How do I index within a structure?

I am trying to take the mean of a structure.field column's number for multiple patients in a structure and store it for a plot of all the means later.

ShiftData(1).meanX = mean([patient(applicablepatients).Xtumorshift(1)])

This causes an error: Expected one output from a curly brace or dot indexing expression, but there were 65 results.

The MATLAB Help Documentation " Access Data in a Structure Array" states, "Note: You can index into part of a field only when you refer to a single element of a structure array. MATLAB® does not support statements such as S(1:2).X(1:50,1:80), which attempt to index into a field for multiple elements of the structure."

It appears that what I am trying to do is not supported. Am I understanding this correctly? Why is it not supported? Must I first store the data in a vector and then take the mean of that vector? Why?

Here is the contents (ID changed) of the second element in structure 'patient', patient(2):

             ID: 999999
     Xboneshift: [1x21 double]
     Yboneshift: [1x21 double]
     Zboneshift: [1x21 double]
    Xtumorshift: [1x21 double]
    Ytumorshift: [1x21 double]
    Ztumorshift: [1x21 double]
       TableLat: [1x21 double]
      TableLong: [1x21 double]
      TableVert: [1x21 double]
         TxDate: {1x21 cell}
       FxNumber: [1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21]

(Is the correct terminology "structure(element).field"?)

Answer 1

[Daniel Bridges]

Instead of indexing within an array stored in a structure element, I sorted the structure elements' arrays into a new structure, essentially David Sanchez's suggestion.

One can index into a structure field the same way one indexes into an array, if that field is an array. Working with ulrickls's dicomrt2matlab files, a DICOM-RT contour can be visualized:

scatter3(contours(9).Points(:,1),contours(9).Points(:,2),contours(9).Points(:,3))

Answer 2

[David Sanchez]

% dummy struct
your_struct.field_1 = rand(5,1);
your_struct.field_2 = rand(5,1);

% array with field-mean
your_means = structfun(@mean,your_struct);

your_means =

      0.6307
      0.5440

[David Sanchez]

By the way, your terminology "structure(element).field" is correct.

For your case, you can also try something like this:

%dummy struct array
patient(1).name = 'John Doe';
patient(1).billing = 127.00;
patient(1).test = [79, 75, 73];

patient(2).name = 'Ann Lane';
patient(2).billing = 28.50;
patient(2).test = [68, 70, 68];

for k=1:length(patient)
    your_means(k) = mean(patient(k).test);
end

your_means =

     75.6667   68.6667

[Daniel Bridges]

Thank you for telling me of structfun. However, I cannot use it directly because the structure contains a non-scalar field, namely, the treatment date.

Your second method is similar to what I implemented: I want to average "across" element fields, e.g. averaging the columns in your test field for all elements: 79 and 68, 75 and 70, 73 and 68. To this end I wrote this code:

for loop = 1:totalfractionnumber
  for loop2 = 1:NumOfAppPat
     shifts(loop).Xtumorshift(loop2) = patient(applicablepatients(loop2)).Xtumorshift(loop);
     shifts(loop).Ytumorshift(loop2) = patient(applicablepatients(loop2)).Ytumorshift(loop);
     shifts(loop).Ztumorshift(loop2) = patient(applicablepatients(loop2)).Ztumorshift(loop);
  end
  meanvectorx(loop) = mean(shifts(loop).Xtumorshift);
  meanvectory(loop) = mean(shifts(loop).Ytumorshift);
  meanvectorz(loop) = mean(shifts(loop).Ztumorshift);
  stdvectorx(loop) = std(shifts(loop).Xtumorshift);
  stdvectory(loop) = std(shifts(loop).Ytumorshift);
  stdvectorz(loop) = std(shifts(loop).Ztumorshift);
end

However, it still appears less-than-ideal to have to create more vectors: It is not really using the data in the structure, but only copying from it.

[Jason W]

I am wondering the same thing - the point of using structures was to help manage the large number of variables I had. Though dynamic indices has improved my code in storing values, I now need to create plots of those values and am forced to create many variables again.

Answer 3

[Mostafa]

I had the same question and I found the neatest solution in arrayfun. While structfun is useful for applying a function across fields, arrayfun lets us move across elements of one filed.

Considering the second example of David Sanchez:

your_means =arrayfun(@(x) mean(x.test), patient)

your_means =

75.6667 68.6667