Comparison of quantitative parameters and radiomic features as inputs into machine learning models to predict the Gleason score of prostate cancer lesions

• Published in: Magnetic resonance imaging Vol. 100; pp. 64 - 72
• Main Authors: Nai, Ying-Hwey, Cheong, Dennis Lai Hong, Roy, Sharmili, Kok, Trina, Stephenson, Mary C., Schaefferkoetter, Josh, Totman, John J., Conti, Maurizio, Eriksson, Lars, Robins, Edward G., Wang, Ziting, Chua, Wynne Yuru, Ang, Bertrand Wei Leng, Singha, Arvind Kumar, Thamboo, Thomas Paulraj, Chiong, Edmund, Reilhac, Anthonin
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier Inc 01.07.2023
• Subjects: Gleason score (GS); Humans; Machine Learning; Machine learning (ML); Magnetic Resonance Imaging - methods; Male; Multiparametric magnetic resonance imaging (mpMRI); Neoplasm Grading; Positron emission tomography (PET); Prostate-Specific Antigen; Prostatic Neoplasms - pathology; Quantitative parameters; Radiomics; Retrospective Studies; PRC; ADC; TZ; PZ; PSADWP; AUC; GLDM; DT; DWI; EP2D; ML; PSA; DCE; ROC; TSE; EM; Gleason score (GS); ROI; GLCM; Ktrans; SUV; GRE; GLRLM; PSAD; SUVmax; SVM; TOF-PSF; GLSZM; Radiomics; PCa; Ve; Multiparametric magnetic resonance imaging (mpMRI); Quantitative parameters; NPV; OP-OSEM; PPV; kep; mRMR; T2w; TURP; WP; SUVmean; PSADTZ; Positron emission tomography (PET); CG; NGTDM; Machine learning (ML); kNN; ADC ratio; GS; CT; PSMA; mpMRI; csPCa; PI-RADS; PSADCG; PET
• ISSN: 0730-725X; 1873-5894
• DOI: 10.1016/j.mri.2023.03.009

The classification of prostate cancer (PCa) lesions using Prostate Imaging Reporting and Data System (PI-RADS) suffers from poor inter-reader agreement. This study compared quantitative parameters or radiomic features from multiparametric magnetic resonance imaging (mpMRI) or positron emission tomography (PET), as inputs into machine learning (ML) to predict the Gleason scores (GS) of detected lesions for improved PCa lesion classification. 20 biopsy-confirmed PCa subjects underwent imaging before radical prostatectomy. A pathologist assigned GS from tumour tissue. Two radiologists and one nuclear medicine physician delineated the lesions on the mpMR and PET images, yielding 45 lesion inputs. Seven quantitative parameters were extracted from the lesions, namely T2-weighted (T2w) image intensity, apparent diffusion coefficient (ADC), transfer constant (KTRANS), efflux rate constant (Kep), and extracellular volume ratio (Ve) from mpMR images, and SUVmean and SUVmax from PET images. Eight radiomic features were selected out of 109 radiomic features from T2w, ADC and PET images. Quantitative parameters or radiomic features, with risk factors of age, prostate-specific antigen (PSA), PSA density and volume, of 45 different lesion inputs were input in different combinations into four ML models - Decision Tree (DT), Support Vector Machine (SVM), k-Nearest-Neighbour (kNN), Ensembles model (EM). SUVmax yielded the highest accuracy in discriminating detected lesions. Among the 4 ML models, kNN yielded the highest accuracies of 0.929 using either quantitative parameters or radiomic features with risk factors as input. ML models' performance is dependent on the input combinations and risk factors further improve ML classification accuracy. [Display omitted] •Among 7 quantitative parameters, SUVmax correlated best with Gleason score (GS).•Machine learning yielded higher accuracies than using cut-points of parameters.•Quantitative parameters (QP) and radiomic features (RF) as inputs for GS prediction.•K-Nearest Neighbour yielded highest accuracy of 0.929 with QP and RF.•Combinations of parameters, risk factors and models affect classification accuracy.