Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- And PET-imaged head and neck cancer patients

Rachel B. Ger, Shouhao Zhou, Baher Elgohari, Hesham Elhalawani, Dennis M. Mackin, Joseph G. Meier, Callistus M. Nguyen, Brian M. Anderson, Casey Gay, Jing Ning, Clifton D. Fuller, Heng Li, Rebecca M. Howell, Rick R. Layman, Osama Mawlawi, R. Jason Stafford, Hugo Aerts, Laurence E. Court

Research output: Contribution to journalArticle

Abstract

Radiomics studies require many patients in order to power them, thus patients are often combined from different institutions and using different imaging protocols. Various studies have shown that imaging protocols affect radiomics feature values. We examined whether using data from cohorts with controlled imaging protocols improved patient outcome models. We retrospectively reviewed 726 CT and 686 PET images from head and neck cancer patients, who were divided into training or independent testing cohorts. For each patient, radiomics features with different preprocessing were calculated and two clinical variables—HPV status and tumor volume—were also included. A Cox proportional hazards model was built on the training data by using bootstrapped Lasso regression to predict overall survival. The effect of controlled imaging protocols on model performance was evaluated by subsetting the original training and independent testing cohorts to include only patients whose images were obtained using the same imaging protocol and vendor. Tumor volume, HPV status, and two radiomics covariates were selected for the CT model, resulting in an AUC of 0.72. However, volume alone produced a higher AUC, whereas adding radiomics features reduced the AUC. HPV status and one radiomics feature were selected as covariates for the PET model, resulting in an AUC of 0.59, but neither covariate was significantly associated with survival. Limiting the training and independent testing to patients with the same imaging protocol reduced the AUC for CT patients to 0.55, and no covariates were selected for PET patients. Radiomics features were not consistently associated with survival in CT or PET images of head and neck patients, even within patients with the same imaging protocol.

Original languageEnglish (US)
Article numbere0222509
JournalPloS one
Volume14
Issue number9
DOIs
StatePublished - Jan 1 2019

Fingerprint

Head and Neck Neoplasms
Tumors
Imaging techniques
neoplasms
prediction
Survival
image analysis
Neoplasms
Area Under Curve
Testing
head and neck neoplasms
Hazards
testing
Tumor Burden
Proportional Hazards Models
neck
Neck
Head

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)

Cite this

Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- And PET-imaged head and neck cancer patients. / Ger, Rachel B.; Zhou, Shouhao; Elgohari, Baher; Elhalawani, Hesham; Mackin, Dennis M.; Meier, Joseph G.; Nguyen, Callistus M.; Anderson, Brian M.; Gay, Casey; Ning, Jing; Fuller, Clifton D.; Li, Heng; Howell, Rebecca M.; Layman, Rick R.; Mawlawi, Osama; Stafford, R. Jason; Aerts, Hugo; Court, Laurence E.

In: PloS one, Vol. 14, No. 9, e0222509, 01.01.2019.

Research output: Contribution to journalArticle

Ger, RB, Zhou, S, Elgohari, B, Elhalawani, H, Mackin, DM, Meier, JG, Nguyen, CM, Anderson, BM, Gay, C, Ning, J, Fuller, CD, Li, H, Howell, RM, Layman, RR, Mawlawi, O, Stafford, RJ, Aerts, H & Court, LE 2019, 'Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- And PET-imaged head and neck cancer patients', PloS one, vol. 14, no. 9, e0222509. https://doi.org/10.1371/journal.pone.0222509
Ger, Rachel B. ; Zhou, Shouhao ; Elgohari, Baher ; Elhalawani, Hesham ; Mackin, Dennis M. ; Meier, Joseph G. ; Nguyen, Callistus M. ; Anderson, Brian M. ; Gay, Casey ; Ning, Jing ; Fuller, Clifton D. ; Li, Heng ; Howell, Rebecca M. ; Layman, Rick R. ; Mawlawi, Osama ; Stafford, R. Jason ; Aerts, Hugo ; Court, Laurence E. / Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- And PET-imaged head and neck cancer patients. In: PloS one. 2019 ; Vol. 14, No. 9.
@article{e9e51d2b23f34400940300156d451183,
title = "Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- And PET-imaged head and neck cancer patients",
abstract = "Radiomics studies require many patients in order to power them, thus patients are often combined from different institutions and using different imaging protocols. Various studies have shown that imaging protocols affect radiomics feature values. We examined whether using data from cohorts with controlled imaging protocols improved patient outcome models. We retrospectively reviewed 726 CT and 686 PET images from head and neck cancer patients, who were divided into training or independent testing cohorts. For each patient, radiomics features with different preprocessing were calculated and two clinical variables—HPV status and tumor volume—were also included. A Cox proportional hazards model was built on the training data by using bootstrapped Lasso regression to predict overall survival. The effect of controlled imaging protocols on model performance was evaluated by subsetting the original training and independent testing cohorts to include only patients whose images were obtained using the same imaging protocol and vendor. Tumor volume, HPV status, and two radiomics covariates were selected for the CT model, resulting in an AUC of 0.72. However, volume alone produced a higher AUC, whereas adding radiomics features reduced the AUC. HPV status and one radiomics feature were selected as covariates for the PET model, resulting in an AUC of 0.59, but neither covariate was significantly associated with survival. Limiting the training and independent testing to patients with the same imaging protocol reduced the AUC for CT patients to 0.55, and no covariates were selected for PET patients. Radiomics features were not consistently associated with survival in CT or PET images of head and neck patients, even within patients with the same imaging protocol.",
author = "Ger, {Rachel B.} and Shouhao Zhou and Baher Elgohari and Hesham Elhalawani and Mackin, {Dennis M.} and Meier, {Joseph G.} and Nguyen, {Callistus M.} and Anderson, {Brian M.} and Casey Gay and Jing Ning and Fuller, {Clifton D.} and Heng Li and Howell, {Rebecca M.} and Layman, {Rick R.} and Osama Mawlawi and Stafford, {R. Jason} and Hugo Aerts and Court, {Laurence E.}",
year = "2019",
month = "1",
day = "1",
doi = "10.1371/journal.pone.0222509",
language = "English (US)",
volume = "14",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "9",

