A natural language processing pipeline for pairing measurements uniquely across free-text CT reports
| Autor: | Andrea Cowhy, Jeffrey Bozeman, William Trost, Merlijn Sevenster |
|---|---|
| Rok vydání: | 2015 |
| Předmět: |
Radiography
Abdominal Similarity (geometry) Computer science Pipeline (computing) Oncologic measurement Health Informatics computer.software_genre Medical Oncology Task (project management) Machine Learning Semantic similarity Data Mining Humans Ground truth business.industry Natural language processing Computational Biology Reproducibility of Results Expression (mathematics) Random forest Computer Science Applications Radiology Information Systems ROC Curve RECIST Information correlation Pairing Area Under Curve Radiographic Image Interpretation Computer-Assisted Artificial intelligence Radiology report business Radiology Tomography X-Ray Computed computer Algorithms Software |
| Zdroj: | Journal of Biomedical Informatics. 53:36-48 |
| ISSN: | 1532-0464 |
| DOI: | 10.1016/j.jbi.2014.08.015 |
| Popis: | Display Omitted Oncology guidelines standardize response assessment based on measurements.Regular expression-based techniques recognize measurements in reports.We introduced the task of pairing lesion measurements across consecutive reports.A natural language processing pipeline can be construed for the pairing task.A post-processor enforces that each measurement is matched with at most one other. ObjectiveTo standardize and objectivize treatment response assessment in oncology, guidelines have been proposed that are driven by radiological measurements, which are typically communicated in free-text reports defying automated processing. We study through inter-annotator agreement and natural language processing (NLP) algorithm development the task of pairing measurements that quantify the same finding across consecutive radiology reports, such that each measurement is paired with at most one other ("partial uniqueness"). Methods and materialsGround truth is created based on 283 abdomen and 311 chest CT reports of 50 patients each. A pre-processing engine segments reports and extracts measurements. Thirteen features are developed based on volumetric similarity between measurements, semantic similarity between their respective narrative contexts and structural properties of their report positions. A Random Forest classifier (RF) integrates all features. A "mutual best match" (MBM) post-processor ensures partial uniqueness. ResultsIn an end-to-end evaluation, RF has precision 0.841, recall 0.807, F-measure 0.824 and AUC 0.971; with MBM, which performs above chance level (P |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect