Dynamic magnetic resonance imaging of tumor perfusion

Structural and practical identifiability of contrast transport models for DCE-MRI

Contrast transport models are widely used to quantify blood flow and transport in dynamic contrast-enhanced magnetic resonance imaging. These models analyze the time course of the contrast agent concentration, providing diagnostic and prognostic value for many biological systems. Thus, ensuring accuracy and repeatability of the model parameter estimation is a fundamental concern. In this work, we analyze the structural and practical identifiability of a class of nested compartment models pervasively used in analysis of MRI data. We combine artificial and real data to study the role of noise in model parameter estimation. We observe that although all the models are structurally identifiable, practical identifiability strongly depends on the data characteristics. We analyze the impact of increasing data noise on parameter identifiability and show how the latter can be recovered with increased data quality. To complete the analysis, we show that the results do not depend on specific tissue characteristics or the type of enhancement patterns of contrast agent signal.

Structural and practical identifiability of contrast transport models for DCE-MRI

Compartment models are widely used to quantify blood flow and transport in dynamic contrast-enhanced magnetic resonance imaging. These models analyze the time course of the contrast agent concentration, providing diagnostic and prognostic value for many biological systems. Thus, ensuring accuracy and repeatability of the model parameter estimation is a fundamental concern. In this work, we analyze the structural and practical identifiability of a class of nested compartment models pervasively used in analysis of MRI data. We combine artificial and real data to study the role of noise in model parameter estimation. We observe that although all the models are structurally identifiable, practical identifiability strongly depends on the data characteristics. We analyze the impact of increasing data noise on parameter identifiability and show how the latter can be recovered with increased data quality. To complete the analysis, we show that the results do not depend on specific tissue characteristics or the type of enhancement patterns of contrast agent signal.

Characterization of breast cancer subtypes based on quantitative assessment of intratumoral heterogeneity using dynamic contrast-enhanced and diffusion-weighted magnetic resonance imaging

OBJECTIVE

To investigate whether intratumoral heterogeneity, assessed via dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and diffusion-weighted imaging (DWI), reflects the molecular subtypes of invasive breast cancers.

MATERIAL AND METHODS

We retrospectively evaluated data from 248 consecutive women (mean age ± standard deviation, 54.6 ± 12.2 years) with invasive breast cancer who underwent preoperative DCE-MRI and DWI between 2019 and 2020. To evaluate intratumoral heterogeneity, kinetic heterogeneity (a measure of heterogeneity in the proportions of tumor pixels with delayed washout, plateau, and persistent components within a tumor) was assessed with DCE-MRI using a commercially available computer-aided diagnosis system. Apparent diffusion coefficients (ADCs) were obtained using a region-of-interest technique, and ADC heterogeneity was calculated using the following formula: (ADC_max-ADC_min)/ADC_mean. Possible associations between imaging-based heterogeneity values and breast cancer subtypes were analyzed.

RESULTS

Of the 248 invasive breast cancers, 61 (24.6%) were classified as luminal A, 130 (52.4%) as luminal B, 25 (10.1%) as HER2-enriched, and 32 (12.9%) as triple-negative breast cancer (TNBC). There were significant differences in the kinetic and ADC heterogeneity values among tumor subtypes (p < 0.001 and p = 0.023, respectively). The TNBC showed higher kinetic and ADC heterogeneity values, whereas the HER2-enriched subtype showed higher kinetic heterogeneity values compared to the luminal subtypes. Multivariate linear analysis showed that the HER2-enriched (p < 0.001) and TNBC subtypes (p < 0.001) were significantly associated with higher kinetic heterogeneity values. The TNBC subtype (p = 0.042) was also significantly associated with higher ADC heterogeneity values.

CONCLUSIONS

Quantitative assessments of heterogeneity in enhancement kinetics and ADC values may provide biological clues regarding the molecular subtypes of breast cancer.

KEY POINTS

• Higher kinetic heterogeneity was associated with HER2-enriched and triple-negative breast cancer. • Higher ADC heterogeneity was associated with triple-negative breast cancer. • Aggressive breast cancer subtypes exhibited higher intratumoral heterogeneity based on MRI.

Texture Analysis Using Semiquantitative Kinetic Parameter Maps from DCE-MRI: Preoperative Prediction of HER2 Status in Breast Cancer

Objective

To evaluate whether texture features derived from semiquantitative kinetic parameter maps based on breast dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) can determine human epidermal growth factor receptor 2 (HER2) status of patients with breast cancer.

Materials and Methods

This study included 102 patients with histologically confirmed breast cancer, all of whom underwent preoperative breast DCE-MRI and were enrolled retrospectively. This cohort included 48 HER2-positive cases and 54 HER2-negative cases. Seven semiquantitative kinetic parameter maps were calculated on the lesion area. A total of 55 texture features were extracted from each kinetic parameter map. Patients were randomly divided into training (n = 72) and test (n = 30) sets. The least absolute shrinkage and selection operator (LASSO) was used to select features in the training set, and then, multivariate logistic regression analysis was conducted to establish the prediction models. The classification performance was evaluated by receiver operating characteristic (ROC) analysis.

Results

Among the seven prediction models, the model with features extracted from the early signal enhancement ratio (ESER) map yielded an area under the ROC curve (AUC) of 0.83 in the training set (sensitivity of 70.59%, specificity of 92.11%, and accuracy of 81.94%), and the highest AUC of 0.83 in the test set (sensitivity of 57.14%, specificity of 100.00%, and accuracy of 80.00%). The model with features extracted from the slope of signal intensity (SI_slope) map yielded the highest AUC of 0.92 in the training set (sensitivity of 82.35%, specificity of 97.37%, and accuracy of 90.28%), and an AUC of 0.79 in the test set (sensitivity of 92.86%, specificity of 68.75%, and accuracy of 80.00%).

Conclusions

Texture features derived from kinetic parameter maps, calculated based on breast DCE-MRI, have the potential to be used as imaging biomarkers to distinguish HER2-positive and HER2-negative breast cancer.

Convolutional neural network for accelerating the computation of the extended Tofts model in dynamic contrast-enhanced magnetic resonance imaging

Quantitative physiological parameters can be obtained from nonlinear pharmacokinetic models, such as the extended Tofts (eTofts) model, applied to dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). However, the computation of such nonlinear models is time consuming. The aim of this study was to develop a convolutional neural network (CNN) for accelerating the computation of fitting eTofts model without sacrificing agreement with conventional nonlinear-least-square (NLLS) fitting. This was a retrospective study, which included 13 patients with brain glioma for training (75%) and validation (25%), and 11 patients (three glioma, four brain metastases, and four lymphoma) for testing. CAIPIRINHA-Dixon-TWIST DCE-MRI and double flip angle T₁ map acquired at 3 T were used. A CNN with both local pathway and global pathway modules was designed to estimate the eTofts model parameters, the volume transfer constant (K^trans ), blood volume fraction (v_p ), and volume fraction of extracellular extravascular space (v_e ), from DCE-MRI data of tumor and normal-appearing voxels. The CNN was trained on mixed dataset consisting of synthetic and patient data. The CNN result and computation speed were compared with NLLS fitting. The robustness to noise variations and generalization to brain metastases and lymphoma data were also evaluated. Statistical tests used were Student's t test on mean absolute error, concordance correlation coefficient (CCC), and normalized root mean squared error. Including global pathway modules in the CNN and training the network with mixed data significantly (p < 0.05) improved the CNN performance. Compared with NLLS fitting, CNN yields an average CCC greater than 0.986 for K^trans , greater than 0.965 for v_p , and greater than 0.948 for v_e . The CNN accelerated computation speed approximately 2000 times compared to NLLS, showed robustness to noise (signal-to-noise ratio >34.42 dB), and had no significant (p > 0.21) difference applied to brain metastases and lymphoma data. In conclusion, the proposed CNN to estimate eTofts parameters showed comparable result as NLLS fitting while significantly reducing the computation time. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 1.

Pre-operative Assessment of Ablation Margins for Variable Blood Perfusion Metrics in a Magnetic Resonance Imaging Based Complex Breast Tumour Anatomy: Simulation Paradigms in Thermal Therapies

BACKGROUND AND OBJECTIVES

Image-guided medical interventions facilitates precise visualization at treatment site. The conformal prediction for sparing healthy tissue fringes precisely in the vicinity of irregular tumour anatomy remains clinically challenging. Pre-clinical image-based computational modelling is imperative as it helps in enhancement of treatment quality, augmenting clinical-decision making, while planning, targeting, controlling, monitoring and assessing treatment response with an effective risk assessment before the onset of treatment in clinical settings. In this study, the influence of heat deposition rate (SAR), exposure duration, and variable blood perfusion metrics for a patient-specific breast tumour is quantified considering the tumour margins thereby suggesting need of geometrically accurate models.

