DICOM Workflows in MATLAB

DWiM provides a complete pipeline for processing DICOM series into analysis-ready 3D volumes:

• Automatic DICOM Discovery: Finds and validates DICOM files in directories
• Intelligent Slice Sorting: Orders slices using ImagePositionPatient for accurate 3D reconstruction
• Volume Assembly: Combines 2D slices into consistent 3D volumes
• Orientation Correction: Standardizes anatomical orientation (RAS, LPS, LAS)
• Isotropic Resampling: Creates uniform voxel spacing for analysis
• Quality Validation: Checks volume integrity and provides preprocessing recommendations

% DICOM folder to ML-ready volume in one call
config = dwim.preprocessPipeline_config('inputType', 'dicomdir', 'steps', {'assemble', 'orient', 'resample'});
[volume, metadata] = dwim.preprocessPipeline('dicom_folder/', config);

% Process a DICOM series with full pipeline
[volume, spacing, metadata] = dwim.preprocess3d.buildVolumeFromSeries('dicom_folder/');

% Custom preprocessing options
[volume, spacing, metadata] = dwim.preprocess3d.buildVolumeFromSeries('dicom_folder/', ...
    'TargetOrientation', 'RAS', ...
    'TargetSpacing', 1.0, ...
    'Validate', true);

See the examples/ directory for detailed usage examples:

• 01_basic_connection/: DICOM server connections
• 02_ml_preprocessing/: 2D image preprocessing for ML
• 03_volume_pipeline/: Complete 3D volume processing pipeline
• 04_ml_dataset_builder/: ML dataset preparation with train/val/test splits
• 05_end_to_end_complete/: Complete architecture demonstration (all 5 layers)
• 06_dataset_validation/: Dataset validation and integrity checks
• QUICK_START_GUIDE.m: Quick reference for common workflows

Comprehensive Guides:

• docs/DICOM_TO_ML_WORKFLOW.md: Complete DICOM to ML workflow tutorial with best practices

Clone the repository and add to MATLAB path:

addpath('path/to/DWiM');

• MATLAB R2020a or later
• Image Processing Toolbox
• Medical Imaging Toolbox (recommended)