Curriculum learning for improved femur fracture classification: scheduling data with prior knowledge and uncertainty

Mostra el registre complet Registre parcial de l'ítem

• dc.contributor.author Jiménez Sánchez, Amelia
• dc.contributor.author Mateus, Diana
• dc.contributor.author Kirchhoff, Sonja
• dc.contributor.author Kirchhoff, Chlodwig
• dc.contributor.author Biberthaler, Peter
• dc.contributor.author Navab, Nassir
• dc.contributor.author González Ballester, Miguel Ángel, 1973-
• dc.contributor.author Piella Fenoy, Gemma
• dc.date.accessioned 2023-03-09T07:24:59Z
• dc.date.issued 2022
• dc.description.abstract An adequate classification of proximal femur fractures from X-ray images is crucial for the treatment choice and the patients’ clinical outcome. We rely on the commonly used AO system, which describes a hierarchical knowledge tree classifying the images into types and subtypes according to the fracture’s location and complexity. In this paper, we propose a method for the automatic classification of proximal femur fractures into 3 and 7 AO classes based on a Convolutional Neural Network (CNN). As it is known, CNNs need large and representative datasets with reliable labels, which are hard to collect for the application at hand. In this paper, we design a curriculum learning (CL) approach that improves over the basic CNNs performance under such conditions. Our novel formulation reunites three curriculum strategies: individually weighting training samples, reordering the training set, and sampling subsets of data. The core of these strategies is a scoring function ranking the training samples. We define two novel scoring functions: one from domain-specific prior knowledge and an original self-paced uncertainty score. We perform experiments on a clinical dataset of proximal femur radiographs. The curriculum improves proximal femur fracture classification up to the performance of experienced trauma surgeons. The best curriculum method reorders the training set based on prior knowledge resulting into a classification improvement of 15%. Using the publicly available MNIST dataset, we further discuss and demonstrate the benefits of our unified CL formulation for three controlled and challenging digit recognition scenarios: with limited amounts of data, under class-imbalance, and in the presence of label noise. The code of our work is available at: https://github.com/ameliajimenez/curriculum-learning-prior-uncertainty.
• dc.description.sponsorship This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 713673 and by the Spanish Ministry of Economy [MDM-2015-0502]. A. Jiménez-Sánchez has received financial support through the “la Caixa” Foundation (ID Q5850017D), fellowship code: LCF/BQ/IN17/11620013. D. Mateus has received funding from Nantes Métropole and the European Regional Development, Pays de la Loire, under the Connect Talent scheme.
• dc.format.mimetype application/pdf
• dc.identifier.citation Jiménez-Sánchez A, Mateus D, Kirchhoff S, Kirchhoff C, Biberthaler P, Navab N, González Ballester MA, Piella G. Curriculum learning for improved femur fracture classification: scheduling data with prior knowledge and uncertainty. Med Image Anal. 2022;75:102273. DOI: 10.1016/j.media.2021.102273
• dc.identifier.doi http://dx.doi.org/10.1016/j.media.2021.102273
• dc.identifier.issn 1361-8415
• dc.identifier.uri http://hdl.handle.net/10230/56119
• dc.language.iso eng
• dc.publisher Elsevier
• dc.relation.ispartof Medical Image Analysis. 2022;75:102273.
• dc.relation.isreferencedby https://github.com/ameliajimenez/curriculum-learning-prior-uncertainty
• dc.relation.isreferencedby https://ars-els-cdn-com.sare.upf.edu/content/image/1-s2.0-S1361841521003182-mmc1.pdf
• dc.relation.projectID info:eu-repo/grantAgreement/EC/H2020/713673
• dc.relation.projectID info:eu-repo/grantAgreement/ES/1PE/MDM-2015-0502
• dc.rights © Elsevier http://dx.doi.org/10.1016/j.media.2021.102273
• dc.rights.accessRights info:eu-repo/semantics/openAccess
• dc.subject.keyword Curriculum learning
• dc.subject.keyword Self-paced learning
• dc.subject.keyword Data scheduler
• dc.subject.keyword Bone fracture
• dc.subject.keyword X-ray
• dc.subject.keyword Multi-class classification
• dc.subject.keyword Limited data
• dc.subject.keyword Class-imbalance
• dc.subject.keyword Noisy labels
• dc.title Curriculum learning for improved femur fracture classification: scheduling data with prior knowledge and uncertainty
• dc.type info:eu-repo/semantics/article
• dc.type.version info:eu-repo/semantics/acceptedVersion