METHODS

A three-dimensional realistic model mimicking dimensions of a female breast, comprising ~1.7 cm irregular tumour, was generated from patient specific two-dimensional DICOM format MRI images through image segmentation tools MIMICS 19.0® and 3-Matic 11.0® which is finally exported to COMSOL Multiphysics 5.2® as a volumetric mesh for finite element analysis. The Pennes bioheat transfer model and Arrhenius thermal damage model of cell-death are integrated to simulate a coupled biophysics problem. A comparative blood perfusion analysis is done to evaluate the response of tumour during heating considering thermal damage extent, including the tumour margins while sparing critical adjoining healthy tissues.

RESULTS

The evaluated thermal damage zones for 1 mm, 2 mm and 3 mm fringe heating region (beyond tumour boundary) reveals 0.09%, 0.21% and 0.34% thermal damage to the healthy tissue (which is <1%) and thus successful necrosis of the tumour. The iterative computational experiments suggests treatment margins < 5 mm are sufficient enough as heating beyond 3 mm fringe layer leads to higher damage surrounding the tumour approximately 1.5 times the tumour volume. Further, the heat-dosage requirements are 22% more for highly perfused tumour as compared to moderately perfused tumour with an approximate double time to ablate the whole tumour volume.

CONCLUSIONS

Depending on the blood perfusion characteristics of a tumour, it is a trade-off between heat-dosage (SAR) and exposure/treatment duration to get desired thermal damage including the irregular tumour boundaries while taking into account, the margin of healthy tissue. The suggested patient-specific integrated multiphysics-model based on MRI-Images may be implemented for pre-treatment planning based on the tumour blood perfusion to evaluate the thermal ablation zone dimensions clinically and thereby avoiding the damage of off-target tissues. Thus, risks involving underestimation or overestimation of thermal coagulation zones may be minimised while preserving the surrounding normal breast parenchyma.

Current status and future prospective of focal therapy for localized prostate cancer: development of multiparametric MRI, MRI-TRUS fusion image-guided biopsy, and treatment modalities

Multiparametric magnetic resonance imaging (mpMRI) has been increasingly used to diagnose clinically significant prostate cancer (csPC) because of its usefulness in combination with anatomic and functional data. MRI-targeted biopsy, such as MRI-transrectal ultrasound (TRUS) fusion image-guided prostate biopsy, has high accuracy in the detection and localization of csPC. This novel diagnostic technique contributes to the development of tailor-made medicine as focal therapy, which cures the csPC while preserving the anatomical structures related to urinary and sexual function. In the early days of focal therapy, TRUS-guided systematic biopsy was used for patient selection, and treatment was performed for patients with low-risk PC. With the introduction of mpMRI and mapping biopsy, the treatment range is now determined based on individualized cancer localization. In recent prospective studies, 87.4% of treated patients had intermediate- and high-risk PC. However, focal therapy has two main limitations. First, a randomized controlled trial would be difficult to design because of the differences in pathological features between patients undergoing focal therapy and radical treatment. Therefore, pair-matched studies and/or historical controlled studies have been performed to compare focal therapy and radical treatment. Second, no long-term (≥ 10-year) follow-up study has been performed. However, recent prospective studies have encouraged the use of focal therapy as a treatment strategy for localized PC because it contributes to high preservation of continence and erectile function.

Role of intravoxel incoherent motion parameters in gastroesophageal cancer: relationship with 18F-FDG-positron emission tomography, computed tomography perfusion and magnetic resonance perfusion imaging parameters

BACKGROUND

Identification of pretherapeutic predictive markers in gastro-esophageal cancer is essential for individual-oriented treatment. This study evaluated the relationship of multimodality parameters derived from intravoxel incoherent motion method (IVIM), 18F-FDG-positron emission tomography (PET), computed tomography (CT) perfusion and dynamic contrast enhanced magnetic resonance imaging (MRI) in patients with gastro-esophageal cancer and investigated their histopathological correlation.

METHODS

Thirty-one consecutive patients (28 males; median age 63.9 years; range 37-84 years) with gastro-esophageal adenocarcinoma (N.=22) and esophageal squamous cell carcinoma (N.=9) were analyzed. IVIM parameters: pseudodiffusion (D*), perfusion fraction (fp), true diffusion (D) and the threshold b-value (bval); PET-parameters: SUV<inf>max</inf>, metabolic tumor volume (MTV) and total lesion glycolysis (TLG); CT perfusion parameters: blood flow (BF), blood volume (BV) and mean transit time (MTT); and MR perfusion parameters: time to enhance, positive enhancement integral, time-to-peak (TTP), maximum-slope-of-increase, and maximum-slope-of-decrease were determined, and correlated to each other and to histopathology.

RESULTS

IVIM and PET parameters showed significant negative correlations: MTV and bval (r<inf>s</inf> =-0.643, P=0.002), TLG and bval (r<inf>s</inf>=-0.699, P<0.01) and TLG and fp (r<inf>s</inf>=-0.577, P=0.006). Positive correlation was found for TLG and D (r<inf>s</inf>=0.705, P=0.000). Negative correlation was found for bval and staging (r<inf>s</inf>=0.590, P=0.005). Positive correlation was found for positive enhancement interval and BV (r<inf>s</inf>=0.547, P=0.007), BF and regression index (r<inf>s</inf>=0.753, P=0.005) and for time-to-peak and staging (r<inf>s</inf>=0.557, P=0.005).

CONCLUSIONS

IVIM parameters (bval, fp, D) provide quantitative information and correlate with PET parameters (MTV, TLG) and staging. IVIM might be a useful tool for additional characterization of gastro-esophageal cancer.

Perfusion-sensitive parameters of intravoxel incoherent motion MRI in rectal cancer: evaluation of reproducibility and correlation with dynamic contrast-enhanced MRI

BACKGROUND

Intravoxel incoherent motion magnetic resonance imaging (IVIM-MRI) acquires tumor perfusion information without injection of contrast medium, which is promising in tumor assessment. However, its consistency with dynamic contrast-enhanced MRI (DCE-MRI), a more widely used method for tumor perfusion evaluation, is not revealed in rectal cancer.

PURPOSE

In this study, we aimed to investigate the correlation of perfusion-sensitive parameters derived from IVIM-MRI with DCE-MRI and measurement reproducibility of IVIM-MRI parameters in rectal cancer.

MATERIAL AND METHODS

Forty-seven rectal cancer patients underwent IVIM-MRI with 16 b-values and DCE-MRI. The perfusion fraction ( f), pseudo-diffusion coefficient ( D*), and f· D* were measured by two radiologists independently and correlated with the transfer constant ( K^trans), reflux constant ( k_ep), and extravascular extracellular fractional volume ( v_e) obtained from DCE-MRI.

RESULTS

Pearson's correlation analyses of IVIM-MRI and DCE-MRI parameters showed fair to moderate correlation between f and K^trans ( r = 0.461, P = 0.001), followed by f and k_ep ( r = 0.430, P = 0.003), f·D*, and K^trans ( r = 0.425, P = 0.003), f·D*, and k_ep ( r = 0.384, P = 0.008). There was no significant correlation between v_e and f, v_e and D*, v_e and f· D*, D* and K^trans, and D* and k_ep. The reproducibility of IVIM-MRI measurements was moderate. For parameter f, intraclass correlation coefficient (ICC) = 0.71 (0.53-0.82), coefficient of variation (CV) = 13.05 ± 0.02%, limit of agreement (LoA) = -0.05-0.04; for parameter D*, ICC = 0.55 (0.32-0.72), CV = 20.28 ± 3.23%, LoA = -9.6-8.4.

CONCLUSION

Perfusion-sensitive parameters derived from IVIM-MRI correlated fairly to moderately with DCE-MRI in rectal cancer patients and showed moderate measurement reproducibility. IVIM-MRI supplements routine high-resolution MRI without contrast enhancement to provide information of tumor microcirculation.

Convolutional Neural Networks for Direct Inference of Pharmacokinetic Parameters: Application to Stroke Dynamic Contrast-Enhanced MRI

Background and Purpose: The T1-weighted dynamic contrast enhanced (DCE)-MRI is an imaging technique that provides a quantitative measure of pharmacokinetic (PK) parameters characterizing microvasculature of tissues. For the present study, we propose a new machine learning (ML) based approach to directly estimate the PK parameters from the acquired DCE-MRI image-time series that is both more robust and faster than conventional model fitting.