}

TY - JOUR

T1 - Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- And PET-imaged head and neck cancer patients

AU - Ger, Rachel B.

AU - Zhou, Shouhao

AU - Elgohari, Baher

AU - Elhalawani, Hesham

AU - Mackin, Dennis M.

AU - Meier, Joseph G.

AU - Nguyen, Callistus M.

AU - Anderson, Brian M.

AU - Gay, Casey

AU - Ning, Jing

AU - Fuller, Clifton D.

AU - Li, Heng

AU - Howell, Rebecca M.

AU - Layman, Rick R.

AU - Mawlawi, Osama

AU - Stafford, R. Jason

AU - Aerts, Hugo

AU - Court, Laurence E.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Radiomics studies require many patients in order to power them, thus patients are often combined from different institutions and using different imaging protocols. Various studies have shown that imaging protocols affect radiomics feature values. We examined whether using data from cohorts with controlled imaging protocols improved patient outcome models. We retrospectively reviewed 726 CT and 686 PET images from head and neck cancer patients, who were divided into training or independent testing cohorts. For each patient, radiomics features with different preprocessing were calculated and two clinical variables—HPV status and tumor volume—were also included. A Cox proportional hazards model was built on the training data by using bootstrapped Lasso regression to predict overall survival. The effect of controlled imaging protocols on model performance was evaluated by subsetting the original training and independent testing cohorts to include only patients whose images were obtained using the same imaging protocol and vendor. Tumor volume, HPV status, and two radiomics covariates were selected for the CT model, resulting in an AUC of 0.72. However, volume alone produced a higher AUC, whereas adding radiomics features reduced the AUC. HPV status and one radiomics feature were selected as covariates for the PET model, resulting in an AUC of 0.59, but neither covariate was significantly associated with survival. Limiting the training and independent testing to patients with the same imaging protocol reduced the AUC for CT patients to 0.55, and no covariates were selected for PET patients. Radiomics features were not consistently associated with survival in CT or PET images of head and neck patients, even within patients with the same imaging protocol.

AB - Radiomics studies require many patients in order to power them, thus patients are often combined from different institutions and using different imaging protocols. Various studies have shown that imaging protocols affect radiomics feature values. We examined whether using data from cohorts with controlled imaging protocols improved patient outcome models. We retrospectively reviewed 726 CT and 686 PET images from head and neck cancer patients, who were divided into training or independent testing cohorts. For each patient, radiomics features with different preprocessing were calculated and two clinical variables—HPV status and tumor volume—were also included. A Cox proportional hazards model was built on the training data by using bootstrapped Lasso regression to predict overall survival. The effect of controlled imaging protocols on model performance was evaluated by subsetting the original training and independent testing cohorts to include only patients whose images were obtained using the same imaging protocol and vendor. Tumor volume, HPV status, and two radiomics covariates were selected for the CT model, resulting in an AUC of 0.72. However, volume alone produced a higher AUC, whereas adding radiomics features reduced the AUC. HPV status and one radiomics feature were selected as covariates for the PET model, resulting in an AUC of 0.59, but neither covariate was significantly associated with survival. Limiting the training and independent testing to patients with the same imaging protocol reduced the AUC for CT patients to 0.55, and no covariates were selected for PET patients. Radiomics features were not consistently associated with survival in CT or PET images of head and neck patients, even within patients with the same imaging protocol.

UR - http://www.scopus.com/inward/record.url?scp=85072372778&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072372778&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0222509

DO - 10.1371/journal.pone.0222509

M3 - Article

VL - 14

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 9

M1 - e0222509

ER -