Using dl+DiReCT on Mac Studio

[SDFsci]

Hello,

I am wondering if you can provide more detailed step by step instructions on how to successfully execute this program in a Mac terminal (I have an M2 Mac Studio). I have installed miniconda, confirmed python3.12, cloned the repository, run setup.py for HD-BET and dl+direct) and am still struggling to get the program to work. My coding skills are limited, as you can see. Appreciate your help and maybe this will help many others as well!

Best,

Shawn

[mrunibe]

Hi Shawn,

It should actually run on Mac as well, but it requires python 3.11 instead of 3.12.

Create conda environment with python 3.11:

conda create -n DL_DiReCT python=3.11
conda activate DL_DiReCT

Then follow the instructions to install DL+DiRect: https://github.com/SCAN-NRAD/DL-DiReCT#install-dldirect

Now you should be able to run with

dl+direct  --bet --lowmem path_to_T1.nii.gz out

This will run on the CPU only and hence rather slow. However, it should soon also be possible to leverage the GPU on Mac with a newer pytorch version.

If you still experience any problems, please let me know and also post any error messages that you see.

[SDFsci]

Hi Michael, Thanks for the tip - I didn't realize using a newer python version would be problematic. I will install python 3.11, follow your instructions, and see if it works. Will let you know shortly. Thanks, Shawn

…

On Thu, Oct 10, 2024 at 3:35 PM Michael Rebsamen ***@***.***> wrote: Hi Shawn, It should actually run on Mac as well, but it requires python 3.11 instead of 3.12. Create conda environment with python 3.11: conda create -n DL_DiReCT python=3.11 conda activate DL_DiReCT Then follow the instructions to install DL+DiRect: https://github.com/SCAN-NRAD/DL-DiReCT#install-dldirect Now you should be able to run with dl+direct --bet --lowmem path_to_T1.nii.gz out This will run on the CPU only and hence rather slow. However, it should soon also be possible to leverage the GPU on Mac with a newer pytorch version. If you still experience any problems, please let me know and also post any error messages that you see. — Reply to this email directly, view it on GitHub <#6 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BFVKD34XRGMZ7SU5GJSIOZTZ23QIJAVCNFSM6AAAAABPXEDCPOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBVHE4TCNJTHA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***>

[SDFsci]

Hi Michael, I installed python 3.11 and followed the steps listed in your email and on the website. When I attempt to run the program, I receive the message "zsh: command not found: dl+direct" Thanks again for your help. Best, Shawn On Thu, Oct 10, 2024 at 4:43 PM Shawn Flanagan < ***@***.***> wrote:

…

Hi Michael, Thanks for the tip - I didn't realize using a newer python version would be problematic. I will install python 3.11, follow your instructions, and see if it works. Will let you know shortly. Thanks, Shawn On Thu, Oct 10, 2024 at 3:35 PM Michael Rebsamen ***@***.***> wrote: > Hi Shawn, > > It should actually run on Mac as well, but it requires python 3.11 > instead of 3.12. > > Create conda environment with python 3.11: > > conda create -n DL_DiReCT python=3.11 > conda activate DL_DiReCT > > Then follow the instructions to install DL+DiRect: > https://github.com/SCAN-NRAD/DL-DiReCT#install-dldirect > > Now you should be able to run with > > dl+direct --bet --lowmem path_to_T1.nii.gz out > > This will run on the CPU only and hence rather slow. However, it should > soon also be possible to leverage the GPU on Mac with a newer pytorch > version. > > If you still experience any problems, please let me know and also post > any error messages that you see. > > — > Reply to this email directly, view it on GitHub > <#6 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/BFVKD34XRGMZ7SU5GJSIOZTZ23QIJAVCNFSM6AAAAABPXEDCPOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBVHE4TCNJTHA> > . > You are receiving this because you authored the thread.Message ID: > ***@***.***> > -- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***>

-- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***>

[SDFsci]

One more thing, if it is helpful - I am trying to run the program in my (default) zsh terminal. On Fri, Oct 11, 2024 at 11:38 AM Shawn Flanagan < ***@***.***> wrote:

