The Utility of Diffusion Weighted Imaging (DWI) Using Apparent Diffusion Coefficient (ADC) Values in Discriminating Between Prostate Cancer and Normal Tissue

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

INTRODUCTION

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.

METHODS

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 (K^TRANS), efflux rate constant (K_ep), and extracellular volume ratio (V_e) from mpMR images, and SUV_mean and SUV_max 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).

RESULTS

SUV_max 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.

CONCLUSIONS

ML models' performance is dependent on the input combinations and risk factors further improve ML classification accuracy.

Advanced Diffusion-Weighted Imaging Sequences for Breast MRI: Comprehensive Comparison of Improved Sequences and Ultra-High B-Values to Identify the Optimal Combination

This study investigated the image quality and choice of ultra-high b-value of two DWI breast-MRI research applications. The study cohort comprised 40 patients (20 malignant lesions). In addition to s-DWI with two m-b-values (b50 and b800) and three e-b-values (e-b1500, e-b2000, and e-b2500), z-DWI and IR m-b1500 DWI were applied. z-DWI was acquired with the same measured b-values and e-b-values as the standard sequence. For IR m-b1500 DWI, b50 and b1500 were measured, and e-b2000 and e-b2500 were mathematically extrapolated. Three readers used Likert scales to independently analyze all ultra-high b-values (b1500-b2500) for each DWI with regards to scan preference and image quality. ADC values were measured in all 20 lesions. z-DWI was the most preferred (54%), followed by IR m-b1500 DWI (46%). b1500 was significantly preferred over b2000 for z-DWI and IR m-b1500 DWI (p = 0.001 and p = 0.002, respectively). Lesion detection was not significantly different among sequences or b-values (p = 0.174). There were no significant differences in measured ADC values within lesions between s-DWI (ADC: 0.97 [±0.09] × 10^-3 mm²/s) and z-DWI (ADC: 0.99 [±0.11] × 10^-3 mm²/s; p = 1.000). However, there was a trend toward lower values in IR m-b1500 DWI (ADC: 0.80 [±0.06] × 10^-3 mm²/s) than in s-DWI (p = 0.090) and z-DWI (p = 0.110). Overall, image quality was superior and there were fewer image artifacts when using the advanced sequences (z-DWI + IR m-b1500 DWI) compared with s-DWI. Considering scan preferences, we found that the optimal combination was z-DWI with a calculated b1500, especially regarding examination time.

A Comparative Study of Prostate Cancer Detection Rate Between Transperineal COG-TB and Transperineal FUS-TB in Patients with PSA ≤20 ng/mL

Background: The combination of prebiopsy MRI and transperineal targeted biopsies is being increasingly used to obtain tissues from patients with suspected prostate cancer (PCa). Objective: To investigate the difference in PCa detection rate between transperineal cognitive fusion TB (COG-TB) and transperineal software fusion TB (FUS-TB). Participants: The present study included 163 male patients with suspected PCa who had not undergone prostate biopsy, had a prostate-specific antigen (PSA) level of ≤20 ng/mL, and had been examined by bi-parameter MRI and confirmed to have prostate nodules by prostate imaging reporting and data system version 2 (PI-RADS V2) scores ≥3 (from December 3, 2018 to October 7, 2019). Intervention: Seventy-one patients underwent transperineal COG-TB, and 92 patients underwent transperineal FUS-TB. The detection rate of the first four needles was compared. Results: No significant difference was found in the overall detection rate of PCa between COG-TB and FUS-TB (60.56% vs 51.08%, p = 0.228). This result was consistent even after stratifying by PI-RADS score. There was also no significant difference between COG-TB and FUS-TB in the detection rate of clinically significant PCa (p = 0.641). Moreover, COG-TB and FUS-TB showed no difference in the detection rate of PCa with different Gleason scores. Conclusions: In patients with suspected PCa with PSA ≤20 ng/mL and PI-RADS ≥3, FUS-TB was comparable to COG-TB in the detection rate of PCa.

Radiological semantics discriminate clinically significant grade prostate cancer

Background

Identification of imaging traits to discriminate clinically significant prostate cancer is challenging due to the multi focal nature of the disease. The difficulty in obtaining a consensus by the Prostate Imaging and Data Systems (PI-RADS) scores coupled with disagreements in interpreting multi-parametric Magnetic Resonance Imaging (mpMRI) has resulted in increased variability in reporting findings and evaluating the utility of this imaging modality in detecting clinically significant prostate cancer. This study assess the ability of radiological traits (semantics) observed on multi-parametric Magnetic Resonance images (mpMRI) to discriminate clinically significant prostate cancer.

