Automatic measurement of hallux valgus angle
Description
This is the dataset for the article, 'Automatic estimation of hallux valgus angle using deep neural network with axis-based annotation '. (Takeda et.al., Skeletal Radiology 2024) https://link.springer.com/article/10.1007/s00256-024-04618-2 Cite this paper when you use this newral network model for research purposes. This dataset contains the developed deep neural network model, automatic HVA measurement application utilizing the neural network, and statistical analysis code used for the validation cohort to evaluate the accuracy of the neural network model. This dataset does not include the patient's radiographs because of the privacy concerns.
Files
Steps to reproduce
The developped neural network is 'trainedNet.mat'. The developped neural network to measure HVA automatically is contained in the folder of 'Application_experienceAI_wo_coding/dist.zip' as ONNX file. If you want to try the neural network with out coding, download and unzip 'dist.zip' in the folder of 'Application_experienceAI_wo_coding'. Execute the main.exe, and the application will start. You should prepare foot radiographs as a dicom file to measure HVA. 'Application_for_manual_measurement' folder contains the MATLAB application used for the annotation data used in the training of neural network. This application was also used for the manual measurement by three orthopaedic surgeons to compare between manual and automatic measurements. You have to install MATLAB, to use this application. 'Preprocessing' contains the code for the image preprocessing before the training of the neural network. 'Validation cohort' contains the code for the statistical analysis and visualization of validation. See 'Comments.txt' in each folder, which provide a brief description of each file contained within that folder.