Knee lateral view radiographs were extracted from The MulticenterOsteoarthritis Study (MOST) (n = 18,436 knees) Patellar region-of-interest(ROI) was first automatically detected, and subsequently, end-to-end deepconvolutional neural networks (CNNs) were trained and validated to detect the status of patellofemoral OA . Results: The CNN model that used only image data significantly improved the prediction of PFOA status . AUC and AP for thereference model including age, sex, body mass index (BMI), the total WesternOntario and McMaster Universities Arthritis Index (WOMAC) score were 0.806 and 0.478,respectively, respectively . We present the first machine learning based automatic PFOAdetection method. Furthermore, our deep learning based model trained on patellaregion from knee lateral viewradiographs performs better at predicting PFO athan models based on patient characteristics and clinical assessments, such as patient characteristics .

Author(s) : Neslihan Bayramoglu, Miika T. Nieminen, Simo Saarakkala

Links : PDF - Abstract

Code :

https://github.com/oktantod/RoboND-DeepLearning-Project


Coursera

Keywords : lateral - knee - learning - model - based -

Leave a Reply

Your email address will not be published. Required fields are marked *