Methods

We obtained multi-parametric MRI studies from 103 prostate cancer patients with 167 targeted biopsies from a single institution. The study was approved by our Institutional Review Board (IRB) for retrospective analysis. The biopsy location had been identified and marked by a clinical radiologist for targeted biopsy based on initial study interpretation. Using the target locations, two study radiologists independently re-evaluated the scans and scored 16 semantic traits on a point scale (up to 5 levels) based on mpMRI images. The semantic traits describe size, shape, and border characteristics of the prostate lesion, as well as presence of disease around lymph nodes (lymphadenopathy). We built a linear classifier model on these semantic traits and related to pathological outcome to identify clinically significant tumors (Gleason Score ≥ 7). The discriminatory ability of the predictors was tested using cross validation method randomly repeated and ensemble values were reported. We then compared the performance of semantic predictors with the PI-RADS predictors.

Results

We found several semantic features individually discriminated high grade Gleason score (ADC-intensity, Homogeneity, early-enhancement, T2-intensity and extraprostatic extention), these univariate predictors had an average area under the receiver operator characteristics (AUROC) ranging from 0.54 to 0.68. Multivariable semantic predictors with three features (ADC-intensity; T2-intensity, enhancement homogenicity) had an average AUROC of 0.7 [0.43, 0.94]. The PI-RADS based predictor had average AUROC of 0.6 [0.47, 0.75].

Conclusion

We find semantics traits are related to pathological findings with relatively higher reproducibility between radiologists. Multivariable predictors formed on these traits shows higher discriminatory ability compared to PI-RADS scores.

Diagnostic efficacy of multiple MRI parameters in differentiating benign vs. malignant thyroid nodules

BACKGROUND

Diffusion weighted imaging (DWI) has a good diagnostic value for malignant thyroid nodules, but the published protocols suffer from flaws and focus on the apparent diffusion coefficient (ADC). This study investigated the diagnostic performance of multiple MRI parameters in differentiating malignant from benign thyroid nodules.

METHODS

This was a retrospective study of 181 consecutive patients (148 benign and 111 malignant nodules, confirmed by pathological results). The patients underwent conventional MRI, DWI, and dynamic contrast-enhanced MRI before surgery. The chi-square test and the Student t test were used to compare the conventional features and ADC value between malignant and benign groups. Multivariate logistic regression was used to identify the independent predictors and to construct a model. Receiver operator characteristic (ROC) curve analysis was used to assess the diagnostic performance of the independent variables and model.

RESULTS

Tumor diameter, ADC value, cystic degeneration, pseudocapsule sign, high signal cystic area on T1-weighted imaging, ring sign in the delayed phase, and irregular shape showed significant differences between two groups (all P < 0.05). The multivariable analysis revealed that ADC value (OR = 694.006, P < 0.001), irregular shape (OR = 32.798, P < 0.001), ring sign in the delayed phase (OR = 20.381, P = 0.004), and cystic degeneration (OR = 8.468, P = 0.016) were independent predictors. Among them, ADC performed the best in discriminating benign from malignant nodules, with an area under the curve (AUC) of 0.95, 0.90 sensitivity, and 0.91 specificity. When the independent factors were combined, the diagnostic performance was improved with an AUC of 0.99, 0.97 sensitivity, and 0.95 specificity.

CONCLUSIONS

ADC value could discriminate between benign and malignant thyroid nodules with a good performance. Subjective features such as the ring sign, irregular shape, and cystic degeneration associated with malignant thyroid nodules could provide complementary information for differentiation.

Role of MRI in the Risk Assessment of Primary Prostate Cancer

A successful paradigm shift toward personalized management strategies for patients with prostate cancer (PCa) is heavily dependent on the availability of noninvasive diagnostic tools capable of accurately establishing the true extent of disease at the time of diagnosis and estimating the risk of subsequent disease progression and related mortality. Although there is still considerable scope for improvement in its diagnostic, predictive, and prognostic capabilities, multiparametric prostate magnetic resonance imaging (MRI) is currently regarded as the imaging modality of choice for local staging of PCa. A negative MRI, that is, the absence of any MRI-visible intraprostatic lesion, has a high negative predictive value for the presence of clinically significant PCa and can substantiate the consideration of active surveillance as a preferred initial management approach. MRI-derived quantitative and semi-quantitative parameters can be utilized to noninvasively characterize MRI-visible prostate lesions and identify those patients who are most likely to benefit from radical treatment, and differentiate them from patients with benign or indolent prostate pathology that may also be visible on MRI. This literature review summarizes current strategies how MRI can be used to determine a tailored management strategy for an individual patient.

