Neuroinformatics

, Volume 1, Issue 2, pp 177–192

Neuroanatomical term generation and comparison between two terminologies

  • Prashanti R. Srinivas
  • Daniel Gusfield
  • Oliver Mason
  • Michael Gertz
  • Michael Hogarth
  • James Stone
  • Edward G. Jones
  • Fredric A. Gorin
Original Article

DOI: 10.1007/s12021-003-0004-z

Cite this article as:
Srinivas, P.R., Gusfield, D., Mason, O. et al. Neuroinform (2003) 1: 177. doi:10.1007/s12021-003-0004-z

Abstract

An approach and software tools are described for identifying and extracting compound terms (CTs), acronyms and their associated contexts from textual material that is associated with neuroanatomical atlases. A set of simple syntactic rules were appended to the output of a commercially available part of speech (POS) tagger (Qtag v 3.01) that extracts CTs and their associated context from the texts of neuroanatomical atlases. This “hybrid” parser appears to be highly sensitive and recognized 96% of the potentially germane neuroanatomical CTs and acronyms present in the cat and primate thalamic atlases.

A comparison of neuroanatomical CTs and acronyms between the cat and primate atlas texts was initially performed using exact-term matching. The implementation of string-matching algorithms significantly improved the identification of relevant terms and acronyms between the two domains. The End Gap Free string matcher identified 98% of CTs and the Needleman Wunsch (NW) string matcher matched 36% of acronyms between the two atlases.

Combining several simple grammatical and lexical rules with the POS tagger (“hybrid parser”) (1) extracted complex neuroanatomical terms and acronyms from selected cat and primate thalamic atlases and (2) and facilitated the semi-automated generation of a highly granular thalamic terminology. The implementation of string-matching algorithms (1) reconciled terminological errors generated by optical character recognition (OCR) software used to generate the neuroanatomical text information and (2) increased the sensitivity of matching neuroanatomical terms and acronyms between the two neuroanatomical domains that were generated by the “hybrid” parser.

Index Entries

Term similarity thalamic atlas neuroanatomical indexing information retrieval string matching statistical parser 

Copyright information

© Humana Press Inc 2003

Authors and Affiliations

  • Prashanti R. Srinivas
    • 1
  • Daniel Gusfield
    • 2
  • Oliver Mason
    • 3
  • Michael Gertz
    • 2
  • Michael Hogarth
    • 4
  • James Stone
    • 1
  • Edward G. Jones
    • 1
  • Fredric A. Gorin
    • 1
    • 5
  1. 1.Center for NeuroscienceUniversity of California at DavisDavis
  2. 2.Department of Computer ScienceUniversity of California at DavisDavis
  3. 3.Department of English, School of HumanitiesThe University of BirminghamBirminghamUK
  4. 4.Departments of Pathology and Internal MedicineUniversity of California at DavisDavis
  5. 5.Department of Neurology, School of MedicineUniversity of California at DavisDavis

Personalised recommendations