Materials and Methods: We specifically utilize deep convolutional neural networks (CNNs) to learn the mapping between the image-time series and corresponding PK parameters. DCE-MRI datasets acquired from 15 patients with clinically evident mild ischaemic stroke were used in the experiments. Training and testing were carried out based on leave-one-patient-out cross- validation. The parameter estimates obtained by the proposed CNN model were compared against the two tracer kinetic models: (1) Patlak model, (2) Extended Tofts model, where the estimation of model parameters is done via voxelwise linear and nonlinear least squares fitting respectively.

Results: The trained CNN model is able to yield PK parameters which can better discriminate different brain tissues, including stroke regions. The results also demonstrate that the model generalizes well to new cases even if a subject specific arterial input function (AIF) is not available for the new data.

Conclusion: A ML-based model can be used for direct inference of the PK parameters from DCE image series. This method may allow fast and robust parameter inference in population DCE studies. Parameter inference on a 3D volume-time series takes only a few seconds on a GPU machine, which is significantly faster compared to conventional non-linear least squares fitting.

Magnetic resonance imaging-transrectal ultrasound fusion image-guided prostate biopsy: Current status of the cancer detection and the prospects of tailor-made medicine of the prostate cancer

Multi-parametric magnetic resonance imaging (mpMRI) has been increasingly used to diagnose clinically significant prostate cancer (csPCa) because of its growing availability and its ability to combine anatomical and functional data. Magnetic resonance imaging (MRI)-transrectal ultrasound (TRUS) fusion imaging provides MRI information with TRUS images for prostate biopsies. This technique combines the superior sensitivity of MRI for targeting suspicious lesions with the practicality and familiarity of TRUS. MRI-TRUS fusion image-guided prostate biopsy is performed with different types of image registration (rigid vs. elastic) and needle tracking methods (electromagnetic tracking vs. mechanical position encoders vs. image-based software tracking). A systematic review and meta-analysis showed that MRI-targeted biopsy detected csPCa at a significantly higher rate than did TRUS-guided biopsy, while it detected significantly fewer cases of insignificant PCas. In addition to the high accuracy of MRI-targeted biopsy for csPCa, localization of csPCa is accurate. The ability to choose the route of biopsy (transperineal vs. transrectal) is required, depending on the patients' risk and the location and size of suspicious lesions on mpMRI. Fusion image-guided prostate biopsy has the potential to allow precise management of prostate cancer, including active surveillance, radical treatment, and focal therapy.

In vitro pharmacokinetic phantom for two-compartment modeling in DCE-MRI

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is an established minimally-invasive method for assessment of extravascular leakage, hemodynamics, and tissue viability. However, differences in acquisition protocols, variety of pharmacokinetic models, and uncertainty on physical sources of MR signal hamper the reliability and widespread use of DCE-MRI in clinical practice. Measurements performed in a controlled in vitro setup could be used as a basis for standardization of the acquisition procedure, as well as objective evaluation and comparison of pharmacokinetic models. In this paper, we present a novel flow phantom that mimics a two-compartmental (blood plasma and extravascular extracellular space/EES) vascular bed, enabling systemic validation of acquisition protocols. The phantom consisted of a hemodialysis filter with two compartments, separated by hollow fiber membranes. The aim of this phantom was to vary the extravasation rate by adjusting the flow in the two compartments. Contrast agent transport kinetics within the phantom was interpreted using two-compartmental pharmacokinetic models. Boluses of gadolinium-based contrast-agent were injected in a tube network connected to the hollow fiber phantom; time-intensity curves (TICs) were obtained from image series, acquired using a T1-weighted DCE-MRI sequence. Under the assumption of a linear dilution system, the TICs obtained from the input and output of the system were then analyzed by a system identification approach to estimate the trans-membrane extravasation rates in different flow conditions. To this end, model-based deconvolution was employed to determine (identify) the impulse response of the investigated dilution system. The flow rates in the EES compartment significantly and consistently influenced the estimated extravasation rates, in line with the expected trends based on simulation results. The proposed phantom can therefore be used to model a two-compartmental vascular bed and can be employed to test and optimize DCE-MRI acquisition sequences in order to determine a standardized acquisition procedure leading to consistent quantification results.

Multiparametric magnetic resonance imaging: Overview of the technique, clinical applications in prostate biopsy and future directions

Multiparametric magnetic resonance imaging (mpMRI) has managed to change the paradigms on prostate cancer detection and risk classification. The most clear-cut indication of mpMRI in guidelines is the patients with a history of negative biopsy/increasing prostate-specific antigen (PSA), and presence of additional findings supporting its use in non biopsied patients and active surveillance. mpMRI complements standard clinical exam, PSA measurements, and systematic biopsy, and will miss some tumors that lack enough size or change in tissue density. Use of mpMRI is likely to increase, and further developments in the technique will be important for safe adoption of focal therapy concepts. Here we present a brief summary about mpMRI and its use in detection, risk classification and follow-up of prostate cancer.

Self-Gated Late Gadolinium Enhancement at 7T to Image Rats with Reperfused Acute Myocardial Infarction

Objective

A failed electrocardiography (ECG)-trigger often leads to a long acquisition time (TA) and deterioration in image quality. The purpose of this study was to evaluate and optimize the technique of self-gated (SG) cardiovascular magnetic resonance (CMR) for cardiac late gadolinium enhancement (LGE) imaging of rats with myocardial infarction/reperfusion.

Materials and Methods

Cardiovascular magnetic resonance images of 10 rats were obtained using SG-LGE or ECG with respiration double-gating (ECG-RESP-gating) method at 7T to compare differences in image interference and TA between the two methods. A variety of flip angles (FA: 10°-80°) and the number of repetitions (NR: 40, 80, 150, and 300) were investigated to determine optimal scan parameters of SG-LGE technique based on image quality score and contrast-to-noise ratio (CNR).

Results

Self-gated late gadolinium enhancement allowed successful scan in 10 (100%) rats. However, only 4 (40%) rats were successfully scanned with the ECG-RESP-gating method. TAs with SG-LGE varied depending on NR used (TA: 41, 82, 154, and 307 seconds, corresponding to NR of 40, 80, 150, and 300, respectively). For the ECG-RESP-gating method, the average TA was 220 seconds. For SG-LGE images, CNR (42.5 ± 5.5, 43.5 ± 7.5, 54 ± 9, 59.5 ± 8.5, 56 ± 13, 54 ± 8, and 41 ± 9) and image quality score (1.85 ± 0.75, 2.20 ± 0.83, 2.85 ± 0.37, 3.85 ± 0.52, 2.8 ± 0.51, 2.45 ± 0.76, and 1.95 ± 0.60) were achieved with different FAs (10°, 15°, 20°, 25°, 30°, 35°, and 40°, respectively). Optimal FAs of 20°-30° and NR of 80 were recommended.

Conclusion

Self-gated technique can improve image quality of LGE without irregular ECG or respiration gating. Therefore, SG-LGE can be used an alternative method of ECG-RESP-gating.

Large-scale retrieval for medical image analytics: A comprehensive review

Over the past decades, medical image analytics was greatly facilitated by the explosion of digital imaging techniques, where huge amounts of medical images were produced with ever-increasing quality and diversity. However, conventional methods for analyzing medical images have achieved limited success, as they are not capable to tackle the huge amount of image data. In this paper, we review state-of-the-art approaches for large-scale medical image analysis, which are mainly based on recent advances in computer vision, machine learning and information retrieval. Specifically, we first present the general pipeline of large-scale retrieval, summarize the challenges/opportunities of medical image analytics on a large-scale. Then, we provide a comprehensive review of algorithms and techniques relevant to major processes in the pipeline, including feature representation, feature indexing, searching, etc. On the basis of existing work, we introduce the evaluation protocols and multiple applications of large-scale medical image retrieval, with a variety of exploratory and diagnostic scenarios. Finally, we discuss future directions of large-scale retrieval, which can further improve the performance of medical image analysis.

Multiparametric MRI in the Detection of Clinically Significant Prostate Cancer

Prostate cancer is the most common cancer among men aged 50 years and older in developed countries and the third leading cause of cancer-related death in men. Multiparametric prostate MR imaging is currently the most accurate imaging modality to detect, localize, and stage prostate cancer. The role of multi-parametric MR imaging in the detection of clinically significant prostate cancer are discussed. In addition, insights are provided in imaging techniques, protocol, and interpretation.

