Pattanasak Mongkolwat PhD, Presenter: Nothing to Disclose

Vladimir Kleper, Abstract Co-Author: Nothing to Disclose

Skip Talbot BS, Abstract Co-Author: Nothing to Disclose

Norman Young MSc, Abstract Co-Author: President, ClearCanvas Incorporated CEO, ClearCanvas Incorporated

David A. Clunie MBBS, Abstract Co-Author: Owner, PixelMed Publishing LLC

For quantitative research and clinical practice, manually authored and automatically generated Regions of Interest (ROIs) may be encoded either as contours or arrays of voxels. The Free Open Source Software AIM-enabled ClearCanvas Workstation creates and displays AIM annotations as DICOM SR or XML. Only contours are stored in AIM annotations. The DICOM Segmentation (SEG) object encodes ROIs as arrays of voxels, and SEG objects may be referenced from AIM annotations (via UIDs). Created SEGs can be consumed by SEG enabled workstations.

SEG capabilities used are sufficient to encode manually created ROIs of unconnected regions on single or successive slices, to make 3D tumor measurements. A shared label is used to identify user-drawn polygonal ROIs to be converted to SEGs. Users select anatomy, category and type properties from an extensive DICOM coded value set, which may be modified via configuration files. A configurable color for each user is recorded. Conformance to the DICOM Segmentation Storage SOP Class was verified mechanically. Creation and display of SEG objects on enhanced multi-frame and series of legacy single frame images is supported and was tested on CT and MR. The workstation loads and superimposes SEGs on referenced images or frames. Binary and thresholded fractional probability or occupancy segments, multiple segments and multiple slices per segment, and sub-regions coplanar with images are supported. Opacity of rendered segments can be adjusted. Correct superimposition and spatial location of sub-regions was verified with synthetic test objects. Computed volume, mean and standard deviation are shown. Statistics of predictable size on imported synthetic test SEG objects were successfully compared with known values.

AIM and SEG offer a complete persistence mechanism for 3D ROIs for tumors, serialized as DICOM and compatible with PACS archives. Broader adoption of DICOM SEG in PACS viewers and research tools (3DSlicer, ePad) will also allow interoperable visualization.

The workstation allows annotation of imaging features found on images. The new segmentation capabilities make AIM a more powerful tool for clinical and research use.

http://abstract.rsna.org/uploads/2014/14001381/14001381_ri8j.jpg

Mongkolwat, P, Kleper, V, Talbot, S, Young, N, Clunie, D, Adding DICOM Segmentation Capability to the National Cancer Informatics Program (NCIP) Annotation and Image Markup (AIM) Enabled Imaging Workstation for Imaging Research.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14001381.html