In the context of cancer diagnosis, treatment monitoring, and active surveillance, there's a need to encode the anatomical location of lesions found via prostate MRI on workstations that perform image segmentation and quantification and record their results as DICOM Structured Reports and DICOM Segmentation objects, according to Clunie, owner of PixelMed Publishing. Although version 2 of the Prostate Imaging Reporting and Data System (PI-RADS) has standardized a description of the anatomical sectors thought to be important, it only includes a visual description and abbreviated names; it also doesn't provide a standard machine-readable code set, he said.
"For interoperability of applications creating and consuming reports and clinical trial case-report forms, one needs a standard set of agreed codes," he said.
Even when coded, one also needs to understand the structural relationship of these coded sectors to draw conclusions, such as recognizing that a very precisely defined sector is "part of" a more coarsely defined sector, and determining if a lesion is in the peripheral zone, regardless of size or more specific location, he said.
Fortunately, "such a 'partology,' as well as spatial relationships (e.g., 'anterior to') can be supported by existing ontologies," Clunie said. The team requested enhancements to standard schemes such as Systematized Nomenclature of Medicine - Clinical Terms (SNOMED CT), Foundational Model of Anatomy (FMA), and the U.S. National Cancer Institute (NCI) Thesaurus. The researchers also obtained permission to use the new SNOMED CT content in the DICOM standard.
"As a result, DICOM Structured Reports and Segmentation objects will now be able to encode standard anatomical location information for prostate multiparametric MRI according to the PI-RADS v2 recommendations in a manner that is interoperable with other systems like [electronic health records]," Clunie told AuntMinnie.com.
