Radiomic features for prostate cancer detection on MRI differ between the transition and peripheral zones: Preliminary findings from a multi‐institutional study

Abstract

Purpose To evaluate in a multi-institutional study whether radiomic features useful for prostate cancer (PCa) detection from 3 Tesla (T) multi-parametric MRI (mpMRI) in the transition zone (TZ) differ from those in the peripheral zone (PZ). Materials and Methods 3T mpMRI, including T2-weighted (T2w), apparent diffusion coefficient (ADC) maps, and dynamic contrast-enhanced MRI (DCE-MRI), were retrospectively obtained from 80 patients at three institutions. This study was approved by the institutional review board of each participating institution. First-order statistical, co-occurrence, and wavelet features were extracted from T2w MRI and ADC maps, and contrast kinetic features were extracted from DCE-MRI. Feature selection was performed to identify ten features for PCa detection in the TZ and PZ, respectively. Two logistic regression classifiers used these features to detect PCa and were evaluated by area under the receiver-operating characteristic curve (AUC). Classifier performance was compared with a zone-ignorant classifier. Results Radiomic features that were identified as useful for PCa detection differed between TZ and PZ. When classification was performed on a per-voxel basis, a PZ-specific classifier detected PZ tumors on an independent test set with significantly higher accuracy (AUC = 0.61-0.71) than a zone-ignorant classifier trained to detect cancer throughout the entire prostate (p<0.05). When classifiers were evaluated on MRI data from multiple institutions, statistically similar AUC values (p > 0.14) were obtained for all institutions. Conclusions A zone-aware classifier significantly improves the accuracy of cancer detection in the PZ.

Transition zone prostate cancer: Logistic regression and machine‐learning models of quantitative ADC, shape and texture features are highly accurate for diagnosis

“…Quantitative ADC and texture features have previously been investigated and found to be significant for diagnosis of cancer using logistic regression and machine-learning models in both the PZ and TZ. [36][37][38][39] Among features studied, mean ADC, skewness, and entropy were all among the reported significant discriminatory features that could differentiate between clinically significant and low-risk (Gleason score = 6) cancers, as well as between Gleason score 3 + 4 vs. Gleason score 4 + 3 tumors using SVM and other machine-learning models. 36,37 Li et al reported AUC values using an initial set of features compared with the best performing subset/technique varying between 0.97 (CI 0.94-0.99) and 0.91 (CI 0.85-0.95) evaluating strictly TZ abnormalities, 36 while Fehr et al reported AUC values between 0.60 and 0.99 (CIs not provided) evaluating the TZ and PZ together.…”

“…Quantitative ADC and texture features have previously been investigated and found to be significant for diagnosis of cancer using logistic regression and machine‐learning models in both the PZ and TZ . Among features studied, mean ADC, skewness, and entropy were all among the reported significant discriminatory features that could differentiate between clinically significant and low‐risk (Gleason score = 6) cancers, as well as between Gleason score 3 + 4 vs. Gleason score 4 + 3 tumors using SVM and other machine‐learning models .…”

“…Multivariate logistic regression models have found mean ADC, minimum ADC, kurtosis, and other ADC/T 2 W texture features to be significant for differentiating between benign tissue and tumors in both the PZ and TZ . The optimal model in one study by Iyama et al evaluating TZ abnormalities yielded an AUC value of 0.97 (CI 0.94–0.99); however, excluded small (<10 mm) lesions, which are the most clinically challenging.…”

A Triple‐Classification Radiomics Model for the Differentiation of Primary Chordoma, Giant Cell Tumor, and Metastatic Tumor of Sacrum Based on T2‐Weighted and Contrast‐Enhanced T1‐Weighted MRI

“…Usually, radiomics workflow contains five steps; data selection, medical imaging, feature extraction, exploratory analysis, and modeling . Compared with classic methods, radiomics is performed based on advanced pattern recognition tools and involves the extraction of a large number of quantitative features from digital images to determine relationships between such features and the underlying pathophysiology, which has been widely used in many fields, especially in cancer . However, most studies focused on differentiating two kinds of tumors and seldom on triple classification.…”

Influence of temporal parameters of DCE‐MRI on the quantification of heterogeneity in tumor vascularization

“…Consequently, quantifying the spatial heterogeneity in vascularization of the whole tumor volume from DCE‐MRI data may be more informative and realistic. In that sense, radiomics approaches on DCE‐MRI have recently shown encouraging results, alone or with other MRI sequences, in order to improve the detection of prostate cancer, to distinguish benign and malignant adnexal masses, to identify relevant molecular subtypes of breast cancers, to detect lymph node metastases in breast cancers, or to predict response to neoadjuvant treatment for rectum, breast, and nasopharyngeal cancers . Soft‐tissue sarcomas (STS) are malignant mesenchymal tumors with important inter‐ and intratumoral heterogeneity known to be associated with high grade .…”