Computed Tomography–based Radiomics for Risk Stratification in Prostate Cancer

Sarah O.S. Osman*, Ralph T.H. Leijenaar, Aidan J. Cole, Ciara A. Lyons, Alan R. Hounsell, Kevin M. Prise, Joe M. O'Sullivan, Philippe Lambin, Conor K. McGarry, Suneil Jain

*Corresponding author for this work

Research output: Contribution to journalArticle

Abstract

Purpose: To explore the role of Computed tomography (CT)-based radiomics features in prostate cancer risk stratification. Methods and Materials: The study population consisted of 506 patients with prostate cancer collected from a clinically annotated database. After applying exclusion criteria, 342 patients were included in the final analysis. CT–based radiomics features were extracted from planning CT scans for prostate gland–only structure, and machine learning was used to train models for Gleason score (GS) and risk group (RG) classifications. Repeated cross-validation was used. The discriminatory performance of the developed models was assessed using receiver operating characteristic area under the curve (AUC) analysis. Results: Classifiers using CT-based radiomics features distinguished between GS ≤ 6 versus GS ≥ 7 with AUC = 0.90 and GS 7(3 + 4) versus GS 7(4 + 3) with AUC = 0.98. Developed classifiers also showed excellent performance in distinguishing low versus high RG (AUC = 0.96) and low versus intermediate RG (AUC = 1.00), but poorer performance was observed for GS 7 versus GS > 7 (AUC = 0.69). An overall modest performance was observed for validation on holdout data sets with the highest AUC of 0.75 for classifiers of low versus high RG and an AUC of 0.70 for GS 7 versus GS > 7. Conclusions: Our results show that radiomics features from routinely acquired planning CT scans could provide insights into prostate cancer aggressiveness in a noninvasive manner. Assessing models on training data sets, the classifiers were especially accurate in discerning high-risk from low-risk patients and in classifying GS 7 versus GS > 7 and GS 7(3 + 4) versus G7(4 + 3); however, classifiers were less adept at distinguishing high RG versus intermediate RG. External validation and prospective studies are warranted to verify the presented findings. These findings could potentially guide targeted radiation therapy strategies in radical intent radiation therapy for prostate cancer.

Original languageEnglish
JournalInternational Journal of Radiation Oncology Biology Physics
Early online date26 Jun 2019
DOIs
Publication statusEarly online date - 26 Jun 2019

Fingerprint

Neoplasm Grading
stratification
Prostatic Neoplasms
cancer
Area Under Curve
classifiers
curves
tomography
Tomography
planning
radiation therapy
machine learning
Radiotherapy
classifying
exclusion
Validation Studies
education
receivers
ROC Curve
Prostate

Cite this

@article{65e731760c08451299a2446a986d5c64,
title = "Computed Tomography–based Radiomics for Risk Stratification in Prostate Cancer",
abstract = "Purpose: To explore the role of Computed tomography (CT)-based radiomics features in prostate cancer risk stratification. Methods and Materials: The study population consisted of 506 patients with prostate cancer collected from a clinically annotated database. After applying exclusion criteria, 342 patients were included in the final analysis. CT–based radiomics features were extracted from planning CT scans for prostate gland–only structure, and machine learning was used to train models for Gleason score (GS) and risk group (RG) classifications. Repeated cross-validation was used. The discriminatory performance of the developed models was assessed using receiver operating characteristic area under the curve (AUC) analysis. Results: Classifiers using CT-based radiomics features distinguished between GS ≤ 6 versus GS ≥ 7 with AUC = 0.90 and GS 7(3 + 4) versus GS 7(4 + 3) with AUC = 0.98. Developed classifiers also showed excellent performance in distinguishing low versus high RG (AUC = 0.96) and low versus intermediate RG (AUC = 1.00), but poorer performance was observed for GS 7 versus GS > 7 (AUC = 0.69). An overall modest performance was observed for validation on holdout data sets with the highest AUC of 0.75 for classifiers of low versus high RG and an AUC of 0.70 for GS 7 versus GS > 7. Conclusions: Our results show that radiomics features from routinely acquired planning CT scans could provide insights into prostate cancer aggressiveness in a noninvasive manner. Assessing models on training data sets, the classifiers were especially accurate in discerning high-risk from low-risk patients and in classifying GS 7 versus GS > 7 and GS 7(3 + 4) versus G7(4 + 3); however, classifiers were less adept at distinguishing high RG versus intermediate RG. External validation and prospective studies are warranted to verify the presented findings. These findings could potentially guide targeted radiation therapy strategies in radical intent radiation therapy for prostate cancer.",
author = "Osman, {Sarah O.S.} and Leijenaar, {Ralph T.H.} and Cole, {Aidan J.} and Lyons, {Ciara A.} and Hounsell, {Alan R.} and Prise, {Kevin M.} and O'Sullivan, {Joe M.} and Philippe Lambin and McGarry, {Conor K.} and Suneil Jain",
year = "2019",
month = "6",
day = "26",
doi = "10.1016/j.ijrobp.2019.06.2504",
language = "English",
journal = "International Journal of Radiation: Oncology - Biology - Physics",
issn = "0360-3016",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - Computed Tomography–based Radiomics for Risk Stratification in Prostate Cancer

AU - Osman, Sarah O.S.

