Adds PHI-Finder pipeline

.github/workflows/release.yml

@@ -41,8 +41,10 @@ jobs:
     - name: Update build tools
       run: python -m pip install --upgrade pip
 
-    - name: Install dependencies
-      run: pip install -r requirements.txt
+    - name: Install dependencies & pipelines package
+      run: |
+        pip install -r requirements.txt
+        pip install .
 
     - uses: docker/login-action@v2
       with:
@@ -59,7 +61,7 @@ jobs:
       env:
         PUSH: "${{ steps.deployable.outputs.PUSH }}"
       run: >
-        pipeline2app make xnat ./specs/australian-imaging-service/mri/human/neuro/t1w/preprocess.yaml
-        --registry ghcr.io --check-registry --clean-up --loglevel info 
+        pydra2app make xnat ./specs/australian-imaging-service/quality-control/phi-finder.yaml
+        --registry ghcr.io --loglevel info
         --resources-dir ./resources --spec-root ./specs --source-package . $PUSH
-      # --release pipelines-metapackage $RELEASE --tag-latest 
+      # --release pipelines-metapackage $RELEASE --tag-latest --check-registry --clean-up 

requirements.txt

@@ -1,10 +1,9 @@
-pipeline2app >=0.12
-pipeline2app-xnat >= 0.6.2
+pydra2app >=0.18.5
+pydra2app-xnat >= 0.8.1
 phi-finder >= 2025.6.0
 fileformats >= 0.8.6
 fileformats-extras >= 0.2.1
 fileformats-datascience >= 0.1.0
 fileformats-datascience-extras >= 0.1.1
 fileformats-medimage >= 0.4.4
 fileformats-medimage-extras >= 0.1.5
-

