Utility of radiomic zones for risk classification and clinical outcome predictions using supervised machine learning during simultaneous ¹¹C‐choline PET/MRI acquisition in prostate cancer patients

Abstract

Purpose In most radiomic studies related to cancer research, the traditional tumor‐centric view has predominated. In this retrospective study, we go beyond the single‐tumor region and investigate the utility of proposed radiomic zones for risk classification and clinical outcome predictions using radiomic features extracted from 11C‐choline positron emission tomography (PET) imaging and supervised machine learning in prostate tumors. Materials and Methods Seventy‐seven prostate tumors were selected and delineated. The prostate organ was divided into three radiomic zones, with zone‐1 being the metabolic tumor zone, zone‐2 the proximal peripheral tumor zone, and zone‐3 the extended peripheral tumor zone. LIFEx was used for PET‐radiomic feature extraction. Risk groups were created using Gleason scores (GS), prostate‐specific antigen (PSA) levels, clinical TNM staging, and progression‐free survival (PFS). Random forest (RF) and AdaBoost advanced machine learning algorithms were used for supervised machine learning. Accuracy, positive predictive value, area under the receiver operating characteristic curve (AreaROC), and other metrics were calculated for comparisons of predictive performance between zones. Results For the GS risk classification group, the accuracies of risk classification predictions were 71%, 71%, and 67% using RF and 65%, 64%, and 63% using AdaBoost for zones −1, −2, and −3, respectively. For the PSA group, the accuracies of risk classification predictions were 74%, 65%, and 64% using RF and 76%, 66%, and 67% using AdaBoost for zones −1, −2, and −3, respectively. For the TNM group, the accuracies of risk classification predictions were 68%, 76%, and 78% using RF and 66%, 75%, and 80% using AdaBoost for zones −1, −2, and −3, respectively. For the PFS group, the accuracies of clinical outcome predictions were 77%, 75%, and 83% using RF and 77%, 74%, and 83% using AdaBoost in zones −1, −2, and −3, respectively. Conclusions We proposed three radiomic zones with different standard uptake value characteristics and created four risk groups of prostate cancer patients for testing this idea. We showed that these radiomic zones have different predicting strengths in classifying risk groups and might allow us to identify a radiomic zone with higher accuracy for patient outcome prediction.

Additional Value of PET Radiomic Features for the Initial Staging of Prostate Cancer: A Systematic Review from the Literature

“…In the study of Tu et al [ 19 ] the authors went beyond the traditional tumor-centric view of radiomic analysis; indeed, they evaluated 77 prostate tumors, but divided the whole prostate organ in three radiomic zones: Zone-1, the metabolic tumor zone; zone-2, the proximal peripheral tumor zone, and zone-3, the extended peripheral tumor zone inside the imaging boundaries of the organ. The radiomic analysis of the three zones was used for risk classification prediction, including the prediction of GS, PSA, TNM, and PFS, and the authors found that these zones have different predicting strengths in classifying risk groups.…”

“…All the considered studies had an RQS lower than 18 (50%) resulting in being non-compliant with the best-practice procedures. In particular, the RQSs ranged from a minimum of 8 (22.22%) for Tu et al [ 19 ] to a maximum of 14 (38.89%) for Zamboglou et al [ 15 ]. The study of Zamboglou et al [ 20 ] received an RQS of 13 (36.11%), followed by Papp et al [ 17 ] and Cysouw et al [ 21 ] with 11 (30.56%) and Solari et al [ 18 ] that scored 10 (27.78%).…”

“…Parameter settings for radiomic feature extraction were rather heterogeneous among studies: Three out of six used isotropic images (two studies resampled the original images to isotropic voxels of 2 × 2 × 2 mm, while in one case, resampling was unnecessary since voxels were already isotropic), five out of six applied intensity discretization with the fixed bin width (FBW) approach (e.g., widths of 0.05 or 0.25 for PET SUV and 5 for ADC) or the fixed bin number (FBN) approach (for MRI data), and five out of six used a 3D feature aggregation method to calculate texture parameters. The study of Tu et al [ 19 ] instead did not report details on feature extraction parameters.…”

“…Regarding tumor delineation, radiomic and clinical features were extracted from the whole prostate in four studies, but other approaches were also proposed. Zamboglou et al [ 15 ] considered manual GTV delineation and the GTVs resulting both from histology and a 40% threshold on SUVmax; Tu et al [ 19 ] defined two more zones based on SUV thresholds (i.e., SUV > 40%, 30% < SUV < 40%), while Cysouw et al [ 21 ] delineated primary tumors using 50–70% background-adapted peak thresholds on images with and without partial-volume correction in order to assess model stability with a threshold.…”

Molecular Imaging in Primary Staging of Prostate Cancer Patients: Current Aspects and Future Trends

“…Also, they were significantly superior over known clinical, laboratory and histopathological adverse features in predicting biochemical recurrence (AUC 0.90) and overall patient risk (AUC 0.94) [ 218 ]. Additionally, a recent study reported predictive values of the radiomics (derived from metabolic tumor and peripheral zones) for Gleason score, PSA group, TNM stage, and progression-free survival [ 219 ].…”