A comprehensive non-invasive framework for diagnosing prostate cancer

Early detection of prostate cancer increases chances of patients' survival. Our automated non-invasive system for computer-aided diagnosis (CAD) of prostate cancer segments the prostate on diffusion-weighted magnetic resonance images (DW-MRI) acquired at different b-values, estimates its apparent diffusion coefficients (ADC), and classifies their descriptors - empirical cumulative distribution functions (CDF) - with a trained deep learning network. To segment the prostate, an evolving geometric (level-set-based) deformable model is guided by a speed function depending on intensity attributes extracted from the DW-MRI with nonnegative matrix factorization (NMF). For a more robust evolution, the attributes are fused with a probabilistic shape prior and estimated spatial dependencies between prostate voxels. To preserve continuity, the ADCs of the segmented prostate volume at different b-values are normalized and refined using a generalized Gauss-Markov random field image model. The CDFs of the refined ADCs at different b-values are considered global water diffusion features and used to distinguish between benign and malignant prostates. A deep learning network of stacked non-negativity-constrained auto-encoders (SNCAE) is trained to classify the benign or malignant prostates on the basis of the constructed CDFs. Our experiments on 53 clinical DW-MRI data sets resulted in 92.3% accuracy, 83.3% sensitivity, and 100% specificity, indicating that the proposed CAD system could be used as a reliable non-invasive diagnostic tool.

Artificial Neural Networks approach to pharmacokinetic model selection in DCE-MRI studies

PURPOSE

In pharmacokinetic analysis of Dynamic Contrast Enhanced MRI data, a descriptive physiological model should be selected properly out of a set of candidate models. Classical techniques suggested for this purpose suffer from issues like computation time and general fitting problems. This article proposes an approach based on Artificial Neural Networks (ANNs) for solving these problems.

METHODS

A set of three physiologically and mathematically nested models generated from the Tofts model were assumed: Model I, II and III. These models cover three possible tissue types from normal to malignant. Using 21 experimental arterial input functions and 12 levels of noise, a set of 27,216 time traces were generated. ANN was validated and optimized by the k-fold cross validation technique. An experimental dataset of 20 patients with glioblastoma was applied to ANN and the results were compared to outputs of F-test using Dice index.

RESULTS

Optimum neuronal architecture ([6:7:1]) and number of training epochs (50) of the ANN were determined. ANN correctly classified more than 99% of the dataset. Confusion matrices for both ANN and F-test results showed the superior performance of the ANN classifier. The average Dice index (over 20 patients) indicated a 75% similarity between model selection maps of ANN and F-test.

CONCLUSIONS

ANN improves the model selection process by removing the need for time-consuming, problematic fitting algorithms; as well as the need for hypothesis testing.

Dynamic contrast-enhanced MRI of gastric cancer: Correlations of the pharmacokinetic parameters with histological type, Lauren classification, and angiogenesis

PURPOSE

To compare the pharmacokinetic parameters derived from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in gastric cancers of different histological type and Lauren classification, and to investigate whether DCE-MRI parameters correlate with vascular endothelial growth factor (VEGF) expression levels in gastric cancer.

METHODS

Included were 32 patients with gastric cancer who underwent DCE-MRI of the upper abdomen before tumor resection. DCE-MRI parameters including the volume transfer coefficient (K^trans), reverse reflux rate constant (K_ep), and extracellular extravascular volume fraction (V_e) were calculated from the tumor region. Post-operative specimens were used for determination of histological differentiation (i.e., non-mucinous, mucinous, or signet-ring-cell adenocarcinoma) as well as Lauren classification (intestinal type or diffuse type). VEGF expression was examined for assessing angiogenesis. DCE-MRI parameters with different histological type and Lauren classification were compared using independent samples t-test and analysis of variance, respectively. Correlations between DCE-MRI parameters and VEGF expression grades were tested using Spearman correlation analysis.

RESULTS

Among gastric adenocarcinomas of three different histological types, mucinous adenocarcinomas showed a higher V_e and lower K^trans than the others (P<0.01). Between the two Lauren classifications, the diffuse type showed a higher V_e than the intestinal type (P<0.001). The mean K^trans showed a significantly positive correlation with VEGF (r=0.762, P<0.001).

CONCLUSION

DCE-MRI permits noninvasive prediction of tumor histological type and Lauren classification and estimation of tumor angiogenesis in gastric cancer. DCE-MRI parameters can be used as imaging biomarkers to predict the biologic aggressiveness of a tumor as well as patient prognosis.

Liver DCE-MRI Registration in Manifold Space Based on Robust Principal Component Analysis

A technical challenge in the registration of dynamic contrast-enhanced magnetic resonance (DCE-MR) imaging in the liver is intensity variations caused by contrast agents. Such variations lead to the failure of the traditional intensity-based registration method. To address this problem, a manifold-based registration framework for liver DCE-MR time series is proposed. We assume that liver DCE-MR time series are located on a low-dimensional manifold and determine intrinsic similarities between frames. Based on the obtained manifold, the large deformation of two dissimilar images can be decomposed into a series of small deformations between adjacent images on the manifold through gradual deformation of each frame to the template image along the geodesic path. Furthermore, manifold construction is important in automating the selection of the template image, which is an approximation of the geodesic mean. Robust principal component analysis is performed to separate motion components from intensity changes induced by contrast agents; the components caused by motion are used to guide registration in eliminating the effect of contrast enhancement. Visual inspection and quantitative assessment are further performed on clinical dataset registration. Experiments show that the proposed method effectively reduces movements while preserving the topology of contrast-enhancing structures and provides improved registration performance.

Mathematical Models of Contrast Transport Kinetics for Cancer Diagnostic Imaging: A Review

Angiogenesis plays a fundamental role in cancer growth and the formation of metastasis. Novel cancer therapies aimed at inhibiting angiogenic processes and/or disrupting angiogenic tumor vasculature are currently being developed and clinically tested. The need for earlier and improved cancer diagnosis, and for early evaluation and monitoring of therapeutic response to angiogenic treatment, have led to the development of several imaging methods for in vivo noninvasive assessment of angiogenesis. The combination of dynamic contrast-enhanced imaging with mathematical modeling of the contrast agent kinetics enables quantitative assessment of the structural and functional changes in the microvasculature that are associated with tumor angiogenesis. In this paper, we review quantitative imaging of angiogenesis with dynamic contrast-enhanced magnetic resonance imaging, computed tomography, and ultrasound.

Quantitative correlational study of microbubble-enhanced ultrasound imaging and magnetic resonance imaging of glioma and early response to radiotherapy in a rat model

PURPOSE

Radiotherapy remains a major treatment method for malignant tumors. Magnetic resonance imaging (MRI) is the standard modality for assessing glioma treatment response in the clinic. Compared to MRI, ultrasound imaging is low-cost and portable and can be used during intraoperative procedures. The purpose of this study was to quantitatively compare contrast-enhanced ultrasound (CEUS) imaging and MRI of irradiated gliomas in rats and to determine which quantitative ultrasound imaging parameters can be used for the assessment of early response to radiation in glioma.

METHODS

Thirteen nude rats with U87 glioma were used. A small thinned skull window preparation was performed to facilitate ultrasound imaging and mimic intraoperative procedures. Both CEUS and MRI with structural, functional, and molecular imaging parameters were performed at preradiation and at 1 day and 4 days postradiation. Statistical analysis was performed to determine the correlations between MRI and CEUS parameters and the changes between pre- and postradiation imaging.

RESULTS

Area under the curve (AUC) in CEUS showed significant difference between preradiation and 4 days postradiation, along with four MRI parameters, T2, apparent diffusion coefficient, cerebral blood flow, and amide proton transfer-weighted (APTw) (all p < 0.05). The APTw signal was correlated with three CEUS parameters, rise time (r = - 0.527, p < 0.05), time to peak (r = - 0.501, p < 0.05), and perfusion index (r = 458, p < 0.05). Cerebral blood flow was correlated with rise time (r = - 0.589, p < 0.01) and time to peak (r = - 0.543, p < 0.05).

CONCLUSIONS

MRI can be used for the assessment of radiotherapy treatment response and CEUS with AUC as a new technique and can also be one of the assessment methods for early response to radiation in glioma.

