Text forms a significant part of the medical record. Being able to map text to concepts in a controlled medical vocabulary can be useful for teaching, patient care, and research. Natural Language Processing systems have figured prominently in attempts to process medical text, particularly radiology dictations. Chart Engine, an existing system for mapping radiology text to Unified Medical Language System Metathesaurus concepts, currently uses barrier words to select candidate noun phrases for mapping. In this thesis, Chart Engine was equipped with a syntactic parser as a preliminary processing step that would then supply candidate noun phrases instead of the phrases supplied by the list of barrier words. Several different types of noun phrases were tested.