detection

detection

Automatic arterial input function detection.

This module implements automatic AIF detection from DCE-MRI data using signal characteristics to identify vascular voxels.

GPU/CPU agnostic using the xp array module pattern. NO scipy dependency - uses custom implementations for filtering and labeling.

References

Mouridsen K et al. (2006). Automatic selection of arterial input function using cluster analysis. Magn Reson Med 55(3):524-531.

Peruzzo D et al. (2011). Automatic selection of arterial input function on dynamic contrast-enhanced MR images. Comput Methods Programs Biomed 104(3):e148-e157.

AIFDetectionParams dataclass

AIFDetectionParams(
    min_peak_enhancement=3.0,
    max_fwhm=15.0,
    early_arrival_percentile=10.0,
    n_candidates=100,
    cluster_size=3,
    use_mask=True,
    smoothing_sigma=0.5,
)

Parameters for automatic AIF detection.

ATTRIBUTE	DESCRIPTION
min_peak_enhancement	Minimum peak enhancement ratio relative to baseline. TYPE: float
max_fwhm	Maximum full-width at half-maximum (seconds). TYPE: float
early_arrival_percentile	Percentile for early arrival time selection. TYPE: float
n_candidates	Number of candidate voxels to consider. TYPE: int
cluster_size	Minimum cluster size for valid AIF region. TYPE: int

AIFDetectionResult dataclass

AIFDetectionResult(
    aif,
    voxel_mask,
    voxel_indices=list(),
    quality_score=0.0,
    detection_params=AIFDetectionParams(),
)

Result of automatic AIF detection.

ATTRIBUTE	DESCRIPTION
aif	Detected arterial input function. TYPE: ArterialInputFunction
voxel_mask	Binary mask of voxels used for AIF. TYPE: NDArray
voxel_indices	List of (x, y, z) indices of selected voxels. TYPE: list
quality_score	Quality metric for the detected AIF (0-1). TYPE: float
detection_params	Parameters used for detection. TYPE: AIFDetectionParams

MultiCriteriaAIFDetector

Bases: BaseAIFDetector

Multi-criteria AIF detection algorithm.

name property

name

Detector display name.

reference property

reference

Literature reference for the detection algorithm.

detect

detect(dataset, params=None, roi_mask=None)

Detect AIF using multi-criteria approach.

PARAMETER	DESCRIPTION
dataset	Input dataset. TYPE: PerfusionDataset
params	Detection parameters. TYPE: AIFDetectionParams | None DEFAULT: None
roi_mask	ROI mask to restrict search. TYPE: NDArray | None DEFAULT: None

RETURNS	DESCRIPTION
AIFDetectionResult	Detection result.

detect_aif

detect_aif(
    dataset, params=None, roi_mask=None, method=None
)

Automatically detect AIF from DCE-MRI data.

Uses a multi-criteria approach: 1. High peak enhancement 2. Early arrival time 3. Narrow peak width (short FWHM) 4. Spatial clustering

PARAMETER	DESCRIPTION
dataset	DCE-MRI dataset with shape (x, y, z, t) or (x, y, t). TYPE: PerfusionDataset
params	Detection parameters. TYPE: AIFDetectionParams DEFAULT: None
roi_mask	Region of interest mask to restrict search. TYPE: NDArray[bool_] DEFAULT: None

RETURNS	DESCRIPTION
AIFDetectionResult	Detection result with AIF and quality metrics.

RAISES	DESCRIPTION
AIFError	If AIF detection fails.