Diagnosis of Prostate Cancer by Use of MRI-Derived Quantitative Risk Maps: A Feasibility Study

Quantitative Prostate MRI, From the AJR Special Series on Quantitative Imaging

Prostate MRI has traditionally relied on qualitative interpretation. However, quantitative components hold the potential to markedly improve performance. The ADC from DWI is probably the most widely recognized quantitative MRI biomarker and has shown strong discriminatory value for clinically significant prostate cancer (csPCa) as well as for recurrent cancer after treatment. Advanced diffusion techniques, including intravoxel incoherent motion, diffusion kurtosis, diffusion tensor imaging, and specific implementations such as restriction spectrum imaging, purport even better discrimination, but are more technically challenging. The inherent T1 and T2 of tissue also provide diagnostic value, with more advanced techniques deriving luminal water imaging and hybrid-multidimensional MRI. Dynamic contrast-enhanced imaging, primarily using a modified Tofts model, also shows independent discriminatory value. Finally, quantitative size and shape features can be combined with the aforementioned techniques and be further refined using radiomics, texture analysis, and artificial intelligence. Which technique will ultimately find widespread clinical use will depend on validation across a myriad of platforms use-cases.

Is There an Added Value of Quantitative DCE-MRI by Magnetic Resonance Dispersion Imaging for Prostate Cancer Diagnosis?

In this multicenter, retrospective study, we evaluated the added value of magnetic resonance dispersion imaging (MRDI) to standard multiparametric MRI (mpMRI) for PCa detection. The study included 76 patients, including 51 with clinically significant prostate cancer (csPCa), who underwent radical prostatectomy and had an mpMRI including dynamic contrast-enhanced MRI. Two radiologists performed three separate randomized scorings based on mpMRI, MRDI and mpMRI+MRDI. Radical prostatectomy histopathology was used as the reference standard. Imaging and histopathology were both scored according to the Prostate Imaging-Reporting and Data System V2.0 sector map. Sensitivity and specificity for PCa detection were evaluated for mpMRI, MRDI and mpMRI+MRDI. Inter- and intra-observer variability for both radiologists was evaluated using Cohen's Kappa. On a per-patient level, sensitivity for csPCa for radiologist 1 (R1) for mpMRI, MRDI and mpMRI+MRDI was 0.94, 0.82 and 0.94, respectively. For the second radiologist (R2), these were 0.78, 0.94 and 0.96. R1 detected 4% additional csPCa cases using MRDI compared to mpMRI, and R2 detected 20% extra csPCa cases using MRDI. Inter-observer agreement was significant only for MRDI (Cohen's Kappa = 0.4250, p = 0.004). The results of this study show the potential of MRDI to improve inter-observer variability and the detection of csPCa.

A radiomics based method for prediction of prostate cancer Gleason score using enlarged region of interest

BACKGROUND

Prostate cancer (PCa) is one of the most common cancers in men worldwide, and its timely diagnosis and treatment are becoming increasingly important. MRI is in increasing use to diagnose cancer and to distinguish between non-clinically significant and clinically significant PCa, leading to more precise diagnosis and treatment. The purpose of this study is to present a radiomics-based method for determining the Gleason score (GS) for PCa using tumour heterogeneity on multiparametric MRI (mp-MRI).

METHODS

Twenty-six patients with biopsy-proven PCa were included in this study. The quantitative T2 values, apparent diffusion coefficient (ADC) and signal enhancement rates (α) were calculated using multi-echo T2 images, diffusion-weighted imaging (DWI) and dynamic contrast-enhanced MRI (DCE-MRI), for the annotated region of interests (ROI). After texture feature analysis, ROI range expansion and feature filtering was performed. Then obtained data were put into support vector machine (SVM), K-Nearest Neighbor (KNN) and other classifiers for binary classification.

