@conference {cGorrizc, title = {Assessing Knee OA Severity with CNN attention-based end-to-end architectures}, booktitle = {International Conference on Medical Imaging with Deep Learning (MIDL) 2019}, year = {2019}, month = {02/2019}, publisher = {JMLR}, organization = {JMLR}, address = {London, United Kingdom}, abstract = {

This work proposes a novel end-to-end convolutional neural network (CNN) architecture to\ automatically quantify the severity of knee osteoarthritis (OA) using X-Ray images, which\ incorporates trainable attention modules acting as unsupervised fine-grained detectors of\ the region of interest (ROI). The proposed attention modules can be applied at different\ levels and scales across any CNN pipeline helping the network to learn relevant attention\ patterns over the most informative parts of the image at different resolutions. We test\ the proposed attention mechanism on existing state-of-the-art CNN architectures as our\ base models, achieving promising results on the benchmark knee OA datasets from the\ osteoarthritis initiative (OAI) and multicenter osteoarthritis study (MOST). All the codes\ from our experiments will be publicly available on the github repository: https://github.com/marc-gorriz/KneeOA-CNNAttention

}, url = {http://proceedings.mlr.press/v102/gorriz19a.html}, author = {G{\'o}rriz, Marc and Antony, Joseph and McGuinness, Kevin and Xavier Gir{\'o}-i-Nieto and O{\textquoteright}Connor, N.} }