Parameter estimation of perfusion models in dynamic contrast-enhanced imaging: a unified framework for model comparison

Quantitative MRI for Evaluation of Musculoskeletal Disease: Cartilage and Muscle Composition, Joint Inflammation, and Biomechanics in Osteoarthritis

ABSTRACT

Magnetic resonance imaging (MRI) is a valuable tool for evaluating musculoskeletal disease as it offers a range of image contrasts that are sensitive to underlying tissue biochemical composition and microstructure. Although MRI has the ability to provide high-resolution, information-rich images suitable for musculoskeletal applications, most MRI utilization remains in qualitative evaluation. Quantitative MRI (qMRI) provides additional value beyond qualitative assessment via objective metrics that can support disease characterization, disease progression monitoring, or therapy response. In this review, musculoskeletal qMRI techniques are summarized with a focus on techniques developed for osteoarthritis evaluation. Cartilage compositional MRI methods are described with a detailed discussion on relaxometric mapping (T 2 , T 2 *, T 1ρ ) without contrast agents. Methods to assess inflammation are described, including perfusion imaging, volume and signal changes, contrast-enhanced T 1 mapping, and semiquantitative scoring systems. Quantitative characterization of structure and function by bone shape modeling and joint kinematics are described. Muscle evaluation by qMRI is discussed, including size (area, volume), relaxometric mapping (T 1 , T 2 , T 1ρ ), fat fraction quantification, diffusion imaging, and metabolic assessment by 31 P-MR and creatine chemical exchange saturation transfer. Other notable technologies to support qMRI in musculoskeletal evaluation are described, including magnetic resonance fingerprinting, ultrashort echo time imaging, ultrahigh-field MRI, and hybrid MRI-positron emission tomography. Challenges for adopting and using qMRI in musculoskeletal evaluation are discussed, including the need for metal artifact suppression and qMRI standardization.

Physics-informed neural networks for myocardial perfusion MRI quantification

Tracer-kinetic models allow for the quantification of kinetic parameters such as blood flow from dynamic contrast-enhanced magnetic resonance (MR) images. Fitting the observed data with multi-compartment exchange models is desirable, as they are physiologically plausible and resolve directly for blood flow and microvascular function. However, the reliability of model fitting is limited by the low signal-to-noise ratio, temporal resolution, and acquisition length. This may result in inaccurate parameter estimates. This study introduces physics-informed neural networks (PINNs) as a means to perform myocardial perfusion MR quantification, which provides a versatile scheme for the inference of kinetic parameters. These neural networks can be trained to fit the observed perfusion MR data while respecting the underlying physical conservation laws described by a multi-compartment exchange model. Here, we provide a framework for the implementation of PINNs in myocardial perfusion MR. The approach is validated both in silico and in vivo. In the in silico study, an overall decrease in mean-squared error with the ground-truth parameters was observed compared to a standard non-linear least squares fitting approach. The in vivo study demonstrates that the method produces parameter values comparable to those previously found in literature, as well as providing parameter maps which match the clinical diagnosis of patients.

A Novel Statistical Optimization Algorithm for Estimating Perfusion Curves in Susceptibility Contrast-Enhanced MRI

Dynamic susceptibility contrast-enhanced magnetic resonance imaging is an important tool for evaluating intravascular indicator dynamics, which in turn is valuable for understanding brain physiology and pathophysiology. This procedure usually involves fitting a gamma-variate function to observed concentration-time curves in order to eliminate undesired effects of recirculation and the leakage of contrast agents. Several conventional curve-fitting approaches are routinely applied. The nonlinear optimization methods typically are computationally expensive and require reliable initial values to guarantee success, whereas a logarithmic linear least-squares (LL-LS) method is more stable and efficient, and does not suffer from the initial-value problem, but it can show degraded performance, especially when a few data or outliers are present. In this paper, we demonstrate, that the original perfusion curve-fitting problem can be transformed into a gamma-distribution-fitting problem by treating the concentration-time curves as a random sample from a gamma distribution with time as the random variable. A robust maximum-likelihood estimation (MLE) algorithm can then be readily adopted to solve this problem. The performance of the proposed method is compared with the nonlinear Levenberg-Marquardt (L-M) method and the LL-LS method using both synthetic and real data. The results show that the performance of the proposed approach is far superior to those of the other two methods, while keeping the advantages of the LL-LS method, such as easy implementation, low computational load, and dispensing with the need to guess the initial values. We argue that the proposed method represents an attractive alternative option for assessing intravascular indicator dynamics in clinical applications. Moreover, we also provide valuable suggestions on how to select valid data points and set the initial values in the two traditional approaches (LL-LS and nonlinear L-M methods) to achieve more reliable estimations.

