orcaz 0.1.5

ORCA: Optimized Registration through Conditional Adversarial networks

Requirements ✅

For an optimal experience with ORCA, ensure the following:

• Operating System: ORCA runs smoothly on Windows, Mac, or Linux.
• Memory: At least 32GB of RAM ensures ORCA operates without a hitch.
• GPU: For blazing-fast predictions, an NVIDIA GPU comes highly recommended. But if you don't have one, fret not! ORCA will still get the job done, just at a more leisurely pace. For training new models, you must have a GPU!
• Python: Version 3.9.2 or above. We like to stay updated!

Installation Guide 🛠️

Navigating the installation process is a breeze. Just follow the steps below:

For Linux and MacOS 🐧🍏

1. Create a Python environment, for example, 'orca-env'.

python3 -m venv orca-env

2. Activate your environment.

source orca-env/bin/activate  # for Linux
source orca-env/bin/activate  # for MacOS

3. Install ORCA.

pip install orcaz

Usage Guide 📚

Command-line tool for data folder processing 💻

orcaz -d <path_to_patient_dir> -m <mode>

Here <path_to_patient_dir> refers to the directory containing your subject's PET and CT images. Where <mode> is the name of the mode for which we want to use the tool, from the available options.

train: Yes we can !! Orca can be used to train your own models as a generic cGAN paltform. More instructions for that to follow !

pred : This option will use orca only to generate a synthetic CT from your PET data.

coreg: Option to generate synethic CT and perform the coregistration pipeline with an output of the co-registered CT. ORCA in its full glory !!

Using ORCA requires your data to be structured according to specific conventions. ORCA supports both DICOM and NIFTI formats.

Required Directory Structure 🌳

Please structure your dataset as follows for coregistration:

EXAMPLE_Data_folder/
├── S1
│   ├── S1_CT
│   │   ├── xyz_1.dcm
│   │   ├── xyz_2.dcm
│   │   ├── .
│   │   ├── .
│   │   ├── .
│   │   └── xyz_532.dcm
│   └── S1_FDG_NAC_PT
│       ├── xyz_1.dcm
│       ├── xyz_2.dcm
│       ├── .
│       ├── .
│       ├── .
│       └── xyz_532.dcm
├── S2
│   ├── S2_CT
│   │   ├── xyz_1.dcm
│   │   ├── xyz_2.dcm
│   │   ├── .
│   │   ├── .
│   │   ├── .
│   │   └── xyz_532.dcm
│   └── S2_FDG_NAC_PT
│       ├── xyz_1.dcm
│       ├── xyz_2.dcm
│       ├── .
│       ├── .
│       ├── .
│       └── xyz_532.dcm
├── S3
│   ├── S3_CT.nii
│   └── S3_FDG_NAC_PT.nii
├── S4
│   ├── S4_CT.nii.gz
│   └── S4_FDG_NAC_PT.nii.gz

In all these cases, ORCA can be executed on the directories one by one

orcaz -d S1 -m coreg
orcaz -d S2 -m coreg
orcaz -d S3 -m coreg  

Note: If the necessary naming conventions are not followed, ORCA will not process the data in the directory.

Naming Conventions for files 📝

There is none! Currently orca requires the naming of the subject subfoders inlcude particular names. The patient identifier can be placed in the start of each subfolder

For instance, S1_CT and S1_FDG_NAC_PT, or S1_CT and S1_FDG_AC_PT.
S2_CT and S2_PSMA_NAC_PT.
S3_CT and S3_FACBC_NAC_PT.
S4_CT and S4_DOTA_NAC_PT.
S5_CT and S5_AGNOSTIC_NAC_PT.

Output

After successful completion, the co-registered CT is saved as dicom data in ORCA_CT_DICOM.

Intermediate images and warp files are stored within the ORCA-VXX-YYYY-MM-DD-HH-MM-SS folder

S1
├── CT
├── NAC_FDG_PET
├── ORCA_CT_DICOM
├── ORCA-V01-2023-09-28-00-02-52

Directory Structure for sCT prediction only

The above dataset structure will also work in prediction mode, but the CT directory is not necessary:

For prediction the following is sufficient.

EXAMPLE_Data_folder/
├── S1
│   └── S1_FDG_NAC_PT
│       ├── xyz_1.dcm
│       ├── xyz_2.dcm
│       ├── .
│       ├── .
│       ├── .
│       └── xyz_532.dcm
├── S2
│   └── S2_FDG_NAC_PT
│       ├── xyz_1.dcm
│       ├── xyz_2.dcm
│       ├── .
│       ├── .
│       ├── .
│       └── xyz_532.dcm
├── S3
│   └── S3_FDG_NAC_PT.nii

In all these cases, ORCA can be executed to generate sCT on the directories one by one

orcaz -d S1 -m pred
orcaz -d S2 -m pred
orcaz -d S3 -m pred