RESULTS

The highest classification accuracy was 73.96% for distinguishing between clinically significant (Gleason 3 + 4 and above) and non-significant cancers (Gleason 3 + 3) and 83.72% for distinguishing between Gleason 3 + 4 from Gleason 4 + 3 and above, which was achieved using initial ROIs drawn by the radiologists. The accuracy improved when using expanded ROIs to 80.67% using SVM and 88.42% using Bayesian classification for distinguishing between clinically significant and non-significant cancers and Gleason 3 + 4 from Gleason 4 + 3 and above, respectively.

CONCLUSIONS

Our results indicate the research significance and value of this study for determining the GS for prostate cancer using the expansion of the ROI region.

Parametric maps of spatial two-tissue compartment model for prostate dynamic contrast enhanced MRI - comparison with the standard tofts model in the diagnosis of prostate cancer

The spatial two-tissue compartment model (2TCM) was used to analyze prostate dynamic contrast enhanced (DCE) MRI data and compared with the standard Tofts model. A total of 29 patients with biopsy-confirmed prostate cancer were included in this IRB-approved study. MRI data were acquired on a Philips Achieva 3T-TX scanner. After T2-weighted and diffusion-weighted imaging, DCE data using 3D T1-FFE mDIXON sequence were acquired pre- and post-contrast media injection (0.1 mmol/kg Multihance) for 60 dynamic scans with temporal resolution of 8.3 s/image. The 2TCM has one fast ([Formula: see text] and [Formula: see text]) and one slow ([Formula: see text] and [Formula: see text]) exchanging compartment, compared with the standard Tofts model parameters (Ktrans and kep). On average, prostate cancer had significantly higher values (p < 0.01) than normal prostate tissue for all calculated parameters. There was a strong correlation (r = 0.94, p < 0.001) between Ktrans and [Formula: see text] for cancer, but weak correlation (r = 0.28, p < 0.05) between kep and [Formula: see text]. Average root-mean-square error (RMSE) in fits from the 2TCM was significantly smaller (p < 0.001) than the RMSE in fits from the Tofts model. Receiver operating characteristic (ROC) analysis showed that fast [Formula: see text] had the highest area under the curve (AUC) than any other individual parameter. The combined four parameters from the 2TCM had a considerably higher AUC value than the combined two parameters from the Tofts model. The 2TCM is useful for quantitative analysis of prostate DCE-MRI data and provides new information in the diagnosis of prostate cancer.

Parametric maps of spatial two-tissue compartment model for prostate dynamic contrast enhanced MRI - comparison with the standard Tofts model in the diagnosis of prostate cancer

The spatial two-tissue compartment model (2TCM) was used to analyze prostate dynamic contrast enhanced (DCE) MRI data and compared with the standard Tofts model. A total of 29 patients with biopsy-confirmed prostate cancer were included in this IRB-approved study. MRI data were acquired on a Philips Achieva 3T-TX scanner. After T2-weighted and diffusion-weighted imaging, DCE data using 3D T1-FFE mDIXON sequence were acquired pre- and post-contrast media injection (0.1 mmol/kg Multihance) for 60 dynamic scans with temporal resolution of 8.3 s/image. The 2TCM has one fast (K 1 trans and k 1 ep ) and one slow (K 2 trans and k 2 ep ) exchanging compartment, compared with the standard Tofts model parameters (K trans and k ep ). On average, prostate cancer had significantly higher values (p < 0.007) than normal prostate tissue for all calculated parameters. There was a strong correlation (r = 0.94, p < 0.0001) between K trans and K 1 trans for cancer, but weak correlation (r = 0.28, p < 0.05) between k ep and k 1 ep . Average root-mean-square error (RMSE) in fits from the 2TCM was significantly smaller (p < 0.001) than the RMSE in fits from the Tofts model. Receiver operating characteristic (ROC) analysis showed that fast K 1 trans had the highest area under the curve (AUC) than any other individual parameter. The combined four parameters from the 2TCM had a considerably higher AUC value than the combined two parameters from the Tofts model. The 2TCM may be useful for quantitative analysis of prostate DCE-MRI data and may provide new information in the diagnosis of prostate cancer.

