Subscripted assignment between dissimilar structures and Error: The variable info in a parfor cannot be classified. I need educating!

1 次查看（过去 30 天）
Patrick Ford
Patrick Ford 2018-4-25
回答： Walter Roberson 2018-4-26
I am having difficulty trying to speed up code by using parfor and pre-defining variable dimensions. The original code looks something like this(abbreviated):
for i = 1 : num_files_in_folder
Flag = true;
foldername = files_in_folder(i).name;
try
info = dicominfo(foldername);
catch ME
Flag = false;
warning("Problem: dicominfo error.");
DB2(i,1) = "Bad DICOM File";
DB2(i,2) = foldername;
% rethrow(ME);
end %%try
...
When trying to speed it up with parfor I get the "cannot be classified" and other similar error messages.
Then I tried moving some of these temporary variables outside and making them an structured array. Note: I have ram to spare, even with a 8G ram disk. So big variable are not an issue yet.
  2 个评论
Matt J
Matt J 2018-4-25
编辑：Matt J 2018-4-25
I get the "cannot be classified" and other similar error messages.
Post them in their entirety please. Also post the complete loop. Do the messages reference any of the variables you've shown above? I can't see anything wrong with that code snippet.
Stephen23
Stephen23 2018-4-26
Patrick Ford's "Answer" moved here:
OK, Matt - just trying to keep the text volume low. As an explanation: move_Dicom.m is the part I was attempting to improve with parfor.
DicomReadMatout.m is the full program - still need work writing out.
pfortestd.m is code I was trying to use to debug.
string2hash.m is a function called.
Thank you for the interest.