In vivo noninvasive preclinical tumor hypoxia imaging methods: a review

Tumors exhibit areas of decreased oxygenation due to malformed blood vessels. This low oxygen concentration decreases the effectiveness of radiation therapy, and the resulting poor perfusion can prevent drugs from reaching areas of the tumor. Tumor hypoxia is associated with poorer prognosis and disease progression, and is therefore of interest to preclinical researchers. Although there are multiple different ways to measure tumor hypoxia and related factors, there is no standard for quantifying spatial and temporal tumor hypoxia distributions in preclinical research or in the clinic. This review compares imaging methods utilized for the purpose of assessing spatio-temporal patterns of hypoxia in the preclinical setting. Imaging methods provide varying levels of spatial and temporal resolution regarding different aspects of hypoxia, and with varying advantages and disadvantages. The choice of modality requires consideration of the specific experimental model, the nature of the required characterization and the availability of complementary modalities as well as immunohistochemistry.

Hierarchical Bayesian myocardial perfusion quantification

Myocardial blood flow can be quantified from dynamic contrast-enhanced magnetic resonance (MR) images through the fitting of tracer-kinetic models to the observed imaging data. The use of multi-compartment exchange models is desirable as they are physiologically motivated and resolve directly for both blood flow and microvascular function. However, the parameter estimates obtained with such models can be unreliable. This is due to the complexity of the models relative to the observed data which is limited by the low signal-to-noise ratio, the temporal resolution, the length of the acquisitions and other complex imaging artefacts. In this work, a Bayesian inference scheme is proposed which allows the reliable estimation of the parameters of the two-compartment exchange model from myocardial perfusion MR data. The Bayesian scheme allows the incorporation of prior knowledge on the physiological ranges of the model parameters and facilitates the use of the additional information that neighbouring voxels are likely to have similar kinetic parameter values. Hierarchical priors are used to avoid making a priori assumptions on the health of the patients. We provide both a theoretical introduction to Bayesian inference for tracer-kinetic modelling and specific implementation details for this application. This approach is validated in both in silico and in vivo settings. In silico, there was a significant reduction in mean-squared error with the ground-truth parameters using Bayesian inference as compared to using the standard non-linear least squares fitting. When applied to patient data the Bayesian inference scheme returns parameter values that are in-line with those previously reported in the literature, as well as giving parameter maps that match the independant clinical diagnosis of those patients.

On the potential use of dynamic contrast-enhanced (DCE) MRI parameters as radiomic features of cervical cancer

PURPOSE

To evaluate whether the analysis of high-temporal resolution DCE-MRI by various tracer kinetic models could yield useful radiomic features in discriminating cervix carcinoma and normal cervix tissue.

METHODS

Forty-three patients (median age 51 yr; range 26-78 yr) diagnosed with cervical cancer based on postoperative pathology were enrolled in this study with informed consent. DCE-MRI data with temporal resolution of 2 s were acquired and analyzed using the Tofts (TOFTS), extended Tofts (EXTOFTS), conventional two-compartment (CC), adiabatic tissue homogeneity (ATH), and distributed parameter (DP) models. Ability of all kinetic parameters in distinguishing tumor from normal tissue was assessed using Mann-Whitney U test and receiver operating characteristic (ROC) curves. Repeatability of parameter estimates due to sampling of arterial input functions (AIFs) was also studied using intraclass correlation (ICC) analysis.