…

Hi Michael, I installed python 3.11 and followed the steps listed in your email and on the website. When I attempt to run the program, I receive the message "zsh: command not found: dl+direct" Thanks again for your help. Best, Shawn On Thu, Oct 10, 2024 at 4:43 PM Shawn Flanagan < ***@***.***> wrote: > Hi Michael, > > Thanks for the tip - I didn't realize using a newer python version would > be problematic. I will install python 3.11, follow your instructions, and > see if it works. Will let you know shortly. > > Thanks, > > Shawn > > On Thu, Oct 10, 2024 at 3:35 PM Michael Rebsamen < > ***@***.***> wrote: > >> Hi Shawn, >> >> It should actually run on Mac as well, but it requires python 3.11 >> instead of 3.12. >> >> Create conda environment with python 3.11: >> >> conda create -n DL_DiReCT python=3.11 >> conda activate DL_DiReCT >> >> Then follow the instructions to install DL+DiRect: >> https://github.com/SCAN-NRAD/DL-DiReCT#install-dldirect >> >> Now you should be able to run with >> >> dl+direct --bet --lowmem path_to_T1.nii.gz out >> >> This will run on the CPU only and hence rather slow. However, it should >> soon also be possible to leverage the GPU on Mac with a newer pytorch >> version. >> >> If you still experience any problems, please let me know and also post >> any error messages that you see. >> >> — >> Reply to this email directly, view it on GitHub >> <#6 (comment)>, >> or unsubscribe >> <https://github.com/notifications/unsubscribe-auth/BFVKD34XRGMZ7SU5GJSIOZTZ23QIJAVCNFSM6AAAAABPXEDCPOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBVHE4TCNJTHA> >> . >> You are receiving this because you authored the thread.Message ID: >> ***@***.***> >> > > > -- > *Shawn Flanagan, PhD* > Associate Professor > Director, Human Performance Laboratory > Department of Podiatric Medicine and Surgery > Dr. William M. Scholl College of Podiatric Medicine > Neuroscience Program > School of Graduate and Postdoctoral Studies > ***@***.*** ***@***.***> > > > -- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***>

-- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***>

[mrunibe]

Hi Shawn,

I suspect something went wrong in the installation, but hard to tell what without all the details.

Would you be able to export the entire shell output? If you still have it all in the terminal where you did the installation and also tried to run 'dl+direct': Choose 'Shell' -> 'Export Text As...' and save as .txt and send to me. That would be very helpful.

If you don't have the terminal output anymore, just run the installation and dl+direct again in a new terminal and export the output.

Best Regards,
Michael

[SDFsci]

Hi Michael, No problem - here you go - really appreciate your help! Thanks again. Best, Shawn

…

On Fri, Oct 11, 2024 at 3:00 PM Michael Rebsamen ***@***.***> wrote: Hi Shawn, I suspect something went wrong in the installation, but hard to tell what without all the details. Would you be able to export the entire shell output? If you still have it all in the terminal where you did the installation and also tried to run 'dl+direct': Choose 'Shell' -> 'Export Text As...' and save as .txt and send to me. That would be very helpful. If you don't have the terminal output anymore, just run the installation and dl+direct again in a new terminal and export the output. Best Regards, Michael — Reply to this email directly, view it on GitHub <#6 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BFVKD3YTPPLERKP7EUXT4LDZ3AU5TAVCNFSM6AAAAABPXEDCPOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBYGA2DMOBQHA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***> Last login: Fri Oct 11 11:23:15 on ttys002 ***@***.*** ~ % conda create -n DL_DiReCT python=3.11 WARNING: A conda environment already exists at '/opt/miniconda3/envs/DL_DiReCT' Remove existing environment (y/[n])? y Channels: - defaults Platform: osx-arm64 Collecting package metadata (repodata.json): done Solving environment: done ## Package Plan ## environment location: /opt/miniconda3/envs/DL_DiReCT added / updated specs: - python=3.11 The following NEW packages will be INSTALLED: bzip2 pkgs/main/osx-arm64::bzip2-1.0.8-h80987f9_6 ca-certificates pkgs/main/osx-arm64::ca-certificates-2024.9.24-hca03da5_0 libffi pkgs/main/osx-arm64::libffi-3.4.4-hca03da5_1 ncurses pkgs/main/osx-arm64::ncurses-6.4-h313beb8_0 openssl pkgs/main/osx-arm64::openssl-3.0.15-h80987f9_0 pip pkgs/main/osx-arm64::pip-24.2-py311hca03da5_0 python pkgs/main/osx-arm64::python-3.11.10-hb885b13_0 readline pkgs/main/osx-arm64::readline-8.2-h1a28f6b_0 setuptools pkgs/main/osx-arm64::setuptools-75.1.0-py311hca03da5_0 sqlite pkgs/main/osx-arm64::sqlite-3.45.3-h80987f9_0 tk pkgs/main/osx-arm64::tk-8.6.14-h6ba3021_0 tzdata pkgs/main/noarch::tzdata-2024b-h04d1e81_0 wheel pkgs/main/osx-arm64::wheel-0.44.0-py311hca03da5_0 xz pkgs/main/osx-arm64::xz-5.4.6-h80987f9_1 zlib pkgs/main/osx-arm64::zlib-1.2.13-h18a0788_1 Proceed ([y]/n)? y Downloading and Extracting Packages: Preparing transaction: done Verifying transaction: done Executing transaction: done # # To activate this environment, use # # $ conda activate DL_DiReCT # # To deactivate an active environment, use # # $ conda deactivate ***@***.*** ~ % conda activate DL_DiReCT ***@***.*** ~ % cd ${HOME} git clone https://github.com/SCAN-NRAD/DL-DiReCT.git cd DL-DiReCT pip install numpy && pip install -e Cloning into 'DL-DiReCT'... ^[[Dremote: Enumerating objects: 201, done. remote: Counting objects: 100% (106/106), done. remote: Compressing objects: 100% (68/68), done. remote: Total 201 (delta 62), reused 80 (delta 38), pack-reused 95 (from 1) Receiving objects: 100% (201/201), 163.53 MiB | 40.37 MiB/s, done. Resolving deltas: 100% (107/107), done. Collecting numpy Using cached numpy-2.1.2-cp311-cp311-macosx_14_0_arm64.whl.metadata (60 kB) Using cached numpy-2.1.2-cp311-cp311-macosx_14_0_arm64.whl (5.4 MB) Installing collected packages: numpy Successfully installed numpy-2.1.2 Usage: pip install [options] <requirement specifier> [package-index-options] ... pip install [options] -r <requirements file> [package-index-options] ... pip install [options] [-e] <vcs project url> ... pip install [options] [-e] <local project path> ... pip install [options] <archive url/path> ...