Magnetic resonance-ultrasound fusion prostate biopsy in the diagnosis of prostate cancer

The advent of multiparametric magnetic resonance imaging (MRI) has ushered in a new era for urologists who perform prostate needle biopsies. The fusion of MRI with transrectal ultrasound (US) allows the direct targeting of suspicious lesions, which has been shown to improve the performance of conventional random biopsy techniques by increasing detection of clinically relevant disease while also decreasing detection of low-risk cancer. However, as with any new technology, many questions regarding effectiveness, reproducibility, and generalizability still remain. In this review, we (1) provide a summary of the various sequences that comprise a MRI of the prostate; (2) evaluate the 3 different ways of incorporating MRI into targeted biopsies of the prostate including in-bore MRI-guided biopsy, cognitive fusion, and device-mediated fusion; (3) review the sensitivity of MR-US fusion in the detection of clinically significant and clinically insignificant disease; and (4) review the barriers to the widespread implementation of MR-US fusion into everyday practice. Whereas other articles in this issue of Urologic Oncology Seminars will discuss other aspects of MRI in the management of prostate cancer, the purpose of this article is to provide an overview of MR-US fusion biopsies in the diagnosis of prostate cancer.

The differential effects of metronomic gemcitabine and antiangiogenic treatment in patient-derived xenografts of pancreatic cancer: treatment effects on metabolism, vascular function, cell proliferation, and tumor growth

BACKGROUND

Metronomic chemotherapy has shown promising activity against solid tumors and is believed to act in an antiangiogenic manner. The current study describes and quantifies the therapeutic efficacy, and mode of activity, of metronomic gemcitabine and a dedicated antiangiogenic agent (DC101) in patient-derived xenografts of pancreatic cancer.

METHODS

Two primary human pancreatic cancer xenograft lines were dosed metronomically with gemcitabine or DC101 weekly. Changes in tumor growth, vascular function, and metabolism over time were measured with magnetic resonance imaging, positron emission tomography, and immunofluorescence microscopy to determine the anti-tumor effects of the respective treatments.

RESULTS

Tumors treated with metronomic gemcitabine were 10-fold smaller than those in the control and DC101 groups. Metronomic gemcitabine, but not DC101, reduced the tumors' avidity for glucose, proliferation, and apoptosis. Metronomic gemcitabine-treated tumors had higher perfusion rates and uniformly distributed blood flow within the tumor, whereas perfusion rates in DC101-treated tumors were lower and confined to the periphery. DC101 treatment reduced the tumor's vascular density, but did not change their function. In contrast, metronomic gemcitabine increased vessel density, improved tumor perfusion transiently, and decreased hypoxia.

CONCLUSION

The aggregate data suggest that metronomic gemcitabine treatment affects both tumor vasculature and tumor cells continuously, and the overall effect is to significantly slow tumor growth. The observed increase in tumor perfusion induced by metronomic gemcitabine may be used as a therapeutic window for the administration of a second drug or radiation therapy. Non-invasive imaging could be used to detect early changes in tumor physiology before reductions in tumor volume were evident.

Quantification of breast tumor heterogeneity for ER status, HER2 status, and TN molecular subtype evaluation on DCE-MRI

PURPOSE

Recognizing molecular markers is helpful for guiding treatment plans for breast cancer. This study correlated estrogen receptor (ER), human epidermal growth factor receptor 2 (HER2), and triple-negative breast cancer (TNBC) statuses to the degree of heterogeneity on breast dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI).

MATERIALS AND METHODS

A total of 102 biopsy-proven cancers from 102 patients between October 2010 and December 2012 were used in this study, including ER (59 positive, 43 negative), HER2 (47 positive, 55 negative), and TNBC (22 TNBC, 80 non-TNBC). At first, the tumor region was segmented by using a region growing method. Then, the region-based features were extracted by the proposed regionalization method to quantify intra-tumoral heterogeneity on breast DCE-MRI. The three-dimensional morphological features (texture features and shape feature) and the pharmacokinetic model were also extracted from the segmented tumor region. After feature extraction, a logistic regression was used to classify ER, HER2, and TNBC statuses respectively. The performances were evaluated by using receiver operating characteristic (ROC) curve analysis.

RESULTS

The proposed region-based features achieved the accuracy of 73.53%, 82.35%, and 77.45% for ER, HER2, and TNBC classifications. The corresponding area under the ROC curves (Az) achieves 0.7320, 0.8458, and 0.8328 that were better than those of texture features, shape features, and Tofts pharmacokinetic model.

CONCLUSION

The intra-tumoral heterogeneity quantified by the region-based features can be used to reflect the vasculature complexity of different molecular markers and to provide prediction information of cell surface receptors on clinical examination.

The current and future role of magnetic resonance imaging in prostate cancer detection and management

Purpose

Accurate detection of clinically significant prostate cancer (PC) and correct risk attribution are essential to individually counsel men with PC. Multiparametric MRI (mpMRI) facilitates correct localization of index lesions within the prostate and MRI-targeted prostate biopsy (TPB) helps to avoid the shortcomings of conventional biopsy such as false-negative results or underdiagnosis of aggressive PC. In this review we summarize the different sequences of mpMRI, characterize the possibilities of incorporating MRI in the biopsy workflow and outline the performance of targeted and systematic cores in significant cancer detection. Furthermore, we outline the potential of MRI in patients undergoing active surveillance (AS) and in the pre-operative setting.

Materials and methods

An electronic MEDLINE/PubMed search up to February 2015 was performed. English language articles were reviewed for inclusion ability and data were extracted, analyzed and summarized.

Results

Targeted biopsies significantly outperform conventional systematic biopsies in the detection of significant PC and are not inferior when compared to transperineal saturation biopsies. MpMRI can detect index lesions in app. 90% of cases as compared to prostatectomy specimen. The diagnostic performance of biparametric MRI (T2w + DWI) is not inferior to mpMRI, offering options to diminish cost- and time-consumption. Since app 10% of significant lesions are still MRI-invisible, systematic cores seem to be necessary. In-bore biopsy and MRI/TRUS-fusion-guided biopsy tend to be superior techniques compared to cognitive fusion. In AS, mpMRI avoids underdetection of significant PC and confirms low-risk disease accurately. In higher-risk disease, pre-surgical MRI can change the clinically-based surgical plan in up to a third of cases.

Conclusions

mpMRI and targeted biopsies are able to detect significant PC accurately and mitigate insignificant PC detection. As long as the negative predictive value (NPV) is still imperfect, systematic cores should not be omitted for optimal staging of disease. The potential to correctly classify aggressiveness of disease in AS patients and to guide and plan prostatectomy is evolving.

High-spatial and high-temporal resolution dynamic contrast-enhanced perfusion imaging of the liver with time-resolved three-dimensional radial MRI

PURPOSE

Detection, characterization, and monitoring the treatment of hepatocellular carcinomas (HCC) in patients with cirrhosis is challenging because of their variable and rapid arterial enhancement. Multiphase dynamic contrast-enhanced MRI is used clinically for HCC assessment; however, the method suffers from limited temporal resolution and difficulty in coordinating imaging and breath-hold timing within a narrow temporal window of interest. In this article, a volumetric, high-spatial resolution, and high-temporal resolution dynamic contrast-enhanced liver imaging method for improved detection and characterization of HCC is demonstrated.

METHODS

A time-resolved three-dimensional radial acquisition with iterative sensitivity-encoding reconstruction images the entire abdomen and thorax with high spatial and temporal resolution, using real-time three-dimensional fluoroscopy to match the breath hold to contrast arrival. The sequence was tested on 17 subjects, including eight patients with HCC or other hypervascular focal lesions.

RESULTS

This technique was successful in acquiring volumetric imaging of the entire liver with 2.1-mm isotropic spatial and true 4-s temporal resolution.

CONCLUSION

This technique may be suitable for detecting, characterizing, and monitoring the treatment of HCC. It also holds significant potential for perfusion modeling, which may provide a noninvasive means to rapidly determine the efficacy of chemotherapeutic agents in these tumors over the entire liver volume.

Integration of DCE-MRI and DW-MRI Quantitative Parameters for Breast Lesion Classification

OBJECTIVE

The purpose of our study was to evaluate the diagnostic value of an imaging protocol combining dynamic contrast-enhanced MRI (DCE-MRI) and diffusion-weighted MRI (DW-MRI) in patients with suspicious breast lesions.

MATERIALS AND METHODS

A total of 31 breast lesions (15 malignant and 16 benign proved by histological examination) in 26 female patients were included in this study. For both DCE-MRI and DW-MRI model free and model based parameters were computed pixel by pixel on manually segmented ROIs. Statistical procedures included conventional linear analysis and more advanced techniques for classification of lesions in benign and malignant.

RESULTS

Our findings indicated no strong correlation between DCE-MRI and DW-MRI parameters. Results of classification analysis show that combining of DCE parameters or DW-MRI parameter, in comparison of single feature, does not yield a dramatic improvement of sensitivity and specificity of the two techniques alone. The best performance was obtained considering a full combination of all features. Moreover, the classification results combining all features are dominated by DCE-MRI features alone.

