v2.1.5

November 21, 2025

Commit: 5943e20

Bug Fixes

v2.1.4

November 19, 2025

Commit: ab80fa2

Bug Fixes

• Fixed minor issues related to RT-STRUCT handling
• Updated logic to use Instance Series UID for RT-STRUCT files to improve consistency and compatibility

v2.1.3

November 7, 2025

Commit: 619225f8

Bug Fixes

• Fixed minor bugs affecting runtime stability
• Modified default CPU core allocation for optimized parallel processing
• Removed additional unnecessary package dependencies

v2.1.2

October 24, 2025

Commit: 0e7c73f

Bug Fixes

• Fixed critical bugs affecting feature extraction stability
• Updated default parameter values for improved performance

v1.0.2

August 20, 2025

Commit: 1c42e6a

Bug Fixes

• Changed package name from PySera to lowercase pysera for consistency across platforms.
• Set default apply_preprocessing to False to prevent unintended preprocessing of ROIs.

v1.0.1

August 20, 2025

Commit: 25d8f71

Bug Fixes

• Fixed bug in NumPy array file processing in in-memory mode, ensuring correct handling of array inputs.

** v1.0.0**

August 19, 2025

Commit: a488ba4

New Features

• Optimized memory management by storing large RoIs as float32 NumPy arrays on disk with memory mapping, ensuring efficient management and preventing RAM overflows.
• Enabled robust feature extraction for very small RoIs (fewer than 10 voxels) using epsilon corrections (ε) and synthetic augmentation, supporting cases as small as single-voxel inputs.
• Corrected mathematical operations prone to errors (e.g., log(0), division by zero, sqrt of negatives, covariance matrix for single voxel, ConvexHull requirements, percentile estimation).
• Improved data loading performance with smoother and more efficient handling of medical imaging inputs.
• Extended supported input formats to include RT-Struct data paths and direct NumPy array inputs for both images and masks.
• Enhanced Excel output with two new tabs: Parameters (record of all parameters used) and Warnings and Errors (detailed runtime issues).
• Added flexible RoI selection controls:
  • roi_num: defines how many RoIs to select.
  • roi_selection_mode: supports per_Img and per_region strategies.
• Exposed radiomics parameters as configurable arguments, replacing hardcoded values.
• Implemented automatic padding to resolve image and mask dimension mismatches.
• Improved RoI naming in Excel outputs by including both label and lesion information when available.
• Performed codebase restructuring: renamed variables for clarity, modularized the implementation into a standalone library, updated requirements, and verified IBSI standard compliance.

initial

New Features

• Initial implementation of the PySERA radiomics pipeline. (#1)
• Established core radiomics feature extraction functionality. (#1)
• Added support for basic file formats. (#1)

v2.1.1

October 22, 2025

Commit: 250a1a0
v2.1.1

Bug Fixes

• Resolved stability issues in batch processing
• Fixed memory management in large dataset handling

v0.0.0

August 13, 2025

New Features

• Standardized radiomics feature extraction per IBSI (Image Biomarker Standardisation Initiative) guidelines. (#3)

Bug Fixes

• Addressed minor bugs in the initial radiomics pipeline implementation. (#3)