A model describing diffusion in prostate cancer

PURPOSE

Quantitative diffusion MRI has frequently been studied as a means of grading prostate cancer. Interpretation of results is complicated by the nature of prostate tissue, which consists of four distinct compartments: vascular, ductal lumen, epithelium, and stroma. Current diffusion measurements are an ill-defined weighted average of these compartments. In this study, prostate diffusion is analyzed in terms of a model that takes explicit account of tissue compartmentalization, exchange effects, and the non-Gaussian behavior of tissue diffusion.

METHOD

The model assumes that exchange between the cellular (ie, stromal plus epithelial) and the vascular and ductal compartments is slow. Ductal and cellular diffusion characteristics are estimated by Monte Carlo simulation and a two-compartment exchange model, respectively. Vascular pseudodiffusion is represented by an additional signal at b = 0. Most model parameters are obtained either from published data or by comparing model predictions with the published results from 41 studies. Model prediction error is estimated using 10-fold cross-validation.

RESULTS

Agreement between model predictions and published results is good. The model satisfactorily explains the variability of ADC estimates found in the literature.

CONCLUSION

A reliable model that predicts the diffusion behavior of benign and cancerous prostate tissue of different Gleason scores has been developed. Magn Reson Med 78:316-326, 2017. © 2016 International Society for Magnetic Resonance in Medicine.

Risk stratification of prostate cancer utilizing apparent diffusion coefficient value and lesion volume on multiparametric MRI

PURPOSE

To evaluate the performance of apparent diffusion coefficient (ADC) and lesion volume in potentially risk-stratifying patients with prostate cancer (PCa).

MATERIALS AND METHODS

Men with elevated prostate-specific antigen or abnormal digital rectal exam underwent a 3T multiparametric magnetic resonance imaging (mpMRI) with endorectal coil. ADC maps were calculated using b values of 0, 500, 1000, and 1500; additional images were obtained with b value of 2000. We prospectively enrolled 312 men with lesions suspicious for cancer (suspicion score 2-5) on mpMRI. MRI/ultrasound fusion-guided prostate biopsies were performed. Mean ADC of suspicious lesions were correlated against lesion volume, Gleason and D'Amico risk.

RESULTS

The cancer detection rate of fusion biopsy per lesion was 45.6% (206/452). Cancerous lesions were larger (median volume: 0.40 vs. 0.30 cm³ ; P = 0.016). The median ADC (×10^-6 mm² /sec) for lesions negative and positive for PCa were 984.5 and 666.5, respectively (P < 0.0001). The AUC of ADC in predicting PCa was 0.79. Larger lesions were associated with higher risk PCa (Gleason and D'Amico) and lower ADC (all P < 0.0001).

CONCLUSION

The mean ADC of suspicious lesions on mpMRI was inversely correlated, while lesion volume had a direct correlation with PCa detection. Future follow-up studies are needed to assess longitudinal cancer risks of suspicious mpMRI lesions.

LEVEL OF EVIDENCE

2 J. Magn. Reson. Imaging 2017;45:610-616.

Current role of multiparametric magnetic resonance imaging in the management of prostate cancer

The purpose of this review was to evaluate the current role of multiparametric magnetic resonance imaging (mp-MRI) in the management of prostate cancer (PC). The diagnosis of PC remains controversial owing to overdetection of indolent disease, which leads to overtreatment and subsequent patient harm. mp-MRI has the potential to equilibrate the imbalance between detection and treatment. The limitation of the data for analysis with this new technology is problematic, however. This issue has been compounded by a paradigm shift in clinical practice aimed at utilizing this modality, which has been rolled out in an ad hoc fashion often with commercial motivation. Despite a growing body of literature, pertinent clinical questions remain. For example, can mp-MRI be calibrated to reliably detect biologically significant disease? As with any new technology, objective evaluation of the clinical applications of mp-MRI is essential. The focus of this review was on the evaluation of mp-MRI of the prostate with respect to clinical utility.