-e option requires 1 argument ***@***.*** DL-DiReCT % dl+direct --subject ACL2 --bet /Users/neurolab/Documents/BBACL/5_DL_DiReCT/Sources/ACL2_T1.nii.gz /Users/neurolab/Documents/BBACL/5_DL_DiReCT/Outputs zsh: command not found: dl+direct ***@***.*** DL-DiReCT %

[mrunibe]

Hi Shawn,

There is a dot (.) missing at the end of the command, it should be exactly like this:

pip install numpy && pip install -e .

The dot (.) refers to the current working directory, telling pip to install it from here.

[SDFsci]

Hi Michael, It worked! Thank you for your help. The power of a simple dot. I tried to run the batch version, but cannot seem to get a working java runtime version in place (the batch command stated that "parallel not found", and applying the recommended script (sudo apt install parallel) led to message "The operation couldn’t be completed. Unable to locate a Java Runtime that supports apt. Please visit http://www.java.com for information on installing Java". Am not sure if I am missing something, or if there is a workaround - from your previous email, it seems that a few things may need to happen before batching/GPU/parallel functions can be done on M2 Mac? Thanks again and cheers, Shawn

…

On Fri, Oct 11, 2024 at 4:44 PM Michael Rebsamen ***@***.***> wrote: Hi Shawn, There is a dot (.) missing at the end of the command, it should be exactly like this: pip install numpy && pip install -e . The dot (.) refers to the current working directory, telling pip to install it from here. — Reply to this email directly, view it on GitHub <#6 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BFVKD33N2PCRCJHTQV5QVDLZ3BBDRAVCNFSM6AAAAABPXEDCPOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBYGE2TQNZVGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- *Shawn Flanagan, PhD* Associate Professor Director, Human Performance Laboratory Department of Podiatric Medicine and Surgery Dr. William M. Scholl College of Podiatric Medicine Neuroscience Program School of Graduate and Postdoctoral Studies ***@***.*** ***@***.***>

[mrunibe]

Hi Shawn,

The purpose of the batch version (with the help of the linux parallel command) is to process several MRI in parallel on a high-performance infrastructure with a GPU powerful enough to run several jobs simultaneously. This is likely not the case with an M2 and I would recommend to run multiple cases sequentially.

Kind Regards,
Michael

[mrunibe]

Should run now also natively on Apple silicon with MPS.

Batch mode requires installing the parallel command using:

conda install conda-forge::parallel