Hyesun Park1, Lei Qin2, Pamela Guerra2, Camden P Bay2, Atul B Shinagare2. 1. Department of Radiology, Brigham and Women's Hospital and Dana-Farber Cancer Institute, 450 Brookline Ave., Boston, MA, 02215, USA. hpark29@bwh.harvard.edu. 2. Department of Radiology, Brigham and Women's Hospital and Dana-Farber Cancer Institute, 450 Brookline Ave., Boston, MA, 02215, USA.
Abstract
PURPOSE: To compare CT texture features of benign and malignant ovarian lesions and to build a machine learning model to detect malignancy in incidental ovarian lesions. METHODS: In this IRB-approved, HIPAA-compliant, retrospective study, 427 consecutive patients with incidental ovarian lesions detected on contrast-enhanced CT (348, 81.5% benign and 79, 18.5% malignant) were included. The following CT texture features were analyzed using commercially available software (TexRAD, Feedback Plc, Cambridge, UK): total pixel, mean, standard deviation (SD), entropy, mean value of positive pixels (MPP), skewness, kurtosis and entropy. Three machine learning models were created by combining texture features and patients' age, and performance of these models was assessed using tenfold cross-validation. Receiver operating characteristics (ROC) were constructed to assess sensitivity and specificity. The cutoff value was picked using a cost-weighted method. RESULTS: Total pixels, mean, SD, entropy, MPP, and skewness were significantly different between benign and malignant groups (p < 0.05). With a selected 10 as a cost factor to optimize cutoff value selection, sensitivity 92%, specificity 60% in the random forest (RF) model, sensitivity 91%, specificity 69% in SVM model, and sensitivity 92%, specificity 61% in the logistic regression, respectively. CONCLUSION: CT texture analysis could provide objective imaging analysis of incidental ovarian lesions and ML models using CT texture features and age demonstrated high sensitivity and moderate specificity for detection of malignant lesions.
PURPOSE: To compare CT texture features of benign and malignant ovarian lesions and to build a machine learning model to detect malignancy in incidental ovarian lesions. METHODS: In this IRB-approved, HIPAA-compliant, retrospective study, 427 consecutive patients with incidental ovarian lesions detected on contrast-enhanced CT (348, 81.5% benign and 79, 18.5% malignant) were included. The following CT texture features were analyzed using commercially available software (TexRAD, Feedback Plc, Cambridge, UK): total pixel, mean, standard deviation (SD), entropy, mean value of positive pixels (MPP), skewness, kurtosis and entropy. Three machine learning models were created by combining texture features and patients' age, and performance of these models was assessed using tenfold cross-validation. Receiver operating characteristics (ROC) were constructed to assess sensitivity and specificity. The cutoff value was picked using a cost-weighted method. RESULTS: Total pixels, mean, SD, entropy, MPP, and skewness were significantly different between benign and malignant groups (p < 0.05). With a selected 10 as a cost factor to optimize cutoff value selection, sensitivity 92%, specificity 60% in the random forest (RF) model, sensitivity 91%, specificity 69% in SVM model, and sensitivity 92%, specificity 61% in the logistic regression, respectively. CONCLUSION: CT texture analysis could provide objective imaging analysis of incidental ovarian lesions and ML models using CT texture features and age demonstrated high sensitivity and moderate specificity for detection of malignant lesions.
Authors: Hebert Alberto Vargas; Harini Veeraraghavan; Maura Micco; Stephanie Nougaret; Yulia Lakhman; Andreas A Meier; Ramon Sosa; Robert A Soslow; Douglas A Levine; Britta Weigelt; Carol Aghajanian; Hedvig Hricak; Joseph Deasy; Alexandra Snyder; Evis Sala Journal: Eur Radiol Date: 2017-03-13 Impact factor: 5.315
Authors: Maitray D Patel; Susan M Ascher; Raj Mohan Paspulati; Alampady K Shanbhogue; Evan S Siegelman; Marjorie W Stein; Lincoln L Berland Journal: J Am Coll Radiol Date: 2013-09 Impact factor: 5.532
Authors: Deborah Levine; Maitray D Patel; Elizabeth J Suh-Burgmann; Rochelle F Andreotti; Beryl R Benacerraf; Carol B Benson; Wendy R Brewster; Beverly G Coleman; Peter M Doubilet; Steven R Goldstein; Ulrike M Hamper; Jonathan L Hecht; Mindy M Horrow; Hye-Chun Hur; Mary L Marnach; Ed Pavlik; Lawrence D Platt; Elizabeth Puscheck; Rebecca Smith-Bindman; Douglas L Brown Journal: Radiology Date: 2019-09-24 Impact factor: 11.105
Authors: Meghan G Lubner; Nicholas Stabo; E Jason Abel; Alejandro Munoz Del Rio; Perry J Pickhardt Journal: AJR Am J Roentgenol Date: 2016-05-04 Impact factor: 3.959