Determining heel strike using only kinematic data

Question

Jak 2021-6-25

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/865135-determining-heel-strike-using-only-kinematic-data

评论： Jak 2021-6-29

I am trying to create gait events to split my data in gait cycles using only kinematic data. To do this i am using the foot velocity algorithm. Whilst toe off is simple enough, findings the peaks, the algorithm involves determining heel strike by measuring the second trough in the foot velocity graph (please see attached). I seem to have created a code that works when the second lowest trough is clear and apparent, however my code completely falls apart when the second trough is minimal and/or doesnt exist, leading it missing heel strike events and not splitting the cycles effectively.

Here is my code so far

heel_marker = tTrajectories{:,47};                                          %heel marker trajectories
heel_marker_ascend = sortrows(heel_marker);                                 %ascends the heel markers
heel_marker_threshold_pos = round(length(heel_marker_ascend)*0.35);         %find the 35% position location
heel_marker_threshold = heel_marker_ascend(heel_marker_threshold_pos,1);    %finds the value of the threshold
tr_locs = [];
trough = [];
for i = 1:8:length(R_Foot_vert_vel_midpoint)                                %iterates between every 8 data points
    try
        minimum = min(R_Foot_vert_vel_midpoint(i:i+7));                     %finds minimum within the 8 points
        e = find(R_Foot_vert_vel_midpoint==minimum);                        %finds location of the minimum within the foot velocity data 
        heel_marker_pos = heel_marker(e);
        if (heel_marker_pos < heel_marker_threshold) && (minimum < -50)     %if the marker in the position is lower than the heel marker threshold
            if minimum < R_Foot_vert_vel_midpoint(e-1)                      %gradient is before the trough so its not catching it on the accend of the main trough
                trough = [trough;minimum];                                  %add minimum to the trough matrix
                tr_locs = [tr_locs;e];                                      %find location of the minimums in the dataset
            end
        end
    end
end
numtrough = length(trough);
start_point = fix(1);
R_Heel_Strike = [trough tr_locs];
for i = 1:length(R_Heel_Strike)
    try
        while (R_Heel_Strike(i+1,2) - R_Heel_Strike(i,2)) < 100;            %finds any trough that is less than n frames from the previous peak. Helps remove two peaks/troughs that may be caused by unsmooth data
            if R_Heel_Strike(i+1,2) < R_Heel_Strike(i,2)
                R_Heel_Strike(i,:) = [];
            else
                R_Heel_Strike(i+1,:) = [];                                  %removes the highest of the two frames
            end                                                             %removes the values that are less than n frames apart
        end
    end
end

Hopefully using the example data you can see what I mean, in that it is not picking up every secondary trough per cycle. I would like to add an element to my code that if a trough is not found within the confines here (the heel marker position is lower than the heel marker threshold, the minimum value found is less than 50, and the gradient before the minimum is negative) that a minimum is instead found within the cycle whereby the first two criteria are met.

I hope I have explain this sufficiently, thanks

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2021-6-29

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/865135-determining-heel-strike-using-only-kinematic-data#answer_735563

在 MATLAB Online 中打开

I am not certain that I understand the problem. In these plots, it appears that the heel strike (orange lines) correspond to the correct foot position in the gait cycle.

What result do you want?

RFoot = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/665215/R_Foot_vert_vel_mipoint.xlsx', 'VariableNamingRule','preserve');

RHeel = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/665220/R_Heel_marker.xlsx', 'VariableNamingRule','preserve');

tF = linspace(0, size(RFoot,1)-1, size(RFoot,1));

tH = linspace(0, size(RHeel,1)-1, size(RHeel,1));

figure

plot(tF, RFoot{:,1})

hold on

plot(tH, RHeel{:,1})

hold off

grid

xlabel('Time Index')

ylabel('Amplitude')

legend('RFoot','RHeel')

figure

for k = 1:10

subplot(5,2,k)

plot(tF, RFoot{:,1})

hold on

plot(tH, RHeel{:,1})

hold off

grid on

% xlabel('Time Index')

% ylabel('Amplitude')

% legend('RFoot','RHeel')

xlim([0 1000]+(k-1)*1000)

end

.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Jak 2021-6-29

Thank you for your reply.

I am looking to extract the location of all the second troughs in the midfoot vertical velocity as that point corresponds to heel strike in the gait cycle. My code extracts some of them, but does not find all of them (especially if the trough is not clear) meaning that when I split my data into gait cycles, sometimes they are not split correctly.

I want to add a caveat to my code that if a trough is not found, then the lowest vertical velocity point whilst the heel marker is below the threshold is extracted instead.

请先登录，再进行评论。