Dual-Tracer PET-MRI-Derived Imaging Biomarkers for Prediction of Clinically Significant Prostate Cancer

PURPOSE

To investigate if imaging biomarkers derived from 3-Tesla dual-tracer [(18)F]fluoromethylcholine (FMC) and [⁶⁸Ga]Ga-PSMA^HBED-CC conjugate 11 (PSMA)-positron emission tomography can adequately predict clinically significant prostate cancer (csPC).

METHODS

We assessed 77 biopsy-proven PC patients who underwent 3T dual-tracer PET/mpMRI followed by radical prostatectomy (RP) between 2014 and 2017. We performed a retrospective lesion-based analysis of all cancer foci and compared it to whole-mount histopathology of the RP specimen. The primary aim was to investigate the pretherapeutic role of the imaging biomarkers FMC- and PSMA-maximum standardized uptake values (SUVmax) for the prediction of csPC and to compare it to the mpMRI-methods and PI-RADS score.

RESULTS

Overall, we identified 104 cancer foci, 69 were clinically significant (66.3%) and 35 were clinically insignificant (33.7%). We found that the combined FMC+PSMA SUVmax were the only significant parameters (p < 0.001 and p = 0.049) for the prediction of csPC. ROC analysis showed an AUC for the prediction of csPC of 0.695 for PI-RADS scoring (95% CI 0.591 to 0.786), 0.792 for FMC SUVmax (95% CI 0.696 to 0.869), 0.852 for FMC+PSMA SUVmax (95% CI 0.764 to 0.917), and 0.852 for the multivariable CHAID model (95% CI 0.763 to 0.916). Comparing the AUCs, we found that FMC+PSMA SUVmax and the multivariable model were significantly more accurate for the prediction of csPC compared to PI-RADS scoring (p = 0.0123, p = 0.0253, respectively).

CONCLUSIONS

Combined FMC+PSMA SUVmax seems to be a reliable parameter for the prediction of csPC and might overcome the limitations of PI-RADS scoring. Further prospective studies are necessary to confirm these promising preliminary results.

Comparing Radiologist Performance in Diagnosing Clinically Significant Prostate Cancer with Multiparametric versus Hybrid Multidimensional MRI

Background Variability of acquisition and interpretation of prostate multiparametric MRI (mpMRI) persists despite implementation of the Prostate Imaging Reporting and Data System (PI-RADS) version 2.1 due to the range of reader experience and subjectivity of lesion characterization. A quantitative method, hybrid multidimensional MRI (HM-MRI), may introduce objectivity. Purpose To compare performance, interobserver agreement, and interpretation time of radiologists using mpMRI versus HM-MRI to diagnose clinically significant prostate cancer. Materials and Methods In this retrospective analysis, men with prostatectomy or MRI-fused transrectal US biopsy-confirmed prostate cancer underwent mpMRI (triplanar T2-weighted, diffusion-weighted, and dynamic contrast-enhanced imaging) and HM-MRI (with multiple echo times and b value combinations) from August 2012 to February 2020. Four readers with 1-20 years of experience interpreted mpMRI and HM-MRI examinations independently, with a 4-week washout period between interpretations. PI-RADS score, lesion location, and interpretation time were recorded. mpMRI and HM-MRI interpretation time, interobserver agreement (Cronbach alpha), and performance of area under the receiver operating characteristic curve (AUC) analysis were compared for each radiologist with use of bootstrap analysis. Results Sixty-one men (mean age, 61 years ± 8 [SD]) were evaluated. Per-patient AUC was higher for HM-MRI for reader 4 compared with mpMRI (AUCs for readers 1-4: 0.61, 0.71, 0.59, and 0.64 vs 0.66, 0.60, 0.50, and 0.46; P = .57, .20, .32, and .04, respectively). Per-patient specificity was higher for HM-MRI for readers 2-4 compared with mpMRI (specificity for readers 1-4: 48%, 78%, 48%, and 46% vs 37%, 26%, 0%, and 7%; P = .34, P < .001, P < .001, and P < .001, respectively). Diagnostic performance improved for the reader least experienced with HM-MRI, reader 4 (AUC, 0.64 vs 0.46; P = .04). HM-MRI interobserver agreement (Cronbach alpha = 0.88 [95% CI: 0.82, 0.92]) was higher than that of mpMRI (Cronbach alpha = 0.26 [95% CI: 0.10, 0.52]; α > .60 indicates reliability; P = .03). HM-MRI mean interpretation time (73 seconds ± 43 [SD]) was shorter than that of mpMRI (254 seconds ± 133; P = .03). Conclusion Radiologists had similar or improved diagnostic performance, higher interobserver agreement, and lower interpretation time for clinically significant prostate cancer with hybrid multidimensional MRI than multiparametric MRI. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Turkbey in this issue.