CONCLUSION

The combination of DWI and DCE-MRI does not show a potential to dramatically increase the sensitivity and specificity of breast MRI. DCE-MRI alone gave the same performance as in combination with DW-MRI.

Distributed capillary adiabatic tissue homogeneity model in parametric multi-channel blind AIF estimation using DCE-MRI

PURPOSE

One of the main challenges in quantitative dynamic contrast-enhanced (DCE) MRI is estimation of the arterial input function (AIF). Usually, the signal from a single artery (ignoring contrast dispersion, partial volume effects and flow artifacts) or a population average of such signals (also ignoring variability between patients) is used.

METHODS

Multi-channel blind deconvolution is an alternative approach avoiding most of these problems. The AIF is estimated directly from the measured tracer concentration curves in several tissues. This contribution extends the published methods of multi-channel blind deconvolution by applying a more realistic model of the impulse residue function, the distributed capillary adiabatic tissue homogeneity model (DCATH). In addition, an alternative AIF model is used and several AIF-scaling methods are tested.

RESULTS

The proposed method is evaluated on synthetic data with respect to the number of tissue regions and to the signal-to-noise ratio. Evaluation on clinical data (renal cell carcinoma patients before and after the beginning of the treatment) gave consistent results. An initial evaluation on clinical data indicates more reliable and less noise sensitive perfusion parameter estimates.

CONCLUSION

Blind multi-channel deconvolution using the DCATH model might be a method of choice for AIF estimation in a clinical setup.

State-of-the-art MRI techniques in neuroradiology: principles, pitfalls, and clinical applications

This article reviews the most relevant state-of-the-art magnetic resonance (MR) techniques, which are clinically available to investigate brain diseases. MR acquisition techniques addressed include notably diffusion imaging (diffusion-weighted imaging (DWI), diffusion tensor imaging (DTI), and diffusion kurtosis imaging (DKI)) as well as perfusion imaging (dynamic susceptibility contrast (DSC), arterial spin labeling (ASL), and dynamic contrast enhanced (DCE)). The underlying models used to process these images are described, as well as the theoretic underpinnings of quantitative diffusion and perfusion MR imaging-based methods. The technical requirements and how they may help to understand, classify, or follow-up neurological pathologies are briefly summarized. Techniques, principles, advantages but also intrinsic limitations, typical artifacts, and alternative solutions developed to overcome them are discussed. In this article, we also review routinely available three-dimensional (3D) techniques in neuro MRI, including state-of-the-art and emerging angiography sequences, and briefly introduce more recently proposed 3D quantitative neuro-anatomy sequences, and new technology, such as multi-slice and multi-transmit imaging.

Advanced magnetic resonance imaging of the physical processes in human glioblastoma

The most common malignant primary brain tumor, glioblastoma multiforme (GBM) is a devastating disease with a grim prognosis. Patient survival is typically less than two years and fewer than 10% of patients survive more than five years. Magnetic resonance imaging (MRI) can have great utility in the diagnosis, grading, and management of patients with GBM as many of the physical manifestations of the pathologic processes in GBM can be visualized and quantified using MRI. Newer MRI techniques such as dynamic contrast enhanced and dynamic susceptibility contrast MRI provide functional information about the tumor hemodynamic status. Diffusion MRI can shed light on tumor cellularity and the disruption of white matter tracts in the proximity of tumors. MR spectroscopy can be used to study new tumor tissue markers such as IDH mutations. MRI is helping to noninvasively explore the link between the molecular basis of gliomas and the imaging characteristics of their physical processes. We, here, review several approaches to MR-based imaging and discuss the potential for these techniques to quantify the physical processes in glioblastoma, including tumor cellularity and vascularity, metabolite expression, and patterns of tumor growth and recurrence. We conclude with challenges and opportunities for further research in applying physical principles to better understand the biologic process in this deadly disease. See all articles in this Cancer Research section, "Physics in Cancer Research."

Use of diffusion-weighted, intravoxel incoherent motion, and dynamic contrast-enhanced MR imaging in the assessment of response to radiotherapy of lytic bone metastases from breast cancer

RATIONALE AND OBJECTIVES

To investigate the value of diffusion-weighted (DW), perfusion-sensitive, and dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) techniques in assessing the response of bone metastases from breast cancer to radiotherapy, with particular emphasis on the role of intravoxel incoherent motion (IVIM)-DW parameters as a potential valuable imaging marker of tumor response.

MATERIALS AND METHODS

Fifteen women having breast cancer and bone metastases underwent MRI before and after radiotherapy (3 weeks [time 1], 2 months [time 2], and 4 months [time 3]), consisting of DW, perfusion-sensitive (IVIM), and DCE acquisitions. MR-based DW and perfusion parameters, including water diffusivity (D), perfusion fraction (f), pseudodiffusion (D*), total apparent diffusion coefficient (ADC-total), fractionated ADCs (ADC-high and ADC-low), and initial area under the gadolinium concentration curve after the first 60 seconds (IAUGC60), were determined. The morphologic MRI findings were also recorded. A one-way repeated measures analysis of variance was used to compare the value of MR-based parameters at the different time points.

RESULTS

A significant variation between pretreatment (time 0) and post-treatment (times 1, 2, and 3) was found for ADC-total and D parameters (P < .001). A statistically significant reduction was also found for IAUGC60 values between times 0 and 3 (P < .001). A significant change across the different time points was observed for D* and IAUGC60 parameters (P < .001). On the contrary, there was no statistically significant change over time for parameters ADC-total, D, f, and IAUGC60 comparing response between each metastasis, that is, the response to therapy was similar for each metastasis.

CONCLUSIONS

DW, IVIM, and DCE-MRI techniques show effectiveness in assessing the response to radiotherapy in bone metastases from breast cancer.

Comprehensive population-averaged arterial input function for dynamic contrast-enhanced vmagnetic resonance imaging of head and neck cancer

PURPOSE

To generate a population-averaged arterial input function (PA-AIF) for quantitative analysis of dynamic contrast-enhanced MRI data in head and neck cancer patients.

METHODS AND MATERIALS

Twenty patients underwent dynamic contrast-enhanced MRI during concurrent chemoradiation therapy. Imaging consisted of 2 baseline scans 1 week apart (B1/B2) and 1 scan after 1 week of chemoradiation therapy (Wk1). Regions of interest (ROIs) in the right and left carotid arteries were drawn on coronal images. Plasma concentration curves of all ROIs were averaged and fit to a biexponential decay function to obtain the final PA-AIF (AvgAll). Right-sided and left-sided ROI plasma concentration curves were averaged separately to obtain side-specific AIFs (AvgRight/AvgLeft). Regions of interest were divided by time point to obtain time-point-specific AIFs (AvgB1/AvgB2/AvgWk1). The vascular transfer constant (Ktrans) and the fractional extravascular, extracellular space volume (Ve) for primaries and nodes were calculated using the AvgAll AIF, the appropriate side-specific AIF, and the appropriate time-point-specific AIF. Median Ktrans and Ve values derived from AvgAll were compared with those obtained from the side-specific and time-point-specific AIFs. The effect of using individual AIFs was also investigated.

RESULTS

The plasma parameters for AvgAll were a1,2 = 27.11/17.65 kg/L, m1,2 = 11.75/0.21 min(-1). The coefficients of repeatability (CRs) for AvgAll versus AvgLeft were 0.04 min(-1) for Ktrans and 0.02 for Ve. For AvgAll versus AvgRight, the CRs were 0.08 min(-1) for Ktrans and 0.02 for Ve. When AvgAll was compared with AvgB1/AvgB2/AvgWk1, the CRs were slightly higher: 0.32/0.19/0.78 min(-1), respectively, for Ktrans; and 0.07/0.08/0.09 for Ve. Use of a PA-AIF was not significantly different from use of individual AIFs.

CONCLUSION

A PA-AIF for head and neck cancer was generated that accounts for differences in right carotid artery versus left carotid artery, day-to-day fluctuations, and early treatment-induced changes. The small CRs obtained for Ktrans and Ve indicate that side-specific AIFs are not necessary. However, a time-point-specific AIF may improve pharmacokinetic accuracy.

Effect of acute hyperglycemia on moderately hypothermic GL261 mouse glioma monitored by T1-weighted DCE MRI

OBJECTIVE

We sought to evaluate the effects of acute hyperglycemia induced by intraperitoneal injection of glucose (2.7 g/kg) on vascular delivery to GL261 mouse gliomas kept at moderate hypothermia (~30 °C).

