Reliable gene mutation prediction in clear cell renal cell carcinoma through multi-classifier multi-objective radiogenomics model

Xi Chen, Zhiguo Zhou, Raquibul Hannan, Kimberly Thomas, Ivan Pedrosa, Payal Kapur, James B Brugarolas, Xuanqin Mou, Jing Wang

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Genetic studies have identified associations between gene mutations and clear cell renal cell carcinoma (ccRCC). Since the complete gene mutational landscape cannot be characterized through biopsy and sequencing assays for each patient, non-invasive tools are needed to determine the mutation status for tumors. Radiogenomics may be an attractive alternative tool to identify disease genomics by analyzing amounts of features extracted from medical images. Most current radiogenomics predictive models are built based on a single classifier and trained through a single objective. However, since many classifiers are available, selecting an optimal model is challenging. On the other hand, a single objective may not be a good measure to guide model training. We proposed a new multi-classifier multi-objective (MCMO) radiogenomics predictive model. To obtain more reliable prediction results, similarity-based sensitivity and specificity were defined and considered as the two objective functions simultaneously during training. To take advantage of different classifiers, the evidential reasoning (ER) approach was used for fusing the output of each classifier. Additionally, a new similarity-based multi-objective optimization algorithm (SMO) was developed for training the MCMO to predict ccRCC related gene mutations (VHL, PBRM1 and BAP1) using quantitative CT features. Using the proposed MCMO model, we achieved a predictive area under the receiver operating characteristic curve (AUC) over 0.85 for VHL, PBRM1 and BAP1 genes with balanced sensitivity and specificity. Furthermore, MCMO outperformed all the individual classifiers, and yielded more reliable results than other optimization algorithms and commonly used fusion strategies.

Original languageEnglish (US)
Article number215008
JournalPhysics in Medicine and Biology
Volume63
Issue number21
DOIs
StatePublished - Oct 24 2018

Fingerprint

Renal Cell Carcinoma
Mutation
Genes
Sensitivity and Specificity
Genomics
ROC Curve
Area Under Curve
Biopsy
Neoplasms

Keywords

  • evidential reasoning
  • multi-classifier
  • multi-objective optimization
  • outcome prediction
  • radiogenomics

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Cite this

@article{ac429793003c43b78c15de1cf28975cb,
title = "Reliable gene mutation prediction in clear cell renal cell carcinoma through multi-classifier multi-objective radiogenomics model",
abstract = "Genetic studies have identified associations between gene mutations and clear cell renal cell carcinoma (ccRCC). Since the complete gene mutational landscape cannot be characterized through biopsy and sequencing assays for each patient, non-invasive tools are needed to determine the mutation status for tumors. Radiogenomics may be an attractive alternative tool to identify disease genomics by analyzing amounts of features extracted from medical images. Most current radiogenomics predictive models are built based on a single classifier and trained through a single objective. However, since many classifiers are available, selecting an optimal model is challenging. On the other hand, a single objective may not be a good measure to guide model training. We proposed a new multi-classifier multi-objective (MCMO) radiogenomics predictive model. To obtain more reliable prediction results, similarity-based sensitivity and specificity were defined and considered as the two objective functions simultaneously during training. To take advantage of different classifiers, the evidential reasoning (ER) approach was used for fusing the output of each classifier. Additionally, a new similarity-based multi-objective optimization algorithm (SMO) was developed for training the MCMO to predict ccRCC related gene mutations (VHL, PBRM1 and BAP1) using quantitative CT features. Using the proposed MCMO model, we achieved a predictive area under the receiver operating characteristic curve (AUC) over 0.85 for VHL, PBRM1 and BAP1 genes with balanced sensitivity and specificity. Furthermore, MCMO outperformed all the individual classifiers, and yielded more reliable results than other optimization algorithms and commonly used fusion strategies.",
keywords = "evidential reasoning, multi-classifier, multi-objective optimization, outcome prediction, radiogenomics",
author = "Xi Chen and Zhiguo Zhou and Raquibul Hannan and Kimberly Thomas and Ivan Pedrosa and Payal Kapur and Brugarolas, {James B} and Xuanqin Mou and Jing Wang",
year = "2018",
month = "10",
day = "24",
doi = "10.1088/1361-6560/aae5cd",
language = "English (US)",
volume = "63",
journal = "Physics in Medicine and Biology",
issn = "0031-9155",
publisher = "IOP Publishing Ltd.",
number = "21",