MR fingerprinting of the prostate

Multiparametric magnetic resonance imaging (mpMRI) has been adopted as the key tool for detection, localization, characterization, and risk stratification of patients suspected to have prostate cancer. Despite advantages over systematic biopsy, the interpretation of prostate mpMRI has limitations including a steep learning curve, leading to considerable interobserver variation. There is growing interest in clinical translation of quantitative imaging techniques for more objective lesion assessment. However, traditional mapping techniques are slow, precluding their use in the clinic. Magnetic resonance fingerprinting (MRF) is an efficient approach for quantitative maps of multiple tissue properties simultaneously. The T1 and T2 values obtained with MRF have been validated with phantom studies as well as in normal volunteers and patients. Studies have shown that MRF-derived T1 and T2 along with ADC values are all significant independent predictors in the differentiation between normal prostate tissue and prostate cancer, and hold promise in differentiating low and intermediate/high-grade cancers. This review seeks to introduce the basics of the prostate MRF technique, discuss the potential applications of prostate MRF for the characterization of prostate cancer, and describes ongoing areas of research.

Effectiveness of Dynamic Contrast Enhanced MRI with a Split Dose of Gadoterate Meglumine for Detection of Prostate Cancer

RATIONALE AND OBJECTIVES

To evaluate whether dynamic contrast enhanced (DCE) MRI with a split injection of 30% followed by 70% of a standard dose (30PSD and 70PSD) of gadoterate meglumine (DOTAREM) can improve diagnosis of prostate cancer (PCa).

MATERIALS AND METHODS

MRI for twenty patients was performed on a Philips Ingenia 3T scanner without an endorectal coil followed by subsequent radical prostatectomy. DCE 3D T1-FFE data were acquired with injection of 0.03 mmol/kg followed after 2 minutes by 0.07 mmol/kg of DOTAREM. Regions-of-interest on histologically verified PCa and normal tissue in different prostate zones and the iliac artery were drawn. Average signal intensity as function of time was calculated for each ROI and fitted by using the signal intensity form of the Tofts (SI-Tofts) model to extract physiological parameters (K^trans and v_e). In addition, the scaled arterial input function (AIF) obtained from 30PSD data was used to analyze 70PSD data.

RESULTS

The AIF obtained from 30PSD data showed both first and second passes clearly and had much higher peak magnitude than AIFs from 70PSD data. K^trans was significantly (p < 0.05) larger in PCa than in normal tissue in peripheral zone (PZ) and central zone (CZ) for both 70PSD and 70PSD data analyzed with a scaled AIF. K^trans in cancer overlapped with that of normal tissue in the transition zone (TZ). There was no statistical difference in v_e between cancer and normal tissue. Receiver operating characteristic analysis showed that use of the AIF from 30PSD data to analyze 70PSD data increased the diagnostic efficacy of K^trans in the PZ and CZ.

