oct_tissuemasking

oct_tissuemasking is a command-line tool for generating tissue masks from 3D Optical Coherence Tomography (OCT) data. It uses lightweight, pre-trained models and patch-based predictions to output binarized tissue masks.

Features

• Predict Tissue Masks: Generate binarized tissue masks for OCT volumes using a trained model.
• Customizable Parameters: Adjust patch size and step size for the prediction process.

1 Installation

1.1 Create a new Mamba Environment

Create a new mamba environment called oct_tissuemasking with python 3.9.

>>> mamba create -n oct_tissuemasking python=3.9
>>> mamba activate oct_tissuemasking

1.2 Install oct_tissuemasking from PyPi

Now we can just install the oct_tissuemasking package from PyPi!

>>> pip install oct_tissuemasking

2 Usage

2.1 Predict Tissue Masks

Use the following command to make a tissue mask from the OCT volume located at--in-path.

python oct_tissuemasking predict --in-path <INPUT_PATH> \
                                 --out-path <OUTPUT_PATH> \
                                [--model <MODEL_VERSION>] \ 	# Optional, default 1
                                [--patch-size <PATCH_SIZE>] \ 	# Optional, default 128
                                [--step-size <STEP_SIZE>] 	# Optional, default 128

Parameters:

• --in-path: Path to the input NIfTI file that contains a 3D OCT image.
• --out-path: Path to save the output binarized tissue mask (NIfTI format).
• --model (optional): Version of the model to use. Defaults to the version specified in the package.
• --patch-size (optional): Size of the model input patch (default: 128).
• --step-size (optional): Step size between adjacent patches during prediction (default: 64).

Example:

To generate a tissue mask:

python oct_tissuemasking predict --in-path input_volume.nii.gz \
                                 --out-path output_mask.nii.gz \
                                 --model 1 \
                                 --patch-size 128 \
                                 --step-size 64