Extremly high memory consumption
Closed this issue · 4 comments
Hi,
I wanted to convert a dataset with your cool tool and for most of it it worked like a charm :)
However for a couple files I got an extremly high memory consumption (used up to 60GB on a machine with 62GB RAM) that the process got terminated by the system.
Do you know why that could be?
Best
Karol
Images are loaded fully into the memory. Whats the dimensions of your dicom? :)
Thanks for the quick reply!
The dimensions for this one problematic case is 1280x1280x249.
Are you converting multiple masks in the rtstruct? If so you could try selecting only the relevant mask using the -s option
dcmrtstruct2nii convert --help
USAGE
console convert [-r <...>] [-d <...>] [-o <...>] [-g [<...>]] [-s [<...>]] [-f [<...>]] [-b [<...>]] [-c [<...>]]
OPTIONS
-r (--rtstruct) Path to DICOM RT Struct file, example: /tmp/DICOM/resources/secondary/rtstruct.dcm
-d (--dicom) Path to original DICOM file, example: /tmp/DICOM/resources/files
-o (--output) Output path, example: /tmp/output
-g (--gzip) Optional, gzip output .nii (default: "true")
-s (--structures) Optional, list of structures that need to be converted, example: Patient, Spinal, Dose-1
-f (--mask-foreground-color) Optional, the foreground color used for the mask. Must be between 0-255. (default: "255")
-b (--mask-background-color) Optional, the background color used for the mask. Must be between 0-255. (default: "0")
-c (--convert-original-dicom) Optional, convert the original dicom to nii (default: "true")
GLOBAL OPTIONS
-h (--help) Display this help message
-q (--quiet) Do not output any message
-v (--verbose) Increase the verbosity of messages: "-v" for normal output, "-vv" for more verbose output and "-vvv" for debug
-V (--version) Display this application version
--ansi Force ANSI output
--no-ansi Disable ANSI output
-n (--no-interaction) Do not ask any interactive question
If you need all the masks you could try a script like below to convert them one-by-one.
#!/bin/bash
# extract_one_by_one.sh
STRUCTS=(`dcmrtstruct2nii list --rtstruct $1`)
dcmrtstruct2nii convert --rtstruct $1 --dicom $2 --output $3 -s ${STRUCTS[0]}
echo $STRUCTS
for el in "${STRUCTS[@]:1}"
do dcmrtstruct2nii convert --rtstruct $1 --dicom $2 --output $3 -c no -s "$el"
done
Which you would then call like ./extract_one_by_one.sh /path/to/rtstruct /path/to/dicom /output/dir
Sorry for the late reply! I managed to avoid the process termination now by using a machine with 255GB of memory. Some of the cases used up to 240GB, but they got converted. I did not test if your approach works, but will probably in the future if I have time again. But thanks for your help! Closing this for now.
Best
Karol