请先登录，再进行评论。

请先登录，再回答此问题。

采纳的回答

Walter Roberson
Walter Roberson 2018-4-26
Here is a marginally cleaned up version that does not have the parfor errors, and which manages directories a little better. Untested.
You should double-check the line
ExamsInfo(i,:) = [{infx(i)},{thisfile}];
I do not understand why you would not use
ExamsInfo(i,:) = {infx(i), thisfile};
%clear
% Script to extract a specified Dicom series from the PET/CT data set
WarmFuzzy = 0;
data_folder = uigetdir;
files_in_folder = dir( fullfile(data_folder, '*.dcm'));
% Discard the current and root file descriptors
files_in_folder = files_in_folder( [files_in_folder.bytes] > 0);
filenames_in_folder = fullfile(data_folder, {files_in_folder.name});
num_files_in_folder = numel(filenamess_in_folder);
DB2 = strings([num_files_in_folder,2]);
dlu = dicomlookup('0008', '103E');
% I searched for a series descriptor called �docs�; You can change it to what you need.
% desired_series_descriptor = 'docs';
ExamsInfo = cell(num_files_in_folder,2);
Flag(1:num_files_in_folder) = true;
Flag1(1:num_files_in_folder) = true;
n = num_files_in_folder;
infx(1:10) = dicominfo(filenames_in_folder{1});
for i = 1:num_files_in_folder
infx(i) = dicominfo(filenames_in_folder{i});
str1(i) = infx(i).(dlu);
end
% foldername(1:num_files_in_folder) = ' ';
parfor i = 1 : num_files_in_folder
newDB2 = strings(1,2);
thisfile = filenames_in_folder{i};
try
infx(i) = dicominfo(thisfile);
catch ME
Flag(i) = false;
warning("Problem: dicominfo error.");
newDB2(1) = "Bad DICOM File";
newDB2(2) = thisfile;
% rethrow(ME);
end %%try
if (Flag(i))
Flag1(i) = true;
try
str1(i) = infx(i).(dlu);
catch ME
warning("No Series Discription");
newDB2(1) = "Nyet";
newDB2(2) = thisfile;
Flag1(i) = false;
end %%try
if(Flag1(i))
newDB2(1) = str1(i);
newDB2(2) = thisfile;
end %% if(Flag1)
% You can convert this section to a series of if statements to match multiple descriptors
% and get the file names for the specific files that you are interested in, and then read the data in
% using DICOM read
ExamsInfo(i,:) = [{infx(i)},{thisfile}];
end % if Flag
DB2(i,:) = newDB2;
% WarmFuzzy = WarmFuzzy +1;
% xt = sprintf('%d Interation\n', WarmFuzzy);
% disp(xt);
end %%parfor i = 1 : num_files_in_folder
disp("Done reading in the directory information");
% parsing the information - creating the database
Dicom_type = {'[GA-N_CTAC_60s] Rest Rb' '[NG_NAC_60s] Stress Rb' 'Cardiac-ldCT' '[NG_NAC_60s] Rest Rb' '[NG_CTAC_60s] Rest Rb' '[NG_CTAC_60s] Stress Rb' '[GA-N_CTAC_60s] Stress Rb'};
% N = size(DB2);
prime = 1033; % 61;
LTA = strings(prime,361);
ptr_lga(prime,361) = int16(0);
% test debug number double due to second run?
for i = 1:prime
ptr_lga(1,1) = 0;
end
nitem(1:prime) = int16(1);
hflag = true;
for i= 1:num_files_in_folder
Logical_Flag = (DB2(i,1) ~= "Bad DICOM File") && (DB2(i,1) ~= "Nyet") ...
&& convertCharsToStrings((ExamsInfo{i, 1}.AccessionNumber) ...
== convertCharsToStrings(ExamsInfo{1, 1}.AccessionNumber));
if (Logical_Flag)
if (hflag)
arr = convertStringsToChars(DB2(i,1));
hash = mod(string2hash(arr), prime)+1;
end %%if (hflag)
hflag = true;
switch LTA(hash,1)
case ""
LTA(hash,1)= DB2(i,1);
nitem(hash) = nitem(hash) + 1;
LTA(hash,nitem(hash)) = DB2(i,2);
% test if this is an image we need
lg2 = false;
lg = ( DB2(i,1) == Dicom_type);
for j = 1:7
lg2 = lg2 | lg(j);
end
if lg2
ptr_lga(hash,nitem(hash)) = int16(ExamsInfo{i,1}.InstanceNumber);
ptr_lga(hash,1) = ptr_lga(hash,1) +1;
end
case DB2(i,1)
nitem(hash) = nitem(hash) + 1;
LTA(hash,nitem(hash)) = DB2(i,2);
% test if this is an image we need
lg2 = false;
lg = ( DB2(i,1) == Dicom_type);
for j = 1:7
lg2 = lg2 | lg(j);
end
if lg2
ptr_lga(hash,nitem(hash)) = int16(ExamsInfo{i,1}.InstanceNumber);
ptr_lga(hash,1) = ptr_lga(hash,1) +1;
end
otherwise
if ~(LTA(hash,1) == DB2(i,1))
hflag = false;
i = i-1;
hash = mod((hash+7),prime);
else
error('Unexpected event - Halting execution');
end % if ~(LTA(hash,1) == DB2(i,1))
end % switch LTA
end % if num_files_in_folder
end
Dicom_name = {"Rest_gated_AC" "StressNG_NAC" "Cardiac_CT" "Rest_NG_NAC" "Rest_NG_AC" "Stress_NG_AC" "Stress_gated_AC"};
exit;
hashtag(1:7) = int16(0);
Rand_Num_Name = string( int16(rand(1,1) * 50000));
for i=1:7
hashtag(i) = mod(string2hash(char(Dicom_type(i))), prime)+1;
while (Dicom_type(i) ~= LTA(hashtag(i)))
hashtag(i) = mod((hashtag(i)+7),prime);
end % while
slicex = dicomread(LTA(hashtag(i),2));
num_slices = ptr_lga(hashtag(i),1);
[x,y] = size(slicex);
Dimage = int16(zeros(num_slices,x,y));
for k = 1:num_slices
indx = ptr_lga(hashtag(i),(k+1));
Dimage(indx, 1:x,1:y) = dicomread(LTA(hashtag(i),(k+1)));
end
% Now write the file out
save(strcat(string(Dicom_name(i)),"_", Rand_Num_Name, ".mat"), 'Dimage')
end
IDt.PatientName = ExamsInfo{1, 1}.PatientName;
IDt.AccessionNumber = ExamsInfo{1, 1}.AccessionNumber;
IDt.StudyDate = ExamsInfo{1, 1}.StudyDate;
save(char(strcat("IDt", Rand_Num_Name, ".mat")), 'IDt')

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Data Import and Analysis 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by