API Reference

This section provides detailed API documentation for all MedTDA modules and classes.

Overview 

MedTDA is organized into several key modules:

featureextractor - Main interface for TDA feature extraction
preprocessor - Image preprocessing utilities
ph_computer - Persistent homology computation
barcodeextractor - Raw barcode extraction
vectorizers - Methods to convert barcodes to feature vectors
loaders - Image and mask loading/saving
utils - General utility functions
cli - Command-line interface

Quick Navigation 

Module	Description
FeatureExtractor	Main FeatureExtractor class for end-to-end feature extraction
Preprocessor	Preprocessor class for image normalization and preparation
PH Computer	Functions for computing persistent homology
BarcodeExtractor	BarcodeExtractor class for raw barcode computation
Vectorizers	All 8 vectorization methods (functions)
Loaders	Image and mask loading/saving functions
Utils	General utility functions for image processing
CLI	Command-line interface functions and argument parsing

Typical Usage Pattern 

The typical workflow uses the FeatureExtractor class as the main interface:

from medtda import FeatureExtractor

# Initialize with configuration
extractor = FeatureExtractor(
    vectorization_method='PersImage',
    normalize=True
)

# Extract features
features = extractor.execute('image.nii.gz', mask='mask.nii.gz')

For more control, you can use individual components:

from medtda import Preprocessor, BarcodeExtractor
from medtda.vectorizers import persistence_image

# Step-by-step processing
preprocessor = Preprocessor(normalize=True)
processed_image, _ = preprocessor.preprocess(image, mask)

extractor = BarcodeExtractor()
barcodes = extractor.execute(processed_image)

features = persistence_image(barcodes['H1'])

Module Documentation 

Core Classes 

FeatureExtractor 

The main interface for TDA feature extraction. See FeatureExtractor for details.

Key methods:

execute(image, mask=None) - Extract features from an image
set_vectorization_method(method, **params) - Change vectorization method
enable_vectorization_methods(methods) - Enable multiple methods

Preprocessor 

Handles image preprocessing operations. See Preprocessor for details.

Key methods:

preprocess(image, mask=None) - Apply all preprocessing steps
Individual methods for normalization, resampling, windowing, etc.

BarcodeExtractor 

Extracts raw persistence barcodes. See BarcodeExtractor for details.

Key methods:

extract(image) - Compute persistence barcodes
set_filtration_type(filtration_type) - Change filtration type
set_construction(construction) - Change construction method

Vectorization Functions 

MedTDA provides 8 different vectorization methods as standalone functions:

Function	Description
`persistence_stats(barcode)`	Statistical summaries (mean, std, min, max, etc.)
`betti_curve(barcode, resolution)`	Betti numbers as function of filtration value
`persistence_image(barcode, bandwidth, resolution)`	2D histogram weighted by persistence
`persistence_landscape(barcode, resolution, num_landscapes)`	Functional representation using landscapes
`persistence_silhouette(barcode, resolution, weight)`	Average persistence landscape
`entropy_summary(barcode, resolution)`	Information-theoretic features
`persistence_lifespan(barcode, resolution)`	Distribution of feature lifespans
`persistence_tropical_coordinates(barcode, r)`	Tropical algebra representation

See Vectorizers for detailed documentation of each method.

Utility Functions 

The utils module provides various helper functions:

resample_image() - Image resampling
normalize_image() - Intensity normalization
apply_mask() - Apply binary mask
crop_to_roi() - Crop to region of interest
extract_label_from_mask() - Extract single label from multi-label mask

See Utils for complete documentation.

Image Loading and Saving 

The loaders module handles I/O operations:

load_image(path) - Load 2D or 3D images (NIFTI, PNG, TIFF, etc.)
load_mask(path) - Load binary or multi-label masks
save_image(array, path) - Save images to disk
Validation functions for images and masks

See Loaders for complete documentation.

Command-Line Interface 

The cli module provides the command-line interface:

create_parser() - Create argument parser
process_single_file(args) - Process single image
process_batch(args) - Process batch of images
main() - Main entry point

See CLI for complete documentation.