Evaluation of a deep learning architecture for MR imaging prediction of ATRX in glioma patients

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Predicting mutation/loss of alpha-thalassemia/mental retardation syndrome X-linked (ATRX) gene utilizing MR imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare a deep neural network approach based on a residual deep neural network (ResNet) architecture and one based on a classical machine learning approach and evaluate their ability in predicting ATRX mutation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture, pre trained on ImageNet data was the best performing model, achieving an accuracy of 0.91 for the test set (classification of a slice as no tumor, ATRX mutated, or mutated) in terms of f1 score in a test set of 35 cases. The SVM classifier achieved 0.63 for differentiating the Flair signal abnormality regions from the test patients based on their mutation status. We report a method that alleviates the need for extensive preprocessing and acts as a proof of concept that deep neural network architectures can be used to predict molecular biomarkers from routine medical images.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2018
Subtitle of host publicationComputer-Aided Diagnosis
PublisherSPIE
Volume10575
ISBN (Electronic)9781510616394
DOIs
StatePublished - Jan 1 2018
EventMedical Imaging 2018: Computer-Aided Diagnosis - Houston, United States
Duration: Feb 12 2018Feb 15 2018

Other

OtherMedical Imaging 2018: Computer-Aided Diagnosis
CountryUnited States
CityHouston
Period2/12/182/15/18

Fingerprint

mutations
Glioma
learning
Tumors
tumors
Learning
Network architecture
Imaging techniques
Mutation
evaluation
predictions
alpha-Thalassemia
X-Linked Genes
machine learning
Aptitude
prognosis
biomarkers
abnormalities
Biomarkers
preprocessing

Keywords

  • ATRX imaging biomarkers
  • Deep learning
  • glioma
  • MRI

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Atomic and Molecular Physics, and Optics
  • Radiology Nuclear Medicine and imaging

Cite this

Evaluation of a deep learning architecture for MR imaging prediction of ATRX in glioma patients. / Korfiatis, Panagiotis; Kline, Timothy; Erickson, Bradley J.

Medical Imaging 2018: Computer-Aided Diagnosis. Vol. 10575 SPIE, 2018. 105752G.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Korfiatis, P, Kline, T & Erickson, BJ 2018, Evaluation of a deep learning architecture for MR imaging prediction of ATRX in glioma patients. in Medical Imaging 2018: Computer-Aided Diagnosis. vol. 10575, 105752G, SPIE, Medical Imaging 2018: Computer-Aided Diagnosis, Houston, United States, 2/12/18. https://doi.org/10.1117/12.2293538
@inproceedings{aef107bd284340a99bf2c42a340b4311,
title = "Evaluation of a deep learning architecture for MR imaging prediction of ATRX in glioma patients",
abstract = "Predicting mutation/loss of alpha-thalassemia/mental retardation syndrome X-linked (ATRX) gene utilizing MR imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare a deep neural network approach based on a residual deep neural network (ResNet) architecture and one based on a classical machine learning approach and evaluate their ability in predicting ATRX mutation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture, pre trained on ImageNet data was the best performing model, achieving an accuracy of 0.91 for the test set (classification of a slice as no tumor, ATRX mutated, or mutated) in terms of f1 score in a test set of 35 cases. The SVM classifier achieved 0.63 for differentiating the Flair signal abnormality regions from the test patients based on their mutation status. We report a method that alleviates the need for extensive preprocessing and acts as a proof of concept that deep neural network architectures can be used to predict molecular biomarkers from routine medical images.",
keywords = "ATRX imaging biomarkers, Deep learning, glioma, MRI",
author = "Panagiotis Korfiatis and Timothy Kline and Erickson, {Bradley J}",
year = "2018",
month = "1",
day = "1",
doi = "10.1117/12.2293538",
language = "English (US)",
volume = "10575",
booktitle = "Medical Imaging 2018",
publisher = "SPIE",

}

TY - GEN

T1 - Evaluation of a deep learning architecture for MR imaging prediction of ATRX in glioma patients

AU - Korfiatis, Panagiotis

AU - Kline, Timothy

AU - Erickson, Bradley J

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Predicting mutation/loss of alpha-thalassemia/mental retardation syndrome X-linked (ATRX) gene utilizing MR imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare a deep neural network approach based on a residual deep neural network (ResNet) architecture and one based on a classical machine learning approach and evaluate their ability in predicting ATRX mutation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture, pre trained on ImageNet data was the best performing model, achieving an accuracy of 0.91 for the test set (classification of a slice as no tumor, ATRX mutated, or mutated) in terms of f1 score in a test set of 35 cases. The SVM classifier achieved 0.63 for differentiating the Flair signal abnormality regions from the test patients based on their mutation status. We report a method that alleviates the need for extensive preprocessing and acts as a proof of concept that deep neural network architectures can be used to predict molecular biomarkers from routine medical images.

AB - Predicting mutation/loss of alpha-thalassemia/mental retardation syndrome X-linked (ATRX) gene utilizing MR imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare a deep neural network approach based on a residual deep neural network (ResNet) architecture and one based on a classical machine learning approach and evaluate their ability in predicting ATRX mutation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture, pre trained on ImageNet data was the best performing model, achieving an accuracy of 0.91 for the test set (classification of a slice as no tumor, ATRX mutated, or mutated) in terms of f1 score in a test set of 35 cases. The SVM classifier achieved 0.63 for differentiating the Flair signal abnormality regions from the test patients based on their mutation status. We report a method that alleviates the need for extensive preprocessing and acts as a proof of concept that deep neural network architectures can be used to predict molecular biomarkers from routine medical images.

KW - ATRX imaging biomarkers

KW - Deep learning

KW - glioma

KW - MRI

UR - http://www.scopus.com/inward/record.url?scp=85046262928&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85046262928&partnerID=8YFLogxK

U2 - 10.1117/12.2293538

DO - 10.1117/12.2293538

M3 - Conference contribution

VL - 10575

BT - Medical Imaging 2018

PB - SPIE

ER -