Pca Guided Idif Steps

This module provides classes and utilities for computing Image-Derived Input Functions (IDIF) using a PCA-guided approach. It is designed to integrate into preprocessing pipelines and offers functionality for optimizing voxel selection, applying fitting techniques, and selecting top-ranked voxels from PCA components.

Classes:

1. PCAGuidedIDIFMixin:

   • Acts as a mixin to provide shared functionality across PCA-guided steps.

   • Manages input/output paths, initializes configurations, and helps connect pipeline steps.

   • Provides methods to infer output paths and set input paths dynamically.

2. PCAGuidedFitIDIFStep:

   • Represents a pipeline step for optimizing voxel selection to fit the IDIF using PCA-guided methods.

   • Extends both ObjectBasedStep and PCAGuidedIDIFMixin to integrate optimization fitting into a pipeline.

   • Includes parameters to control optimization behavior, such as:

     • alpha: Weight for the peak term to encourage a high TAC peak.

     • beta: Weight for the smoothness term to ensure low total variation.

     • method: The optimization method for fitting.

3. PCAGuidedTopVoxelsIDIFStep:

   • Represents a pipeline step to calculate the IDIF by selecting the top N voxels from a specific PCA component.

   • Extends both ObjectBasedStep and PCAGuidedIDIFMixin for pipeline functionality.

   • Allows customizing the selected PCA component and the number of voxels to include.

Key Features:

• Pipeline Integration:

  All classes are designed to be modular and compatible with preprocessing pipelines. They inherit functionalities from ObjectBasedStep and StepsAPI.

• BIDS-Compatible Output:

  Output paths for TAC files are generated in a BIDS-like format using subject and session information extracted from the input image paths.

• Optimization and Voxel Selection:

  Classes include functionality to:

  • Select voxels based on PCA component ranking.

  • Optimize voxel selection iteratively for better IDIF calculation.

  • Use terms like smoothness, noise reduction, and peak emphasis for guiding optimization.

Classes

PCAGuidedIDIFMixin	Mixin class to provide shared functionality for PCA-guided IDIF pipeline steps.
PCAGuidedFitIDIFStep	Pipeline step for fitting the PCA-guided IDIF using an optimization-based approach.
PCAGuidedTopVoxelsIDIFStep	Pipeline step for PCA-guided IDIF calculation using top-ranked voxels.