CONCLUSION

The split dose protocol for injection of Dotarem increased diagnostic accuracy of quantitative analysis with the SI-Tofts model.

Lesion segmentation in breast ultrasound images using the optimized marked watershed method

BACKGROUND

Breast cancer is one of the most serious diseases threatening women's health. Early screening based on ultrasound can help to detect and treat tumours in the early stage. However, due to the lack of radiologists with professional skills, ultrasound-based breast cancer screening has not been widely used in rural areas. Computer-aided diagnosis (CAD) technology can effectively alleviate this problem. Since breast ultrasound (BUS) images have low resolution and speckle noise, lesion segmentation, which is an important step in CAD systems, is challenging.

RESULTS

Two datasets were used for evaluation. Dataset A comprises 500 BUS images from local hospitals, while dataset B comprises 205 open-source BUS images. The experimental results show that the proposed method outperformed its related classic segmentation methods and the state-of-the-art deep learning model RDAU-NET. Its accuracy (Acc), Dice similarity coefficient (DSC) and Jaccard index (JI) reached 96.25%, 78.4% and 65.34% on dataset A, and its Acc, DSC and sensitivity reached 97.96%, 86.25% and 88.79% on dataset B, respectively.

CONCLUSIONS

We proposed an adaptive morphological snake based on marked watershed (AMSMW) algorithm for BUS image segmentation. It was proven to be robust, efficient and effective. In addition, it was found to be more sensitive to malignant lesions than benign lesions.

METHODS

The proposed method consists of two steps. In the first step, contrast limited adaptive histogram equalization (CLAHE) and a side window filter (SWF) are used to preprocess BUS images. Lesion contours can be effectively highlighted, and the influence of noise can be eliminated to a great extent. In the second step, we propose adaptive morphological snake (AMS). It can adjust the working parameters adaptively according to the size of the lesion. Its segmentation results are combined with those of the morphological method. Then, we determine the marked area and obtain candidate contours with a marked watershed (MW). Finally, the best lesion contour is chosen by the maximum average radial derivative (ARD).

Nakagami imaging with related distributions for advanced thermogram pseudocolorization

Pseudocoloring algorithms embedded in the software of thermal cameras gradually colorize original intensity thermograms generated by detecting temperatures and contrast. Maximum and minimum based algorithms, however, executed by thresholding, applied to intensity thermograms for revealing and coloring the outliers instead. Although the common pseudocoloring protocols employed for general purposes may provide crucial information on the superficial contrast between radiation emitted by various sources; their common kernel is not sufficient for detecting and differentiating high radiated regions from surrounding areas, which is mandatory for recognition of abnormalities. Therefore, we propose novel imaging methodology based on Nakagami and related distributions, including gamma, Rayleigh, Weibull, chi-square and exponential, for enhancing thermal images and also for creating adequate discrimination. We initially define the boundaries of tumor and surrounding area in a synthetically generated breast thermogram already diagnosed as retroareolar tumor. Using Nakagami and transformations supported by mathematical foundations, we conducted several experiments to find the discrimination factor of the pseudocoloring techniques by calculating difference of average contrast between the tumor and the surrounding area. The performance is greatly encouraging that we achieved considerably better discrimination factor, designated for this study, up to 106.80 compared to the results of existing built-in pseudocolorization methods computed as 11.56.

Multi-slice representational learning of convolutional neural network for Alzheimer's disease classification using positron emission tomography

BACKGROUND

Alzheimer's Disease (AD) is a degenerative brain disorder that often occurs in people over 65 years old. As advanced AD is difficult to manage, accurate diagnosis of the disorder is critical. Previous studies have revealed effective deep learning methods of classification. However, deep learning methods require a large number of image datasets. Moreover, medical images are affected by various environmental factors. In the current study, we propose a deep learning-based method for diagnosis of Alzheimer's disease (AD) that is less sensitive to different datasets for external validation, based upon F-18 fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT).

