numpyradiomics.firstorder#

numpyradiomics.firstorder(image, mask, voxelVolume=1, binWidth=25, voxelArrayShift=0, extend=True)[source]#

Compute first-order (intensity-based) statistics for a given image and mask, replicating Pyradiomics first-order features.

Parameters:

  • image (np.ndarray) – 3D image array containing voxel intensities.

  • mask (np.ndarray) – 3D mask array (same shape as image), where non-zero values indicate the ROI.

  • voxelVolume (float, optional) – Volume of a single voxel (used to scale TotalEnergy). Default is 1.

  • binWidth (float, optional) – Width of bins for histogram-based features (entropy, uniformity). Default is 25.

  • voxelArrayShift (float, optional) – Value to add to intensities before computing Energy/RMS. Default is 0.

  • extend (bool, optional) – If True, returns extended features not strictly in the PyRadiomics standard set (05/95 Percentile, CoeffOfVar, Heterogeneity). Default is True.

Returns:

Dictionary containing first-order feature names and their computed values.

Base Features:

  • Energy: Sum of squared voxel intensities.

  • TotalEnergy: Energy scaled by voxelVolume.

  • Entropy: Histogram-based Shannon entropy.

  • Minimum: Minimum intensity.

  • Maximum: Maximum intensity.

  • Mean: Mean intensity.

  • Median: Median intensity.

  • Range: Maximum - Minimum.

  • Variance: Variance of intensities.

  • StandardDeviation: Standard deviation of intensities.

  • Skewness: Measure of asymmetry of the distribution.

  • Kurtosis: Measure of the “tailedness” of the distribution.

  • MeanAbsoluteDeviation: Mean distance of all intensity values from the Mean.

  • RobustMeanAbsoluteDeviation: Mean distance of intensities (10-90th percentile) from the Mean.

  • RootMeanSquared: Square root of the mean of all the squared intensity values.

  • 10Percentile: 10th percentile of intensities.

  • 90Percentile: 90th percentile of intensities.

  • InterquartileRange: 75th percentile - 25th percentile.

  • Uniformity: Sum of squared histogram probabilities.

Extended Features (if extend=True):

  • 05Percentile: 5th percentile of intensities.

  • 95Percentile: 95th percentile of intensities.

  • CoefficientOfVariation: Standard Deviation / Mean.

  • Heterogeneity: Interquartile Range / Median.

Return type:

dict

Example

>>> import numpyradiomics as npr
>>> # Generate a noisy ellipsoid (simulation of a tumor)
>>> img, mask = npr.dro.noisy_ellipsoid(radii_mm=(20, 10, 5), intensity_range=(20, 80))
>>>
>>> # Compute first order statistics
>>> feats = npr.firstorder(img, mask)
>>>
>>> print(f"Mean: {feats['Mean']:.2f}")
Mean: 50.02
>>> print(f"Range: {feats['Range']:.2f}")
Range: 59.91
On this page

This Page