MATERIALS AND METHODS

Seven GL261 glioma-bearing mice were studied by T1-weighted DCE MRI before and after an injection of glucose (n = 4) or saline (n = 3). Maximum relative contrast enhancement (RCE) and initial area under the enhancement curve (IAUC) were determined in each pixel.

RESULTS

The mean tumor parameter values showed no significant changes after injecting either saline (RCE -5.9 ± 5.0 %; IAUC -3.7 ± 3.6 %) or glucose (RCE -1.6 ± 9.0 %; IAUC +0.6 ± 6.4 %). Pixel-by-pixel analysis revealed small post-injection changes in RCE and IAUC between the glucose and saline groups, all within 13 % range of their baseline values.

CONCLUSION

Perturbing the metabolism of GL261 tumors kept at moderate hypothermia with hyperglycemia did not induce significant changes in the permeability/perfusion of these tumors. This is relevant for future studies with this model since regional differences in glucose accumulation could thus reflect basal heterogeneities in vasculature and/or metabolism of GL261 tumors.

In vivo assessment of the vascular disrupting effect of M410 by DCE-MRI biomarker in a rabbit model of liver tumor

The present study aimed to prospectively monitor the vascular disrupting effect of M410 by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in rabbits with VX2 liver tumors. Twenty-eight rabbits bearing VX2 tumors in the left lobe of the liver were established and randomly divided into treatment and control groups, intravenously injected with 25 mg/kg M410 or sterile saline, respectively. Conventional and DCE-MRI data were acquired on a 3.0-T MR unit at pretreatment, 4 h, 1, 4, 7 and 14 days post-treatment. Histopathological examinations [hematoxylin and eosin (H&E) and CD34 immunohistochemisty staining] were performed at each time point. The dynamic changes in tumor volume, kinetic DCE-MRI parameter [volume transfer constant (Ktrans)] and histological data were evaluated. Tumors grew slower in the M410 group 4-14 days following treatment, compared with rapidly growing tumors in the control group (P<0.05). At 4 h, 1 and 4 days, Ktrans significantly decreased in the M410 group compared with that in the control group (P<0.05). However, Ktrans values were similar in the two groups for the other time points studied. The changes in DCE-MRI parameters were consistent with the results obtained from H&E and CD34 staining of the tumor tissues. DCE-MRI parameter Ktrans may be used as a non-invasive imaging biomarker to monitor the dynamic histological changes in tumors following treatment with the vascular targeting agent M410.

Screening and detection advances in magnetic resonance image-guided prostate biopsy

Multiparametric magnetic resonance imaging (MRI) has provided a method for visualizing prostate cancer. MRI-ultrasonography fusion allows prostate biopsy to be performed quickly, on an outpatient basis, using the transrectal technique. Targeted biopsies are more sensitive for detection of prostate cancer than nontargeted, systematic biopsies and detect more significant prostate cancers and fewer insignificant cancers than conventional biopsies. A negative MRI scan should not defer biopsy. Two groups who will especially benefit from targeted prostate biopsy are men with low-risk lesions in active surveillance and men with increased prostate-specific antigen levels and previous negative conventional biopsies.

Nature-inspired nanoformulations for contrast-enhanced in vivo MR imaging of macrophages

Magnetic resonance imaging (MRI) of macrophages in atherosclerosis requires the use of contrast-enhancing agents. Reconstituted lipoprotein particles that mimic native high-density lipoproteins (HDL) are a versatile delivery platform for Gd-based contrast agents (GBCA) but require targeting moieties to direct the particles to macrophages. In this study, a naturally occurring methionine oxidation in the major HDL protein, apolipoprotein (apo) A-I, was exploited as a novel way to target HDL to macrophages. We also tested if fully functional GBCA-HDL can be generated using synthetic apo A-I peptides. The fluorescence and MRI studies reveal that specific oxidation of apo A-I or its peptides increases the in vitro macrophage uptake of GBCA-HDL by 2-3 times. The in vivo imaging studies using an apo E-deficient mouse model of atherosclerosis and a 3.0 T MRI system demonstrate that this modification significantly improves atherosclerotic plaque detection using GBCA-HDL. At 24 h post-injection of 0.05 mmol Gd kg(-1) GBCA-HDL containing oxidized apo A-I or its peptides, the atherosclerotic wall/muscle normalized enhancement ratios were 90 and 120%, respectively, while those of GBCA-HDL containing their unmodified counterparts were 35 and 45%, respectively. Confocal fluorescence microscopy confirms the accumulation of GBCA-HDL containing oxidized apo A-I or its peptides in intraplaque macrophages. Together, the results of this study confirm the hypothesis that specific oxidation of apo A-I targets GBCA-HDL to macrophages in vitro and in vivo. Furthermore, our observation that synthetic peptides can functionally replace the native apo A-I protein in HDL further encourages the development of these contrast agents for macrophage imaging.

Quantitative hepatic perfusion modeling using DCE-MRI with sequential breathholds

PURPOSE

To develop and demonstrate the feasibility of a new formulation for quantitative perfusion modeling in the liver using interrupted DCE-MRI data acquired during multiple sequential breathholds.

MATERIALS AND METHODS

A new mathematical formulation to estimate quantitative perfusion parameters using interrupted data was developed. Using this method, we investigated whether a second degree-of-freedom in the tissue residue function (TRF) improves quality-of-fit criteria when applied to a dual-input single-compartment perfusion model. We subsequently estimated hepatic perfusion parameters using DCE-MRI data from 12 healthy volunteers and 9 cirrhotic patients with a history of hepatocellular carcinoma (HCC); and examined the utility of these estimates in differentiating between healthy liver, cirrhotic liver, and HCC.

RESULTS

Quality-of-fit criteria in all groups were improved using a Weibull TRF (2 degrees-of-freedom) versus an exponential TRF (1 degree-of-freedom), indicating nearer concordance of source DCE-MRI data with the Weibull model. Using the Weibull TRF, arterial fraction was greater in cirrhotic versus normal liver (39 ± 23% versus 15 ± 14%, P = 0.07). Mean transit time (20.6 ± 4.1 s versus 9.8 ± 3.5 s, P = 0.01) and arterial fraction (39 ± 23% versus 73 ± 14%, P = 0.04) were both significantly different between cirrhotic liver and HCC, while differences in total perfusion approached significance.

CONCLUSION

This work demonstrates the feasibility of estimating hepatic perfusion parameters using interrupted data acquired during sequential breathholds.

DCE@urLAB: a dynamic contrast-enhanced MRI pharmacokinetic analysis tool for preclinical data

Background

DCE@urLAB is a software application for analysis of dynamic contrast-enhanced magnetic resonance imaging data (DCE-MRI). The tool incorporates a friendly graphical user interface (GUI) to interactively select and analyze a region of interest (ROI) within the image set, taking into account the tissue concentration of the contrast agent (CA) and its effect on pixel intensity.

Results

Pixel-wise model-based quantitative parameters are estimated by fitting DCE-MRI data to several pharmacokinetic models using the Levenberg-Marquardt algorithm (LMA). DCE@urLAB also includes the semi-quantitative parametric and heuristic analysis approaches commonly used in practice. This software application has been programmed in the Interactive Data Language (IDL) and tested both with publicly available simulated data and preclinical studies from tumor-bearing mouse brains.

Conclusions

A user-friendly solution for applying pharmacokinetic and non-quantitative analysis DCE-MRI in preclinical studies has been implemented and tested. The proposed tool has been specially designed for easy selection of multi-pixel ROIs. A public release of DCE@urLAB, together with the open source code and sample datasets, is available at http://www.die.upm.es/im/archives/DCEurLAB/.

Using intraoperative dynamic contrast-enhanced T1-weighted MRI to identify residual tumor in glioblastoma surgery

OBJECT

The goal of surgery in high-grade gliomas is to maximize the resection of contrast-enhancing tumor without causing additional neurological deficits. Intraoperative MRI improves surgical results. However, when using contrast material intraoperatively, it may be difficult to differentiate between surgically induced enhancement and residual tumor. The purpose of this study was to assess the usefulness of intraoperative dynamic contrast-enhanced T1-weighted MRI to guide this differential diagnosis and test it against tissue histopathology.

