NA-MIC National Alliance for Medical Image Computing Slicer3 Update
National Alliance for Medical Image Computing When last we saw you… Steve took a beating on Slicer 2.6 Steve presented Slicer3 Execution Model Calmer heads prevailed “There must be something wrong with this…” - Polina
National Alliance for Medical Image Computing Review of base architecture Logic MRML Nodes GUI WidgetsRenderers Edit Observe Edit
National Alliance for Medical Image Computing Gratuitous screenshots
National Alliance for Medical Image Computing Status MRML Slices 3D View Data GUI Volumes Models Transforms Application GUI Fiducials Execution model Colors Model maker Editor Slicer2 modules Build/Package Testing Bug tracking Grid system System
National Alliance for Medical Image Computing
June 2006
National Alliance for Medical Image Computing September 2006 GUI Thread Processing Thread
National Alliance for Medical Image Computing AHM 2007 Many Core 1 presentations with 2D plotting –Key component of clinical analysis Slicer 3 had no 2D plotting –VTK very lacking in 2D plotting
National Alliance for Medical Image Computing February 2007 Python, SciPy, PyLab, Matplotlib
National Alliance for Medical Image Computing March 2007 Large scale experiment control Scenario #1 –Use Slicer3 interactively to select parameters –Run across N datasets Scenario #2 –Use Slicer3 interactively to select some parameters –Design of experiments across remaining parameters Assumptions –Single computer, cluster, grid –No direct access to compute nodes –Staged data –Staged programs –Experiment scheduling –Job submission –Experiment control –Job execution robustness –Platform Options Loopy launcher BatchMake Grid tools
National Alliance for Medical Image Computing Current Slicer3 activities Performance –Resampling, observer madness, intervention Visualization breadth –Cine, lightbox, tensors, plots, CSV, tables Interaction breadth –Tracked cursors, editors, measurements, WWidgets, SWidgets Execution model breadth –Scenes, transforms, coordinate frames, CSV, MRMLImageIO, … MRML breadth –Model hierarchies, grouping, CSV, tables Modules –IGT
National Alliance for Medical Image Computing Three integration strategies Base Slicer3 modules (interactive modules) –Tight integration with GUI and Visualization subsystems –Need MRML Node, Logic class, GUI class Command line modules (execution model modules) –XML descriptions of executable and shared library modules Wizard based modules –Sophisticated wizard to guide user to setting parameters –Need many MRML, Logic, and GUI classes –EM Segmenter
National Alliance for Medical Image Computing Interactive module
National Alliance for Medical Image Computing Command line module
National Alliance for Medical Image Computing Wizard based module