brainchop-v4.1.0

In v4.1.0, following fixes has been added :

• Fix brainchop parameters (b3372b8)

• Fix 104 high memory and fast model parameter (e36437a)

• Cleaning code (e.g. 1fc13a0).

brainchop-v4.0.0

Release Notes

Version 4.0.0

We are thrilled to welcome Chris Rorden (@neurolabusc) and Taylor Hanayik (@hanayik) as new contributors to brainchop. Special thanks to them for their main contributions to version 4. (67396d2).

Release supervision: @sergeyplis

Merging and configuration: @Mmasoud1

New Features

• Integration of the NiiVue viewer for loading and visualization. This allows drag and drop support for voxel formats including NIfTI, NRRD, MRtrix MIF, AFNI HEAD/BRIK, MGH/MGZ, ITK MHD, ECAT7 as well as limited DICOM support.

• Allows models to be run either on the main thread (as before) or a web worker

• Allows classifications to be edited by drawing to append or erase voxels.

• Allows entires scenes (images, classification results, contrast, crosshair position, drawings) to be saved as NiiVue documents that can be opened anywhere.

Enhancements

• Refactoring for improved performance
• Optimized pre and post-processing steps
• Removes the Python dependencies (pyodide) for the conform function, and replaces this function with native JavaScript.
• Reduces static memory usage.
• Dramatically reduces peak heap memory usage (for low memory 104 region model the reduction is x173).

brainchop-v3.4.0

In v3.4.0, following fixes has been added :

• 2x speedup for sequential low memory models (2e555f4)

brainchop-v3.3.0

In v3.3.0, following fixes has been added :

• Accelerate connected components post processing (89a01c9)

brainchop-v3.2.1

In v3.2.1, following fixes has been added :

• Fix buffer leakage (03d4019)

• Fix Normalization issue #22 (b17d9c6)

brainchop-v3.2.0

In v3.2.0, following features and fixes has been added :

• Add new failsafe subcortical model use preModel for cropping (0fe66a3)

• Activate production mode (0fe66a3)

• Turn on enableCrop for light model (0fe66a3)

brainchop-v3.1.0

In v3.1.0, following fixes has been added :

Fixes

• Fixed 3D connected components bug with textures f16 bits (eff81f3)

• Turn off preModel cropping for Sub-cortical models (71a1a49)

brainchop-v3.0.0

In v3.0.0, we added the following features :

Features:

• Add more robust models (2b061eb)

• Add Pytorch to TFJS conversion example (7840eec)

brainchop-v2.4.0

In v2.4.0, we add more options :

Features:

• Add 18 classes model (f0d3153 ).

brainchop-v2.3.0

In v2.3.0, we add more options and addressed the following fixes :

Fixes:

• Fix memory (baa236b)

• Add sequential convolution option for low memory resources (1c4f657)