Parcitron Module Release
Overview
This release marks the significant addition of the **Parcitron** module to our neuroimaging toolkit. Parcitron is a powerful tool designed for creating random parcellations of NIfTI mask images, offering various options tailored to specific research needs in neuroimaging.
Key Features of Parcitron
1. Random Parcellation
- **Parcitron** allows users to generate random parcellations of a given NIfTI mask. It supports flexible inputs, including specific parcel sizes or the number of parcels, with the option to create contiguous or non-contiguous clusters.
2. Multiple Clustering Methods
- **KMeans Clustering**: Creates evenly distributed clusters with a traditional KMeans approach.
- **Custom Compactor Method**: Offers precise control over cluster sizes with two strategies:
- **Fixed Size**: Ensures that clusters are created with a set number of voxels, regardless of how the clusters are distributed spatially.
- **Balanced Size**: Distributes clusters more evenly, adjusting sizes slightly to avoid creating very small, leftover clusters.
3. Custom Cluster Sizes
- Users can now specify exact cluster sizes, ensuring that clusters are formed with the desired voxel counts. This is particularly useful for creating synthetic datasets or controlled null models in neuroimaging studies.
4. Command-Line Usability
- **Parcitron** is designed for ease of use via the command line, making it easily integrable into automated workflows or pipelines.
- Example commands provided in the script illustrate various use cases, from basic KMeans clustering to advanced, strategy-driven parcellations using the compactor method.
Additional Updates
Codebase Cleanup
- Removed obsolete files and reorganized utility functions to enhance maintainability and accessibility across different projects.
- The functions now follow consistent naming conventions and improved documentation, making them easier to integrate and use in other contexts.
Use Cases
1. Synthetic Lesion Generation
- **Parcitron** can be used to generate synthetic lesions with controlled size and distribution, aiding in the validation of neuroimaging methods.
2. Null Models for Statistical Testing
- The tool is ideal for creating null models with specific spatial characteristics, allowing for robust statistical testing in studies involving brain parcellation.
Getting Started
To explore the capabilities of **Parcitron**, check out the example commands provided in the documentation, or run `parcitron --help` for a full list of options and configurations.