Learning normal asymmetry representations for homologous brain structures

Although normal homologous brain structures are approximately symmetrical by definition, they also have shape differences due to e.g. natural ageing. On the other hand, neurodegenerative conditions induce their own changes in this asymmetry, making them more pronounced or altering their location....

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Bendersky, Mariana, Iarussi, Emmanuel, Deangeli, Duilio, Princich, Juan Pablo, Larrabide, Ignacio, Orlando, José Ignacio
Formato: info:eu-repo/semantics/preprint submittedVersion
Lenguaje:Inglés
Publicado: Universidad Torcuato Di Tella 2023
Materias:
Normal Asymmetry
Brain MRI
Anomaly detection
Machine Learning
Acceso en línea:https://repositorio.utdt.edu/handle/20.500.13098/12140
Aporte de:
Repositorio Digital Universidad Torcuato Di Tella de Universidad Torcuato Di Tella
id I57-R163-20.500.13098-12140
record_format dspace
spelling I57-R163-20.500.13098-121402023-11-17T07:00:20Z Learning normal asymmetry representations for homologous brain structures Bendersky, Mariana Iarussi, Emmanuel Deangeli, Duilio Princich, Juan Pablo Larrabide, Ignacio Orlando, José Ignacio Normal Asymmetry Brain MRI Anomaly detection Machine Learning Although normal homologous brain structures are approximately symmetrical by definition, they also have shape differences due to e.g. natural ageing. On the other hand, neurodegenerative conditions induce their own changes in this asymmetry, making them more pronounced or altering their location. Identifying when these alterations are due to a pathological deterioration is still challenging. Current clinical tools rely either on subjective evaluations, basic volume measurements or disease-specific deep learning models. This paper introduces a novel method to learn normal asymmetry patterns in homologous brain structures based on anomaly detection and representation learning. Our framework uses a Siamese architecture to map 3D segmentations of left and right hemispherical sides of a brain structure to a normal asymmetry embedding space, learned using a support vector data description objective. Being trained using healthy samples only, it can quantify deviations-from-normal-asymmetry patterns in unseen samples by measuring the distance of their embeddings to the center of the learned normal space. We demonstrate in public and in-house sets that our method can accurately characterize normal asymmetries and detect pathological alterations due to Alzheimer’s disease and hippocampal sclerosis, even though no diseased cases were accessed for training. Our source code is available at https://github.com/duiliod/DeepNORHA 2023-11-16T17:41:06Z 2023-11-16T17:41:06Z 2023 info:eu-repo/semantics/preprint info:eu-repo/semantics/submittedVersion https://repositorio.utdt.edu/handle/20.500.13098/12140 eng info:eu-repo/semantics/openAccess https://creativecommons.org/licenses/by-sa/2.5/ar/ 10 p. application/pdf application/pdf Universidad Torcuato Di Tella
institution Universidad Torcuato Di Tella
institution_str I-57
repository_str R-163
collection Repositorio Digital Universidad Torcuato Di Tella
language Inglés
orig_language_str_mv eng
topic Normal Asymmetry
Brain MRI
Anomaly detection
Machine Learning
spellingShingle Normal Asymmetry
Brain MRI
Anomaly detection
Machine Learning
Bendersky, Mariana
Iarussi, Emmanuel
Deangeli, Duilio
Princich, Juan Pablo
Larrabide, Ignacio
Orlando, José Ignacio
Learning normal asymmetry representations for homologous brain structures
topic_facet Normal Asymmetry
Brain MRI
Anomaly detection
Machine Learning
description Although normal homologous brain structures are approximately symmetrical by definition, they also have shape differences due to e.g. natural ageing. On the other hand, neurodegenerative conditions induce their own changes in this asymmetry, making them more pronounced or altering their location. Identifying when these alterations are due to a pathological deterioration is still challenging. Current clinical tools rely either on subjective evaluations, basic volume measurements or disease-specific deep learning models. This paper introduces a novel method to learn normal asymmetry patterns in homologous brain structures based on anomaly detection and representation learning. Our framework uses a Siamese architecture to map 3D segmentations of left and right hemispherical sides of a brain structure to a normal asymmetry embedding space, learned using a support vector data description objective. Being trained using healthy samples only, it can quantify deviations-from-normal-asymmetry patterns in unseen samples by measuring the distance of their embeddings to the center of the learned normal space. We demonstrate in public and in-house sets that our method can accurately characterize normal asymmetries and detect pathological alterations due to Alzheimer’s disease and hippocampal sclerosis, even though no diseased cases were accessed for training. Our source code is available at https://github.com/duiliod/DeepNORHA
format info:eu-repo/semantics/preprint
submittedVersion
author Bendersky, Mariana
Iarussi, Emmanuel
Deangeli, Duilio
Princich, Juan Pablo
Larrabide, Ignacio
Orlando, José Ignacio
author_facet Bendersky, Mariana
Iarussi, Emmanuel
Deangeli, Duilio
Princich, Juan Pablo
Larrabide, Ignacio
Orlando, José Ignacio
author_sort Bendersky, Mariana
title Learning normal asymmetry representations for homologous brain structures
title_short Learning normal asymmetry representations for homologous brain structures
title_full Learning normal asymmetry representations for homologous brain structures
title_fullStr Learning normal asymmetry representations for homologous brain structures
title_full_unstemmed Learning normal asymmetry representations for homologous brain structures
title_sort learning normal asymmetry representations for homologous brain structures
publisher Universidad Torcuato Di Tella
publishDate 2023
url https://repositorio.utdt.edu/handle/20.500.13098/12140
work_keys_str_mv AT benderskymariana learningnormalasymmetryrepresentationsforhomologousbrainstructures
AT iarussiemmanuel learningnormalasymmetryrepresentationsforhomologousbrainstructures
AT deangeliduilio learningnormalasymmetryrepresentationsforhomologousbrainstructures
AT princichjuanpablo learningnormalasymmetryrepresentationsforhomologousbrainstructures
AT larrabideignacio learningnormalasymmetryrepresentationsforhomologousbrainstructures
AT orlandojoseignacio learningnormalasymmetryrepresentationsforhomologousbrainstructures
_version_ 1808040620656164864

Ejemplares similares