2014 Summer Project Week:DRAMMS Slicer
- MGH: Yangming Ou, Randy Gollub, Jayashree Kalpathy-Cramer
- Isomics: Steve Pieper
- UPenn: Christos Davatzikos
- BWH: Andriy Fedorov, Tina Kapur, Ron Kikinis
[DRAMMS] is a general-purpose, fully-automated, deformable image registration algorithm and software package. It was designed for 2D-to-2D and 3D-to-3D cross-subject, longitudinal, multi-modal registration tasks, and has been recently extended for spatial-temporal (4D) registration and population-wise registration. Some typical applications include:
- quantification of 4D tumor changes;
- study of early-life longitudinal neuro-deveolopment;
- maintenance of 4D segmentation consistency;
- atlas construction in a population (infants, neonates, adults, ...);
- multi-atlas based structure segmentation (brain, prostate, ...);
- multi-modal registration (we had some success in histology-MRI registration of the prostate and mouse brain, looking for data for validation);
- registration under a partial loss of correspondences(such as tumor/lesion-bearing images to normal-appearance images).
Several evaluation studies comparing DRAMMS with many state-of-the-art registration algorithms are also being documented for brain [DRAMMS_in_Brain_Registration], cardiac [DRAMMS_in_Cardiac_Registration], breast registration (a journal article accepted pending minor revision) tasks.
The integration of DRAMMS into Slicer is expected to bring more diverse deformable registration options into Slicer, especially an option to deal with registration of both brain and non-brain structures, and to deal with registration under a partial loss of correspondences.
- To integrate the DRAMMS deformable registration software into 3DSlicer.
- Starting from the current stand-alone DRAMMS version (academic- but not commercial-friendly), figure out the necessary interface to integrate it into Slicer.
- [Evaluation of DRAMMS in Longitudinal Breast Registration (a journal article accepted pending minor revision)]