}

TY - JOUR

T1 - Reliable gene mutation prediction in clear cell renal cell carcinoma through multi-classifier multi-objective radiogenomics model

AU - Chen, Xi

AU - Zhou, Zhiguo

AU - Hannan, Raquibul

AU - Thomas, Kimberly

AU - Pedrosa, Ivan

AU - Kapur, Payal

AU - Brugarolas, James B

AU - Mou, Xuanqin

AU - Wang, Jing

PY - 2018/10/24

Y1 - 2018/10/24

N2 - Genetic studies have identified associations between gene mutations and clear cell renal cell carcinoma (ccRCC). Since the complete gene mutational landscape cannot be characterized through biopsy and sequencing assays for each patient, non-invasive tools are needed to determine the mutation status for tumors. Radiogenomics may be an attractive alternative tool to identify disease genomics by analyzing amounts of features extracted from medical images. Most current radiogenomics predictive models are built based on a single classifier and trained through a single objective. However, since many classifiers are available, selecting an optimal model is challenging. On the other hand, a single objective may not be a good measure to guide model training. We proposed a new multi-classifier multi-objective (MCMO) radiogenomics predictive model. To obtain more reliable prediction results, similarity-based sensitivity and specificity were defined and considered as the two objective functions simultaneously during training. To take advantage of different classifiers, the evidential reasoning (ER) approach was used for fusing the output of each classifier. Additionally, a new similarity-based multi-objective optimization algorithm (SMO) was developed for training the MCMO to predict ccRCC related gene mutations (VHL, PBRM1 and BAP1) using quantitative CT features. Using the proposed MCMO model, we achieved a predictive area under the receiver operating characteristic curve (AUC) over 0.85 for VHL, PBRM1 and BAP1 genes with balanced sensitivity and specificity. Furthermore, MCMO outperformed all the individual classifiers, and yielded more reliable results than other optimization algorithms and commonly used fusion strategies.

AB - Genetic studies have identified associations between gene mutations and clear cell renal cell carcinoma (ccRCC). Since the complete gene mutational landscape cannot be characterized through biopsy and sequencing assays for each patient, non-invasive tools are needed to determine the mutation status for tumors. Radiogenomics may be an attractive alternative tool to identify disease genomics by analyzing amounts of features extracted from medical images. Most current radiogenomics predictive models are built based on a single classifier and trained through a single objective. However, since many classifiers are available, selecting an optimal model is challenging. On the other hand, a single objective may not be a good measure to guide model training. We proposed a new multi-classifier multi-objective (MCMO) radiogenomics predictive model. To obtain more reliable prediction results, similarity-based sensitivity and specificity were defined and considered as the two objective functions simultaneously during training. To take advantage of different classifiers, the evidential reasoning (ER) approach was used for fusing the output of each classifier. Additionally, a new similarity-based multi-objective optimization algorithm (SMO) was developed for training the MCMO to predict ccRCC related gene mutations (VHL, PBRM1 and BAP1) using quantitative CT features. Using the proposed MCMO model, we achieved a predictive area under the receiver operating characteristic curve (AUC) over 0.85 for VHL, PBRM1 and BAP1 genes with balanced sensitivity and specificity. Furthermore, MCMO outperformed all the individual classifiers, and yielded more reliable results than other optimization algorithms and commonly used fusion strategies.

KW - evidential reasoning

KW - multi-classifier

KW - multi-objective optimization

KW - outcome prediction

KW - radiogenomics

UR - http://www.scopus.com/inward/record.url?scp=85056270983&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85056270983&partnerID=8YFLogxK

U2 - 10.1088/1361-6560/aae5cd

DO - 10.1088/1361-6560/aae5cd

M3 - Article

C2 - 30277889

AN - SCOPUS:85056270983

VL - 63

JO - Physics in Medicine and Biology

JF - Physics in Medicine and Biology

SN - 0031-9155

IS - 21

M1 - 215008

ER -