Abstract Archives of the RSNA, 2012
LL-INE1222-SUB
DICOM SR, PS, RT, Segmentation and Proprietary Format Result Review System for Clinical, Research and Clinical Trial Imaging
Education Exhibits
Presented on November 25, 2012
Presented as part of LL-INE-SU: Informatics Lunch Hour CME Exhibits
Selected for RadioGraphics
David A. Clunie MBBS, Presenter: Officer, CoreLab Partners, Inc
Owner, PixelMed Publishing LLC
Research support, Siemens AG
Advisory Board, CoreLab Partners, Inc
Use-cases for distribution of results for those without access to the authoring tool include referring physicians, external reviewers, public access, central review of site reads, review by regulators, & data mining. Either a standard format, or a review tool that is agnostic to the format is needed. A single universal format for results from all systems remains elusive.
A general-purpose review tool, the Result Annotation Tabulator & Basic Evaluation Review Tool (RATBERT), is not constrained to a specific format. It emphasizes the table as a means of presentation, & converts disparate input formats into tabular form.
Advantages of tables include ease of:
-navigation
-sorting & filtering by columns
-communication with statistical tools
-conversion to Study Data Tabulation Model
export for publication in scientific journals
Additional design goals include
-remote accessibility
-hyperlinking of selections to image displays to show ROIs & outliers
A generalized remotely deployable review tool tolerant of different formats is feasible & solves the problem of reviewing & sharing results information.
It is feasible to accept lack of a single common format & move on to making use of the data without obsessing over enforcement of the use of a single format.
Testing included DICOM PS, SR, RT & SEG objects, & non-DICOM data sets from PACS, clinical trial software, LIDC & TCGA projects in XML (including AIM).Java was used for multiple platforms, remote deployment, built-in data structures & widgets for tables, XML & DICOM toolkits for XSL-T, SR, 2D & 3D coordinate & rasterized segmentation support in the open-source PixelMed tools.Declarative & procedural approaches were tried, with XML representation of DICOM SR as a common intermediate format, XSL-T stylesheets for XML to SR conversion prior to procedural table extraction.Patterns translated include:- simple concept-value pairs (measurements & question/answer pairs)- concepts with explicit or silbling modifiers- multi-column siblings- identifiers correlating different sub-trees- parent context to expand as repeated values across rows- single versus multiple input files
Clunie, D,
DICOM SR, PS, RT, Segmentation and Proprietary Format Result Review System for Clinical, Research and Clinical Trial Imaging. Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL.
http://archive.rsna.org/2012/12037058.html