Name
Abstract | ||
|---|---|---|
Radiology reports generally consist of narrative text. It has been envisioned that structured medical content can be leveraged to clinical applications. Text-mining techniques can be utilized to realize this vision. We created a pipeline for automatic sentence classification of narrative breast cancer radiology reports. A corpus of 353 reports and 8166 sentences was annotated with seven sentence classes related to laterality, modality and recommendation. Sentences have been represented by four types of feature sets, characterizing various levels of linguistic complexity and domain knowledge. We conducted an evaluation to find the optimal combination of features and the optimal classification paradigm. The classification accuracy ranges between 92 and 98% for the different classes. |
Year | DOI | Venue |
|---|---|---|
2012 | 10.1109/CBMS.2012.6266374 | Computer-Based Medical Systems |
Keywords | Field | DocType |
cancer,data mining,medical computing,radiology,text analysis,automatic sentence classification,domain knowledge,free-text breast cancer radiology reports,linguistic complexity,narrative breast cancer radiology,sentence-based classification,text-mining techniques | Data mining,Breast cancer,Domain knowledge,Computer science,Narrative,Linguistic sequence complexity,Natural language processing,Artificial intelligence,Radiology,Sentence | Conference |
ISSN | ISBN | Citations |
1063-7125 | 978-1-4673-2049-8 | 0 |
PageRank | References | Authors |
0.34 | 7 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Aisan Maghsoodi | 1 | 0 | 0.68 |
| Merlijn Sevenster | 2 | 98 | 13.33 |
| Johannes Scholtes | 3 | 0 | 0.34 |
| Georgi I. Nalbantov | 4 | 0 | 0.34 |