RESULTS

The accuracy, sensitivity, and specificity of our proposed network were 86.09%, 80.00%, and 92.96% (respectively) using our dataset, and 91.02%, 87.93%, and 93.57% (respectively) using the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We observed that our model classified AD and normal cognitive (NC) cases based on the posterior cingulate cortex (PCC), where pathological changes occur in AD. The performance of the GAP layer was considered statistically significant compared to the fully connected layer in both datasets for accuracy, sensitivity, and specificity (p < 0.01). In addition, performance comparison between the ADNI dataset and our dataset showed no statistically significant differences in accuracy, sensitivity, and specificity (p > 0.05).

CONCLUSIONS

The proposed model demonstrated the effectiveness of AD classification using the GAP layer. Our model learned the AD features from PCC in both the ADNI and Severance datasets, which can be seen in the heatmap. Furthermore, we showed that there were no significant differences in performance using statistical analysis.

Elaboration of a multimodal MRI-based radiomics signature for the preoperative prediction of the histological subtype in patients with non-small-cell lung cancer

BACKGROUND

Non-invasive discrimination between lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD) subtypes of non-small-cell lung cancer (NSCLC) could be very beneficial to the patients unfit for the invasive diagnostic procedures. The aim of this study was to investigate the feasibility of utilizing the multimodal magnetic resonance imaging (MRI) radiomics and clinical features in classifying NSCLC. This retrospective study involved 148 eligible patients with postoperative pathologically confirmed NSCLC. The study was conducted in three steps: (1) feature extraction was performed using the online freely available package with the multimodal MRI data; (2) feature selection was performed using the Student's t test and support vector machine (SVM)-based recursive feature elimination method with the training cohort (n = 100), and the performance of these selected features was evaluated using both the training and the validation cohorts (n = 48) with a non-linear SVM classifier; (3) a Radscore model was then generated using logistic regression algorithm; (4) Integrating the Radscore with the semantic clinical features, a radiomics-clinical nomogram was developed, and its overall performance was evaluated with both cohorts.

RESULTS

Thirteen optimal features achieved favorable discrimination performance with both cohorts, with area under the curve (AUC) of 0.819 and 0.824, respectively. The radiomics-clinical nomogram integrating the Radscore with the independent clinical predictors exhibited more favorable discriminative power, with AUC improved to 0.901 and 0.872 in both cohorts, respectively. The Hosmer-Lemeshow test and decision curve analysis results furtherly showed good predictive precision and clinical usefulness of the nomogram.

CONCLUSION

Non-invasive histological subtype stratification of NSCLC can be done favorably using multimodal MRI radiomics features. Integrating the radiomics features with the clinical features could further improve the performance of the histological subtype stratification in patients with NSCLC.

Measurement and application of patient similarity in personalized predictive modeling based on electronic medical records

Background

Conventional risk prediction techniques may not be the most suitable approach for personalized prediction for individual patients. Therefore, individualized predictive modeling based on similar patients has emerged. This study aimed to propose a comprehensive measurement of patient similarity using real-world electronic medical records data, and evaluate the effectiveness of the individualized prediction of a patient’s diabetes status based on the patient similarity.

Results

When using no more than 30% of the whole training sample, the personalized predictive models outperformed corresponding traditional models built on randomly selected training samples of the same size as the personalized models (P < 0.001 for all). With only the top 1000 (10%), 700 (7%) and 1400 (14%) similar samples, personalized random forest, k-nearest neighbor and logistic regression models reached the globally optimal performance with the area under the receiver-operating characteristic (ROC) curve of 0.90, 0.82 and 0.89, respectively.

Conclusions

The proposed patient similarity measurement was effective when developing personalized predictive models. The successful application of patient similarity in predicting a patient’s diabetes status provided useful references for diagnostic decision-making support by investigating the evidence on similar patients.