Python

Neuroslice provides a Python API for integration into your scripts and processing pipelines.

Core Functions

The main functions available are:

predict() - Generate mask from array data
predict_mask() - Generate mask from NIfTI file
predict_multi_axis() - Process multiple orientations
mask2cuboid() - Convert mask to bounding cuboid
unite_masks() - Generate the union of multiple masks

Basic Usage

Generate Mask from File

from neuroslice import predict_mask
import nibabel as nib

# Generate mask
mask = predict_mask("input.nii.gz", axis=1, verbose=True)

# Save output
nifti = nib.load("input.nii.gz")
output = nib.Nifti1Image(mask.astype("uint8"), nifti.affine, nifti.header)
nib.save(output, "output_mask.nii.gz")

Process from Array

from neuroslice import predict
import nibabel as nib

# Load data
nifti = nib.load("input.nii.gz")
data = nifti.get_fdata()

# Generate mask
mask = predict(data, axis=1, verbose=True)

Advanced Usage

Multiple Orientations

Process along multiple axes and combine results:

from neuroslice import predict_multi_axis
import nibabel as nib

# Load data
nifti = nib.load("input.nii.gz")
data = nifti.get_fdata()

# Process multiple axes
mask = predict_multi_axis(data, axis=[0, 1, 2], verbose=True)

# Save result
output = nib.Nifti1Image(mask.astype("uint8"), nifti.affine, nifti.header)
nib.save(output, "combined_mask.nii.gz")

Or using predict_mask:

from neuroslice import predict_mask

# Process multiple axes directly
mask = predict_mask("input.nii.gz", axis=[0, 1, 2], verbose=True)

Convert to Cuboid

Convert a mask to its bounding cuboid:

from neuroslice import mask2cuboid

# Convert existing mask
cuboid_mask = mask2cuboid(mask)

Combine Multiple Masks

Unite masks from different processing runs:

from neuroslice import unite_masks

# Generate masks from different orientations
mask1 = predict(data, axis=0, verbose=False)
mask2 = predict(data, axis=1, verbose=False)
mask3 = predict(data, axis=2, verbose=False)

# Combine using union
combined = unite_masks(mask1, mask2, mask3)

Complete Workflow Example

Processing Pipeline

import nibabel as nib
from neuroslice import predict_mask, mask2cuboid
from pathlib import Path

def process_brain_scan(input_path, output_dir, use_cuboid=False):
    """
    Process a brain scan and save the tumor mask.

    Args:
        input_path: Path to input NIfTI file
        output_dir: Directory to save output
        use_cuboid: Whether to use cuboid mode
    """
    # Create output directory
    output_dir = Path(output_dir)
    output_dir.mkdir(parents=True, exist_ok=True)

    # Generate mask
    print(f"Processing {input_path}...")
    mask = predict_mask(input_path, axis=1, verbose=True)

    # Convert to cuboid if requested
    if use_cuboid:
        mask = mask2cuboid(mask)

    # Save output
    input_name = Path(input_path).stem.replace('.nii', '')
    output_path = output_dir / f"{input_name}_mask.nii.gz"

    nifti = nib.load(input_path)
    output = nib.Nifti1Image(mask.astype("uint8"), nifti.affine, nifti.header)
    nib.save(output, output_path)

    print(f"Mask saved to: {output_path}")
    return mask

# Use the function
mask = process_brain_scan("brain.nii.gz", "results", use_cuboid=True)