A Deep Learning-Based Radiomics Model for Prediction of Survival in Glioblastoma Multiforme

Traditional radiomics models mainly rely on explicitly-designed handcrafted features from medical images. This paper aimed to investigate if deep features extracted via transfer learning can generate radiomics signatures for prediction of overall survival (OS) in patients with Glioblastoma Multiform...

Full description

Saved in:
Bibliographic Details
Published in:Scientific reports 2017-09, Vol.7 (1), p.10353-8, Article 10353
Main Authors: Lao, Jiangwei, Chen, Yinsheng, Li, Zhi-Cheng, Li, Qihua, Zhang, Ji, Liu, Jing, Zhai, Guangtao
Format: Article
Language:English
Subjects:
59/57
631/114/1305
631/114/1564
639/166/985
Adolescent
Adult
Aged
Aged, 80 and over
Algorithms
Archives & records
Brain cancer
Cancer
Child
Datasets
Deep Learning
Edema
Feature selection
Female
Geometry
Glioblastoma
Glioblastoma - diagnostic imaging
Glioblastoma - mortality
Humanities and Social Sciences
Humans
Image Interpretation, Computer-Assisted
Image Processing, Computer-Assisted
Kaplan-Meier Estimate
Magnetic Resonance Imaging - methods
Male
Medical imaging
Medical prognosis
Middle Aged
Models, Theoretical
multidisciplinary
Neural networks
Nomograms
Oncology
Open source software
Patients
Prognosis
Public domain
Radiomics
Reproducibility of Results
Risk factors
ROC Curve
Science
Science (multidisciplinary)
Survival
Transfer learning
Young Adult
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Traditional radiomics models mainly rely on explicitly-designed handcrafted features from medical images. This paper aimed to investigate if deep features extracted via transfer learning can generate radiomics signatures for prediction of overall survival (OS) in patients with Glioblastoma Multiforme (GBM). This study comprised a discovery data set of 75 patients and an independent validation data set of 37 patients. A total of 1403 handcrafted features and 98304 deep features were extracted from preoperative multi-modality MR images. After feature selection, a six-deep-feature signature was constructed by using the least absolute shrinkage and selection operator (LASSO) Cox regression model. A radiomics nomogram was further presented by combining the signature and clinical risk factors such as age and Karnofsky Performance Score. Compared with traditional risk factors, the proposed signature achieved better performance for prediction of OS (C-index = 0.710, 95% CI: 0.588, 0.932) and significant stratification of patients into prognostically distinct groups (P 
ISSN:2045-2322
2045-2322
DOI:10.1038/s41598-017-10649-8