METHODS

Preoperative and intraoperative dynamic contrast-enhanced MRI was performed in 21 patients with histopathologically confirmed WHO Grade IV gliomas using intraoperative 3-T MRI. Standardized regions of interest (ROIs) were placed manually at 2 separate contrast-enhancing areas at the resection border for each patient. Time-intensity curves (TICs) were generated for each ROI. All ROIs were biopsied and the TIC types were compared with histopathological results. Pharmacokinetic modeling was performed in the last 10 patients to confirm nonparametric TIC analysis findings.

RESULTS

Of the 42 manually selected ROIs in 21 patients, 25 (59.5%) contained solid tumor tissue and 17 (40.5%) retained the brain parenchymal architecture but contained infiltrating tumor cells. Time-intensity curves generated from residual contrast-enhancing tumor and their preoperative counterparts were comparable and showed a quick and persistently increasing slope ("climbing type"). All 17 TICs obtained from regions that did not contain solid tumor tissue were undulating and low in amplitude, compared with those obtained from residual tumors ("low-amplitude type"). Pharmacokinetic findings using the transfer constant, extravascular extracellular volume fraction, rate constant, and initial area under the curve parameters were significantly different for the tumor mass, nontumoral regions, and surgically induced contrast-enhancing areas.

CONCLUSIONS

Intraoperative dynamic contrast-enhanced MRI provides quick, reproducible, high-quality, and simply interpreted dynamic MR images in the intraoperative setting and can aid in differentiating surgically induced enhancement from residual tumor.

Stacked autoencoders for unsupervised feature learning and multiple organ detection in a pilot study using 4D patient data

Medical image analysis remains a challenging application area for artificial intelligence. When applying machine learning, obtaining ground-truth labels for supervised learning is more difficult than in many more common applications of machine learning. This is especially so for datasets with abnormalities, as tissue types and the shapes of the organs in these datasets differ widely. However, organ detection in such an abnormal dataset may have many promising potential real-world applications, such as automatic diagnosis, automated radiotherapy planning, and medical image retrieval, where new multimodal medical images provide more information about the imaged tissues for diagnosis. Here, we test the application of deep learning methods to organ identification in magnetic resonance medical images, with visual and temporal hierarchical features learned to categorize object classes from an unlabeled multimodal DCE-MRI dataset so that only a weakly supervised training is required for a classifier. A probabilistic patch-based method was employed for multiple organ detection, with the features learned from the deep learning model. This shows the potential of the deep learning model for application to medical images, despite the difficulty of obtaining libraries of correctly labeled training datasets and despite the intrinsic abnormalities present in patient datasets.

Prostatic arterial embolization: post-procedural follow-up

Prostatic arterial embolization (PAE) gained special attention in the past years as a potential minimally invasive technique for benign prostatic hyperplasia. Treatment decisions are based on morbidity and quality-of-life issues and the patient has a central role in decision-making. Medical therapy is a first-line treatment option and surgery is usually performed to improve symptoms and decrease the progression of disease in patients who develop complications or who have inadequately controlled symptoms on medical treatment. The use of validated questionnaires to assess disease severity and sexual function, uroflowmetry studies, prostate-specific antigen and prostate volume measurements are essential when evaluating patients before PAE and to evaluate response to treatment. PAE may be performed safely with minimal morbidity and without associated mortality. The minimally invasive nature of the technique inducing a significant improvement in symptom severity associated with prostate volume reduction and a slight improvement in the sexual function are major advantages. However, as with other surgical therapies for benign prostatic hyperplasia, up to 15% of patients fail to show improvement significantly after PAE, and there is a modest improvement of the peak urinary flow.

Quantitative dynamic contrast-enhanced MRI of pelvic and lumbar bone marrow: effect of age and marrow fat content on pharmacokinetic parameter values

OBJECTIVE

The purpose of this study was to determine the effects of age and fat content on quantitative dynamic contrast-enhanced MRI (DCE-MRI) parameters in the bone marrow of the lumbar spine and pelvis. The interreader reproducibility of this technique will also be assessed.

MATERIALS AND METHODS

Forty-three DCE-MRI studies of the female pelvis defined the study group. Quantitative pharmacokinetic perfusion parameters of lumbar and pelvic marrow were analyzed by three readers on a DCE-MRI postprocessing platform. Linear regression analysis was performed to determine the effect of age and marrow fat fraction on the parameters of transfer constant (K(trans)), efflux rate constant (K(ep)), extravascular extracellular space (V(e)), and initial area under the gadolinium curve at 60 seconds (iAUGC(60)). Interreader agreement was assessed by means of intraclass correlation coefficient calculation.

RESULTS

A weak but statistically significant correlation was established between both age and fat fraction and the parameters K(trans) (R(2) = 0.14) and K(ep) (R(2) = 0.09). There was also a weak but statistically significant correlation between fat fraction and V(e) (R(2) = 0.116) and iAUGC(60) (R(2) = 0.108), but no correlation between age and these parameters. Intraclass correlation coefficients of parameter measurements by different readers were all greater than 0.7 at the p < 0.05 level.

CONCLUSION

Age and fat fraction have small measurable effects on quantitative DCE-MRI parameters in bone marrow. However, given the wide interindividual variation of these parameters, these effects are unlikely to confound changes related to malignancy or treatment. Also of note, there was strong interreader reproducibility of parameter measurements among a range of experience levels, suggesting that the reader-reader experience level may not represent a significant source of variability in bone marrow DCE-MRI.

Assessment of feasibility to use computer aided texture analysis based tool for parametric images of suspicious lesions in DCE-MR mammography

Our aim was to analyze the feasibility of computer aided malignant tumor detection using the traditional texture analysis applied on two-compartment-based parameter pseudoimages of dynamic contrast-enhanced magnetic resonance (DCE-MR) breast image data. A major contribution of this research will be the through-plane assessment capability. Texture analysis was performed on two-compartment-based pseudo images of DCE-MRI datasets of breast data of eight subjects. The resulting texture parameter pseudo images were inputted to a feedforward neural network classification system which uses the manual segmentations of a primary radiologist as a gold standard, and each voxel was assigned as malignant or nonmalignant. The classification results were compared with the lesions manually segmented by a second radiologist. Results show that the mean true positive fraction (TPF) and false positive fraction (FPF) performance of the classifier vs. primary radiologist is statistically as good as the mean TPF and FPF performance of the second radiologist vs. primary radiologist with a confidence interval of 95% using a one-sample t-test with α = 0.05. In the experiment implemented on all of the eight subjects, all malignant tumors marked by the primary radiologist were classified to be malignant by the computer classifier. Our results have shown that neural network classification using the textural parameters for automated screening of two-compartment-based parameter pseudo images of DCE-MRI as input data can be a supportive tool for the radiologists in the preassessment stage to show the possible cancerous regions and in the postassessment stage to review the segmentations especially in analyzing complex DCE-MRI cases.

Comparison of two vascular-disrupting agents at a clinically relevant dose in rodent liver tumors with multiparametric magnetic resonance imaging biomarkers

We sought to compare the therapeutic efficacy between two vascular-disrupting agents, combretastatin A4 phosphate (CA4P) and ZD6126, at a clinically relevant dose on tumor models with magnetic resonance imaging (MRI). Thirty rats with liver rhabdomyosarcoma were randomized into CA4P (10 mg/kg), ZD6126 (10 mg/kg), and control group (n=10 for each group). Multiparametric MRI biomarkers including tumor volume, enhancement ratio, necrosis ratio, apparent diffusion coefficient (ADC), and K (volume transfer constant) derived from T2-weighted, T1-weighted, contrast-enhanced T1-weighted, and diffusion-weighted imaging, and dynamic contrast-enhanced MRI were compared at pretreatment, 1 h, 6 h, 24 h, 48 h, and 120 h posttreatment; they were validated using ex-vivo techniques. Relative to rapidly growing tumors without necrosis in control rats, tumors grew slower in the CA4P group compared with the ZD6126 group with a higher necrosis ratio at 120 h (P<0.05), as proven by histopathology. In the CA4P group, K decreased from 1 h until 6 h, and partially recovered at 120 h. In the ZD6126 group, the reduced K at 1 h began to rebound from 6 h and exceeded the baseline value at 120 h (P<0.05), parallel to evolving enhancement ratios (P<0.05). ADC revealed more necrotic tumors with CA4P versus ZD6126 at 120 h (P<0.05). The different tumor responses were confirmed by ex-vivo microangiography and histopathology. CA4P was more effective than ZD6126 in impairing blood supply, inducing necrosis, and delaying growth in rat liver tumors at a clinically relevant dose. A single dose of vascular-disrupting agent was insufficient to destroy the tumor. The multiparametric MRI biomarkers enabled in-vivo noninvasive comparison of therapeutic efficacy between CA4P and ZD6126.