Learning Interclass Relations for Intravenous Contrast Phase Classification in CT

Raouf Muhamedrahimov, Amir Bar, Ayelet Akselrod-Ballin
Proceedings of the Fourth Conference on Medical Imaging with Deep Learning, PMLR 143:507-519, 2021.

Abstract

In classification, categories are typically treated as independent of one-another. In many problems, however, this neglects the natural relations that exist between categories, which are often dictated by an underlying biological or physical process. In this work, we propose novel formulations of the classification problem, aimed at reintroducing class relations into the training process. We demonstrate the benefit of these approaches for the classification of intravenous contrast enhancement phase in CT images, an important task in the medical imaging domain. First, we propose manual ways reintroduce knowledge about problem-specific interclass relations into the training process. Second, we propose a general approach to jointly learn categorical label representations that can implicitly encode natural interclass relations, alleviating the need for strong prior assumptions or knowledge. We show that these improvements are most significant for smaller training sets, typical in the medical imaging domain where access to large amounts of labelled data is often not trivial.

Cite this Paper

BibTeX
@InProceedings{pmlr-v143-muhamedrahimov21a, title = {Learning Interclass Relations for Intravenous Contrast Phase Classification in {CT}}, author = {Muhamedrahimov, Raouf and Bar, Amir and Akselrod-Ballin, Ayelet}, booktitle = {Proceedings of the Fourth Conference on Medical Imaging with Deep Learning}, pages = {507--519}, year = {2021}, editor = {Heinrich, Mattias and Dou, Qi and de Bruijne, Marleen and Lellmann, Jan and Schläfer, Alexander and Ernst, Floris}, volume = {143}, series = {Proceedings of Machine Learning Research}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v143/muhamedrahimov21a/muhamedrahimov21a.pdf}, url = {https://proceedings.mlr.press/v143/muhamedrahimov21a.html}, abstract = {In classification, categories are typically treated as independent of one-another. In many problems, however, this neglects the natural relations that exist between categories, which are often dictated by an underlying biological or physical process. In this work, we propose novel formulations of the classification problem, aimed at reintroducing class relations into the training process. We demonstrate the benefit of these approaches for the classification of intravenous contrast enhancement phase in CT images, an important task in the medical imaging domain. First, we propose manual ways reintroduce knowledge about problem-specific interclass relations into the training process. Second, we propose a general approach to jointly learn categorical label representations that can implicitly encode natural interclass relations, alleviating the need for strong prior assumptions or knowledge. We show that these improvements are most significant for smaller training sets, typical in the medical imaging domain where access to large amounts of labelled data is often not trivial.} }
Endnote
%0 Conference Paper %T Learning Interclass Relations for Intravenous Contrast Phase Classification in CT %A Raouf Muhamedrahimov %A Amir Bar %A Ayelet Akselrod-Ballin %B Proceedings of the Fourth Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2021 %E Mattias Heinrich %E Qi Dou %E Marleen de Bruijne %E Jan Lellmann %E Alexander Schläfer %E Floris Ernst %F pmlr-v143-muhamedrahimov21a %I PMLR %P 507--519 %U https://proceedings.mlr.press/v143/muhamedrahimov21a.html %V 143 %X In classification, categories are typically treated as independent of one-another. In many problems, however, this neglects the natural relations that exist between categories, which are often dictated by an underlying biological or physical process. In this work, we propose novel formulations of the classification problem, aimed at reintroducing class relations into the training process. We demonstrate the benefit of these approaches for the classification of intravenous contrast enhancement phase in CT images, an important task in the medical imaging domain. First, we propose manual ways reintroduce knowledge about problem-specific interclass relations into the training process. Second, we propose a general approach to jointly learn categorical label representations that can implicitly encode natural interclass relations, alleviating the need for strong prior assumptions or knowledge. We show that these improvements are most significant for smaller training sets, typical in the medical imaging domain where access to large amounts of labelled data is often not trivial.
APA
Muhamedrahimov, R., Bar, A. & Akselrod-Ballin, A.. (2021). Learning Interclass Relations for Intravenous Contrast Phase Classification in CT. Proceedings of the Fourth Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 143:507-519 Available from https://proceedings.mlr.press/v143/muhamedrahimov21a.html.

Related Material