specs/australian-imaging-service/mri/human/neuro/t1w/preprocess.yaml

@@ -63,58 +63,58 @@ docs:
     Yeo7 and Yeo17) from a single input anatomical image. Input image should satisfy
     the input requirements for FastSurfer
 commands:
-  single_parc:
-    task: australianimagingservice.mri.human.neuro.t1w.preprocess:single_parc
-    row_frequency: common:Clinical[session] # the pipeline is desgined to run on imaging "sessions" as opposed to "subjects" or "projects"
-    inputs: # List the inputs that are presented to end-user in UI
-      t1w:
-        datatype: medimage/nifti-gz # MIME-type or "MIME-like" format
-        help: "T1-weighted anatomical scan" # description of field presented in U
-        column_defaults:
-          datatype: medimage/nifti-gz # medimage/dicom-series  # the default
-    outputs: # List the outputs generated by the pipeline
-      parcellation:
-        field: parc_image
-        datatype: medimage-mrtrix3/image-format # MIME-type or "MIME-like" format
-        help: parcellation image generated # description of field presented in UI
-      vis_image_fsl:
-        field: vis_image_fsl
-        datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
-        help: 5ttvis image FSL # description of field presented in UI
-      ftt_image_fsl:
-        field: ftt_image_fsl
-        datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
-        help: 5tt image FSL # description of field presented in UI
-      vis_image_freesurfer:
-        field: vis_image_freesurfer
-        datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
-        help: 5ttvis image FreeSurfer # description of field presented in UI
-      ftt_image_freesurfer:
-        field: ftt_image_freesurfer
-        datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
-        help: 5tt image FreeSurfer # description of field presented in UI
-      vis_image_hsvs:
-        field: vis_image_hsvs
-        datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
-        help: 5ttvis image HSVS # description of field presented in UI
-      ftt_image_hsvs:
-        field: ftt_image_hsvs
-        datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
-        help: 5tt image HSVS # description of field presented in UI
-    parameters:
-      parcellation:
-        datatype: field/text # Format of the field
-        help: parcellation type # description of field presented in UI
-        default: desikan # Default value, filled in on the UI
-    configuration: # Additional args passed to arcana.common:shell_cmd
-      mrtrix_lut_dir: /parcellations # /opt/mrtrix3-3.0.2/share/mrtrix3/labelconvert
-      freesurfer_home: &freesurfer_home !join ["/opt/freesurfer-", *freesurfer_version]
-      fs_license: *freesurfer_license
-      cache_dir: /work/pydra-cache
-      fastsurfer_executable:
-        - /fastsurfer/Docker/entrypoint.sh
-        - /fastsurfer/run_fastsurfer.sh
-        - --allow_root
+  # single_parc:
+  #   task: australianimagingservice.mri.human.neuro.t1w.preprocess:single_parc
+  #   row_frequency: common:Clinical[session] # the pipeline is desgined to run on imaging "sessions" as opposed to "subjects" or "projects"
+  #   inputs: # List the inputs that are presented to end-user in UI
+  #     t1w:
+  #       datatype: medimage/nifti-gz # MIME-type or "MIME-like" format
+  #       help: "T1-weighted anatomical scan" # description of field presented in U
+  #       column_defaults:
+  #         datatype: medimage/nifti-gz # medimage/dicom-series  # the default
+  #   outputs: # List the outputs generated by the pipeline
+  #     parcellation:
+  #       field: parc_image
+  #       datatype: medimage-mrtrix3/image-format # MIME-type or "MIME-like" format
+  #       help: parcellation image generated # description of field presented in UI
+  #     vis_image_fsl:
+  #       field: vis_image_fsl
+  #       datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
+  #       help: 5ttvis image FSL # description of field presented in UI
+  #     ftt_image_fsl:
+  #       field: ftt_image_fsl
+  #       datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
+  #       help: 5tt image FSL # description of field presented in UI
+  #     vis_image_freesurfer:
+  #       field: vis_image_freesurfer
+  #       datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
+  #       help: 5ttvis image FreeSurfer # description of field presented in UI
+  #     ftt_image_freesurfer:
+  #       field: ftt_image_freesurfer
+  #       datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
+  #       help: 5tt image FreeSurfer # description of field presented in UI
+  #     vis_image_hsvs:
+  #       field: vis_image_hsvs
+  #       datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
+  #       help: 5ttvis image HSVS # description of field presented in UI
+  #     ftt_image_hsvs:
+  #       field: ftt_image_hsvs
+  #       datatype: medimage-mrtrix3/image-format-gz # MIME-type or "MIME-like" format
+  #       help: 5tt image HSVS # description of field presented in UI
+  #   parameters:
+  #     parcellation:
+  #       datatype: field/text # Format of the field
+  #       help: parcellation type # description of field presented in UI
+  #       default: desikan # Default value, filled in on the UI
+  #   configuration: # Additional args passed to arcana.common:shell_cmd
+  #     mrtrix_lut_dir: /parcellations # /opt/mrtrix3-3.0.2/share/mrtrix3/labelconvert
+  #     freesurfer_home: &freesurfer_home !join ["/opt/freesurfer-", *freesurfer_version]
+  #     fs_license: *freesurfer_license
+  #     cache_dir: /work/pydra-cache
+  #     fastsurfer_executable:
+  #       - /fastsurfer/Docker/entrypoint.sh
+  #       - /fastsurfer/run_fastsurfer.sh
+  #       - --allow_root
   all_parcs:
     task: australianimagingservice.mri.human.neuro.t1w.preprocess:all_parcs
     row_frequency: common:Clinical[session] # the pipeline is desgined to run on imaging "sessions" as opposed to "subjects" or "projects"
@@ -155,7 +155,7 @@ commands:
         help: 5tt image HSVS # description of field presented in UI
     configuration: # Additional args passed to arcana.common:shell_cmd
       mrtrix_lut_dir: /parcellations # /opt/mrtrix3-3.0.2/share/mrtrix3/labelconvert ## should this be 
-      freesurfer_home: *freesurfer_home
+      freesurfer_home: &freesurfer_home !join ["/opt/freesurfer-", *freesurfer_version] ## freesurfer_home: *freesurfer_home
       fs_license: *freesurfer_license
       cache_dir: /work/pydra-cache
       fastsurfer_executable:

specs/australian-imaging-service/quality-control/phi-finder.yaml

@@ -0,0 +1,34 @@
+schema_version: '2.0'
+title: PHI-Finder
+version: &version '2025.6.0'
+authors:
+  - name: Pedro Faustini
+    email: pedro.faustini@mq.edu.au
+  - name: Thomas G. Close
+    email: tom.g.close@gmail.com
+docs:
+  description: >-
+    PHI-Finder is a tool for de-identifying DICOM files. It uses Tesseract OCR to
+    extract text from images and then applies a set of rules to identify and remove
+    sensitive information. The tool is designed to be easy to use and can be run
+    from the command line or as part of a larger pipeline.
+  info_url: https://github.com/Australian-Imaging-Service/phi-finder
+packages:
+  system:
+    - libglib2.0-0 
+    - libgl1  
+    - poppler-utils 
+    - ghostscript
+    - qpdf
+  conda:
+    tesseract: 5.5.0
+  pip:
+    phi-finder: *version
+    torch: 2.6.0
+    thinc: 8.3.4
+commands:
+  phi-finder:
+    row_frequency: common:Clinical[session]
+    task:
+      type: python
+      function: phi_finder.dicom_tools.utils:deidentify_dicom_files