AU - Leijenaar, Ralph T.H.

AU - Cole, Aidan J.

AU - Lyons, Ciara A.

AU - Hounsell, Alan R.

AU - Prise, Kevin M.

AU - O'Sullivan, Joe M.

AU - Lambin, Philippe

AU - McGarry, Conor K.

AU - Jain, Suneil

PY - 2019/6/26

Y1 - 2019/6/26

N2 - Purpose: To explore the role of Computed tomography (CT)-based radiomics features in prostate cancer risk stratification. Methods and Materials: The study population consisted of 506 patients with prostate cancer collected from a clinically annotated database. After applying exclusion criteria, 342 patients were included in the final analysis. CT–based radiomics features were extracted from planning CT scans for prostate gland–only structure, and machine learning was used to train models for Gleason score (GS) and risk group (RG) classifications. Repeated cross-validation was used. The discriminatory performance of the developed models was assessed using receiver operating characteristic area under the curve (AUC) analysis. Results: Classifiers using CT-based radiomics features distinguished between GS ≤ 6 versus GS ≥ 7 with AUC = 0.90 and GS 7(3 + 4) versus GS 7(4 + 3) with AUC = 0.98. Developed classifiers also showed excellent performance in distinguishing low versus high RG (AUC = 0.96) and low versus intermediate RG (AUC = 1.00), but poorer performance was observed for GS 7 versus GS > 7 (AUC = 0.69). An overall modest performance was observed for validation on holdout data sets with the highest AUC of 0.75 for classifiers of low versus high RG and an AUC of 0.70 for GS 7 versus GS > 7. Conclusions: Our results show that radiomics features from routinely acquired planning CT scans could provide insights into prostate cancer aggressiveness in a noninvasive manner. Assessing models on training data sets, the classifiers were especially accurate in discerning high-risk from low-risk patients and in classifying GS 7 versus GS > 7 and GS 7(3 + 4) versus G7(4 + 3); however, classifiers were less adept at distinguishing high RG versus intermediate RG. External validation and prospective studies are warranted to verify the presented findings. These findings could potentially guide targeted radiation therapy strategies in radical intent radiation therapy for prostate cancer.

AB - Purpose: To explore the role of Computed tomography (CT)-based radiomics features in prostate cancer risk stratification. Methods and Materials: The study population consisted of 506 patients with prostate cancer collected from a clinically annotated database. After applying exclusion criteria, 342 patients were included in the final analysis. CT–based radiomics features were extracted from planning CT scans for prostate gland–only structure, and machine learning was used to train models for Gleason score (GS) and risk group (RG) classifications. Repeated cross-validation was used. The discriminatory performance of the developed models was assessed using receiver operating characteristic area under the curve (AUC) analysis. Results: Classifiers using CT-based radiomics features distinguished between GS ≤ 6 versus GS ≥ 7 with AUC = 0.90 and GS 7(3 + 4) versus GS 7(4 + 3) with AUC = 0.98. Developed classifiers also showed excellent performance in distinguishing low versus high RG (AUC = 0.96) and low versus intermediate RG (AUC = 1.00), but poorer performance was observed for GS 7 versus GS > 7 (AUC = 0.69). An overall modest performance was observed for validation on holdout data sets with the highest AUC of 0.75 for classifiers of low versus high RG and an AUC of 0.70 for GS 7 versus GS > 7. Conclusions: Our results show that radiomics features from routinely acquired planning CT scans could provide insights into prostate cancer aggressiveness in a noninvasive manner. Assessing models on training data sets, the classifiers were especially accurate in discerning high-risk from low-risk patients and in classifying GS 7 versus GS > 7 and GS 7(3 + 4) versus G7(4 + 3); however, classifiers were less adept at distinguishing high RG versus intermediate RG. External validation and prospective studies are warranted to verify the presented findings. These findings could potentially guide targeted radiation therapy strategies in radical intent radiation therapy for prostate cancer.

UR - http://www.scopus.com/inward/record.url?scp=85069869631&partnerID=8YFLogxK

U2 - 10.1016/j.ijrobp.2019.06.2504

DO - 10.1016/j.ijrobp.2019.06.2504

M3 - Article

AN - SCOPUS:85069869631

JO - International Journal of Radiation: Oncology - Biology - Physics

JF - International Journal of Radiation: Oncology - Biology - Physics

SN - 0360-3016

ER -