Keywords: radiographs, quality assessment, anatomical features, deep learning
Abstract: The quality of radiographs is of major importance for diagnosis and treatment planning. While most research regarding automated radiograph quality assessment uses technical features such as noise or contrast, we propose to use anatomical structures as more appropriate features. We show that based on such anatomical features, a modular deep-learning framework can serve as a quality control mechanism for the diagnostic quality of ankle radiographs. For evaluation, a dataset consisting of 950 ankle radiographs was collected and their quality was labeled by radiologists. We obtain an average accuracy of 94.1%, which is better than the expert radiologists are on average.
Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.
Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
Paper Type: both
Primary Subject Area: Application: Radiology
Secondary Subject Area: Detection and Diagnosis
Source Latex: zip