressure distribution pattern may differ by its length, width and angle of the inclination of the foot axis to the horizontal it was necessary to solve a problem of transformation of any pressure picture to the given length and width (for vertical location of the foot axis). METHODS:: The transformed coordinates X(*) and Y(*) are defined from the following system of equations: X = Y(*) x width/width(*) x sina + X(*) x length/length(*) x cosa + X(o) Y = Y(*) x width/width(*) x cosa - X(*) x length/length(*) x sina + Y(o) where X(o), Y(o) -- the coordinates of the origin; width, length, a -- parameters of the initial pressure picture; width(*), length(*) -- parameters of the transformed pressure picture. X, Y -- are real values but the values of pressure are defined only in the integer coordinates. One of the main parts of average is a reduction of the time process to any number of frames. If N(max) is a maximum number of frames in one of the data files than all other data files must be transformed with accordance to the following formulas: for j-th frame the value of pressure for sensor with coordinates x, y is: P(*1)(xy) (j) = P(1)(xy) (i) x ki + P(1)(xy) (i+1) x k(i+1), where P(1)(xy) (i) -- pressure for sensor with coordinates x, y of the 1-th initial time process; j = 0, ellipsis, N(max) -1; j-th frame corresponds to i-th frame in the initial time process as i = integer(j) x (N -1)/(N(max) -1), N -- a number of frames in initial time process; coefficients k(i) and k(i+1) are equal to k(i) = j x (N -1)/(N(max) -1) -- I, k(i+1) = 1 - ki. Then the values of pressure are averaged: P(*)(xy) (j) = 1 / L x S P(1)(xy) (j), where L is a number of data files. Corresponding software was developed. RESULTS AND DISCUSSION:: 38 files of pressure measurement data of the teenagers from military school were chosen for transformation and average. Statistical errors of transformation and average were calculated for each data file (for peak pressure, force and area) and for all parameters from two software packages: Novel-win v.95.121 and Novel-ortho v.95.121. The maximum errors are for peak pressures (about 7%-10%) and contact areas (about 6%). CONCLUSION:: Transformation and average programs can be successfully used for data analysis additionally to traditional statistical software: groupmask evaluation program.