baseline

baseline

Baseline correction for time-series data.

This module provides functions for baseline estimation and correction in perfusion time-series data.

GPU/CPU agnostic using the xp array module pattern.

baseline_correction

baseline_correction(data, baseline_frames=5, method='mean')

Correct baseline in time-series data.

Estimates baseline from initial frames and subtracts or normalizes the signal relative to baseline.

GPU/CPU agnostic - operates on same device as input data.

PARAMETER	DESCRIPTION
data	Time-series data. Last dimension is time. TYPE: NDArray[floating]
baseline_frames	Number of initial frames for baseline estimation. TYPE: int DEFAULT: 5
method	Correction method: - "mean": Subtract mean baseline (additive correction) - "median": Subtract median baseline - "normalize": Divide by mean baseline (multiplicative) - "percent": Convert to percent change from baseline TYPE: str DEFAULT: "mean"

RETURNS	DESCRIPTION
NDArray[floating]	Baseline-corrected data with same shape as input.

RAISES	DESCRIPTION
ValueError	If baseline_frames exceeds data length or invalid method.

Examples:

>>> import numpy as np
>>> from osipy.common.signal.baseline import baseline_correction
>>> data = np.random.rand(64, 64, 20, 30) + 100  # Signal with offset
>>> corrected = baseline_correction(data, baseline_frames=5)
>>> print(corrected.shape)
(64, 64, 20, 30)

estimate_baseline_std

estimate_baseline_std(data, baseline_frames=5)

Estimate noise standard deviation from baseline frames.

GPU/CPU agnostic - operates on same device as input data.

PARAMETER	DESCRIPTION
data	Time-series data. Last dimension is time. TYPE: NDArray[floating]
baseline_frames	Number of initial frames for noise estimation. TYPE: int DEFAULT: 5

RETURNS	DESCRIPTION
NDArray[floating]	Standard deviation estimate for each voxel. Shape is data.shape[:-1].

Examples:

>>> import numpy as np
>>> from osipy.common.signal.baseline import estimate_baseline_std
>>> data = np.random.randn(64, 64, 20, 30)
>>> noise_std = estimate_baseline_std(data, baseline_frames=5)
>>> print(noise_std.shape)
(64, 64, 20)