Sometimes these double as a general toolbox, like voxelmorph.
- VoxelMorph
We introduced learning-based image registration/alignment, including VoxelMorph, SynthMorph, AtlasMorph, HyperMorph, etc, and general utilities. - Synth*
In our Synth* line of work we develop a procedural data generation engine to produce synthetic training data that yields methods generalizing reliably across scanners, contrasts, populations, and even modalities. This includes SynthMorph, SynthSeg, SynthStrip, BabySeg, Anatomix - Automatic segmentation of new data at scale
We've spearheaded methods for automatic segmentation of new domains/images/regions in new datasets. These include ScribblePrompt
Interactive medical image segmentation with bounding boxes, clicks and scribbles, UniverSeg -- Universal Medical Image Segmentation from new examples without re-training, - MultiverSeg -- Scalable interactive segmentation of biomedical imaging datasets with in-context guidance, tyche -- multi-rater segmentation at scale, etc - Topofit
Learning-basedfitting of surfaces to cerebral cortex in brain MRI - Thunderpack
Blazingly fast multi-modal data format - Hyperlight
Hypernetworks in Pytorch made easy
Libraries with tools offering support to the sort of research we do
- voxel
torch-focused general purpose medical volume processing - neurite
medical image analysis, mostly intensorflow/kerasfor now.
Related more experimental library: neurite-sandbox - voxynth
Synthesizing voxel data, based on the generative modeling in SynthSeg, SynthMorph, SynthStrip, etc
- teramedical
large-scale medical data processing library, with dataset-specific processing [in conjunction with Sabuncu Lab at Cornell] - daldata internal library for quickly loading/handling data for torch-based ML projects
- surfa
General purpose medical image handling and volume processing (general/non-DL)
- matlib
extensivematlabgeneral purpose libraries
- resources lab resources (internal datasets, etc)
- idealab light jupyters on ideas and tutorials
- dalcalab_resources
[name TBD?] internal easy-access resources, includingdataandmodelsupporting utilities, for the dalcalab