Preprocessing for extracting information from handwritten medical record to add XML tags

Abstract

The purpose of this paper is to propose preprocessing procedures from handwritten medical records to translate contents of the records into formatted data such as XML data. From the viewpoint, we performed an experiment, which that physicians freely described patients' problems in order to reveal the characteristics in the handwritten record. Based on the handwritten data, we obtained five characteristics and considered eight preprocessing steps. Then, we confirmed that our proposed preprocessing worked well from a computational experiment although not all sub-procedures are effective due to lack of a large amount of corpora.