CTSC:CHBresources:CRL

Back to CHB Imaging Resources

Mission

The Computational Radiology Laboratory (CRL) is a computer science oriented laboratory in the Department of Radiology at Children’s Hospital. Faculty have appointments at Harvard Medical School and Massachusetts Institute of Technology, and work closely with colleagues at Brigham and Women’s Hospital and Massachusetts General Hospital.

The CRL was founded in 2001 with the mission of improving our understanding of the structure and function of the brain and other organs of the human body, in order to improve our capacity to diagnose and treat disease. Members of the CRL achieve this by developing novel technologies and computational modeling strategies for understanding and interpreting radiological images.

Resources

• The CRL has dedicated file servers with a combined total of over 12 terabytes of online storage, providing rapid access to data and a capacity to simultaneously analyze large databases of imaging data. The local network of the radiology department at Children’s Hospital is connected by fiber optic links operating at 1 Gigabit per second, and are all connected to Harvard Medical School, which is a gateway to the Harvard University network and the internet.

• The primary workstation configuration on each user desktop is a symmetric multi-processor machine running linux with dual 3.0 gigahertz Intel Xeon CPUs or a dual core Pentium D CPU, 4 gigabytes of RAM, 400 GB of local storage and a high end hardware accelerated 3D graphics card. This provides ample local compute capacity for the analysis and visualization of three and four-dimensional imaging data. The primary compute servers (2) have two dual 3.0 gigahertz Intel Xeon 5100 CPUs with EM64T (total of 4 CPUs per machine), 32 gigabytes of RAM and 1.5 terabytes of local storage, and the CRL has recently acquired a high end symmetric multiprocessor machine with two quad core Intel Xeon X5535 CPUs (total of 8 CPU cores) with EM64T and 64 gigabytes of RAM and 2.0 terabytes of local storage.

• The software environment for developers includes highly optimizing C, C++ and Fortran compilers from Intel, optimized linear algebra and FFT libraries (LAPACK, BLAS, FFTPACK etc.) and parallel computing libraries (MPICH, PETSc). We routinely develop software using the open-source visualization toolkit VTK and the Insight toolkit ITK for segmentation and registration algorithms. We have Qt toolkit licenses with commercial support for GUI development to enable end users to make use of the algorithms we develop.

• The user environment includes access to locally developed software packages, including CRkit (Computational Radiology Kit) for automatic and interactive segmentation, registration and visualization, as well as our older environment called IMAGER used locally and nationally, and standard software packages from other groups commonly used in medical image analysis (SPM, FSL, 3Dslicer). Standard scientific packages such as Matlab are also installed.

Services

Collaboration and consulting for the design of imaging and imaging analysis experiments