Refine
Document Type
- Article (1)
- Conference Proceeding (1)
Language
- English (2)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- Automatische Sprachanalyse (2) (remove)
Institute
- Fakultät III - Medien, Information und Design (2) (remove)
After kidney transplantation graft rejection must be prevented. Therefore, a multitude of parameters of the patient is observed pre- and postoperatively. To support this process, the Screen Reject research project is developing a data warehouse optimized for kidney rejection diagnostics. In the course of this project it was discovered that important information are only available in form of free texts instead of structured data and can therefore not be processed by standard ETL tools, which is necessary to establish a digital expert system for rejection diagnostics. Due to this reason, data integration has been improved by a combination of methods from natural language processing and methods from image processing. Based on state-of-the-art data warehousing technologies (Microsoft SSIS), a generic data integration tool has been developed. The tool was evaluated by extracting Banff-classification from 218 pathology reports and extracting HLA mismatches from about 1700 PDF files, both written in german language.
Purpose: Radiology reports mostly contain free-text, which makes it challenging to obtain structured data. Natural language processing (NLP) techniques transform free-text reports into machine-readable document vectors that are important for creating reliable, scalable methods for data analysis. The aim of this study is to classify unstructured radiograph reports according to fractures of the distal fibula and to find the best text mining method.
Materials & Methods: We established a novel German language report dataset: a designated search engine was used to identify radiographs of the ankle and the reports were manually labeled according to fractures of the distal fibula. This data was used to establish a machine learning pipeline, which implemented the text representation methods bag-of-words (BOW), term frequency-inverse document frequency (TF-IDF), principal component analysis (PCA), non-negative matrix factorization (NMF), latent Dirichlet allocation (LDA), and document embedding (doc2vec). The extracted document vectors were used to train neural networks (NN), support vector machines (SVM), and logistic regression (LR) to recognize distal fibula fractures. The results were compared via cross-tabulations of the accuracy (acc) and area under the curve (AUC).
Results: In total, 3268 radiograph reports were included, of which 1076 described a fracture of the distal fibula. Comparison of the text representation methods showed that BOW achieved the best results (AUC = 0.98; acc = 0.97), followed by TF-IDF (AUC = 0.97; acc = 0.96), NMF (AUC = 0.93; acc = 0.92), PCA (AUC = 0.92; acc = 0.9), LDA (AUC = 0.91; acc = 0.89) and doc2vec (AUC = 0.9; acc = 0.88). When comparing the different classifiers, NN (AUC = 0,91) proved to be superior to SVM (AUC = 0,87) and LR (AUC = 0,85).
Conclusion: An automated classification of unstructured reports of radiographs of the ankle can reliably detect findings of fractures of the distal fibula. A particularly suitable feature extraction method is the BOW model.
Key Points:
- The aim was to classify unstructured radiograph reports according to distal fibula fractures.
- Our automated classification system can reliably detect fractures of the distal fibula.
- A particularly suitable feature extraction method is the BOW model.