NETASSETS

The massive collections of radiology links
on portal sites such as AuntMinnie
.com (http://www.auntminnie.com)
and RadiologyEducation.com (http:
//radiologyeducation.com) are a testament
to the enthusiasm with which
radiologists have embraced the Internet
as an “e-learning” platform (1).
The ideal is clear: fast relevant answers
from reliable sources on beautifully
illustrated Web sites that are
easy to navigate. However, when one
is hunting for a quick answer during a
hectic day or a lonely overnight shift,
the deficiencies of the Web for pointof-
care decision support become glaring.
Although general search engines
have improved, one still often encounters
too much irrelevant information
for the radiologist, including advertisements,
patient-oriented information,
and outdated or unreliable material.
Luckily, we have now entered the
age of the “radiology-centric” search.

Yottalook: “Y’oughta Look”
Fusing proprietary technologies with a
highly customized Google search engine,
the developers of Yottalook (http:
//www.yottalook.com) at iVirtuoso (Washington,
DC) have optimized the radiology
search experience by incorporating
a semantic, or concept-based, approach
(2). A user’s query is analyzed with
reference to various medical ontologic
taxonomies, including an enhanced version
of Radlex, the radiology lexicon developed
with sponsorship from the Radiological
Society of North America
(RSNA). The search engine can consequently
expand the query to all possible
relevant terms and index the results in a
way that enables intelligent filtering. For
example, if one were to enter “PE” into
the Yottalook search box, the site understands
that this could refer to pulmonary
embolism, preeclampsia, or pericardial
effusion and allows one to
choose appropriately. The semantic
search engine cleverly recognizes synonyms
such as gallstones and cholelithiasis
and utilizes hierarchic relation between medical concepts to return the most comprehensive and relevant search results. For example, when one enters the disease category “phakomatosis” in Yottalook, the related terms tuberous sclerosis, neurofibromatosis, and ataxia-telangiectasia are offered as additional search options

Radiology-specific filters are another key strategy by which Yottalook enhances the search experience. By restricting the scope to such relevant categories as teaching files, journal articles, continuing medical education materials or to specific modalities such as computed tomography (CT) or magnetic resonance (MR) imaging, one can much more quickly get to the right answer (Fig 1). The “Anatomy” filter will preferentially return search queries from dedicated online atlases such as Gray's Anatomy and journal articles likely to have high-yield anatomic content, such as one example titled “The Cystic Duct: Normal Anatomy and Disease Processes” (3) from an online education exhibit in RadioGraphics. Combining the various filters is possible in order to make even more specific searches practical. For example, if one were trying to update one's practice's MR imaging protocol for carotid dissection, a Yottalook search could be performed with the “MRI” and “Protocol” filters chosen. Top results would include the links to various institutions' Web sites that have made their protocols freely available online (eg, Massachusetts General Hospital Neuroradiology—MRI Protocols, http://www.mghneuroradiology.org/NewFiles/mrip.html). If one were to choose the filter for a specific hardware manufacturer, only journal articles that listed the appropriate equipment in their methods section would be listed.