RESULTS

Fractional extravascular, extracellular volume (Ve) of all models were significantly smaller in cervix carcinoma than normal cervix tissue, and were associated with large values of area under ROC curve (AUC 0.884-0.961). Capillary permeability PS derived from the ATH, CC, and DP models also yielded large AUC values (0.730, 0.860, and 0.797). Transfer constant Ktrans derived from TOFTS and EXTOFTS models yielded smaller AUC (0.587 and 0.701). Repeatability of parameters derived from all models was robust to AIF sampling, with ICC coefficients typically larger than 0.80.

CONCLUSIONS

With the use of high-temporal resolution DCE-MRI, all tracer kinetic models could reflect pathophysiological differences between cervix carcinoma and normal tissue (with significant differences in Ve and PS) and potentially yield radiomic features with diagnostic value.

A multi-model framework to estimate perfusion parameters using contrast-enhanced ultrasound imaging

PURPOSE

Contrast-enhanced ultrasound imaging has expanded the diagnostic potential of ultrasound by enabling real-time imaging and quantification of tissue perfusion. Several perfusion models and curve fitting methods have been developed to quantify the temporal behavior of tracer signal and standardize perfusion quantification. While the least-squares approach has traditionally been applied for curve fitting, it can be inadequate for noisy and complex data. Moreover, previous research suggests that certain perfusion models may be more relevant depending on the organ or tissue imaged. We propose a multi-model framework to select the most appropriate perfusion model and curve fitting method for each diagnostic application.

METHODS

Our multi-model approach uses a system identification method, which estimates perfusion parameters from the model with the best fit to a given time-intensity curve. We compared current perfusion quantification methods that use a single perfusion model and curve fitting method and our proposed multi-model framework on bolus 3D dynamic contrast-enhanced ultrasound (DCE-US) in vivo images obtained in mice implanted with a colon cancer, as well as on simulation data. The quality of fit in estimating perfusion parameters was evaluated using the Spearman correlation coefficient, the coefficient of determination (R² ), and the normalized root-mean-square error (NRMSE) to ensure that the multi-model framework finds the best perfusion model and curve fitting algorithm.

RESULTS

Our multi-model framework outperforms conventional single perfusion model approaches with least-squares optimization, providing more robust perfusion parameter estimation. R² and NRMSE are 0.98 and 0.18, respectively, for our proposed method. By comparison, the performance of the traditional approach is much more dependent upon the selection of the appropriate model. The R² and NRMSE are 0.91 and 0.31, respectively.

CONCLUSIONS

The proposed multi-model framework for perfusion modeling outperforms the current approach of single perfusion modeling using least-squares optimization and more robustly estimates perfusion parameters when using empiric data labeled by an expert as the gold standard. Our technique is minimally sensitive to issues affecting the accuracy of perfusion parameter estimation, including rise time, noise, region of interest size, and frame rate. This framework could be of key utility in modeling different perfusion systems in different tissues and organs.

CT Perfusion in Patients with Lung Cancer: Squamous Cell Carcinoma and Adenocarcinoma Show a Different Blood Flow

Objectives

To characterize tumour baseline blood flow (BF) in two lung cancer subtypes, adenocarcinoma (AC) and squamous cell carcinoma (SCC), also investigating those “borderline” cases whose perfusion value is closer to the group mean of the other histotype.

Materials and Methods

26 patients (age range 36-81 years) with primary Non-Small Cell Lung Cancer (NSCLC), subdivided into 19 AC and 7 SCC, were enrolled in this study and underwent a CT perfusion, at diagnosis. BF values were computed according to the maximum-slope method and unreliable values (e.g., arising from artefacts or vessels) were automatically removed. The one-tail Welch's t-test (p-value <0.05) was employed for statistical assessment.

Results

At diagnosis, mean BF values (in [mL/min/100g]) of AC group [(83.5 ± 29.4)] are significantly greater than those of SCC subtype [(57.0 ± 27.2)] (p-value = 0.02). However, two central SCCs undergoing artefacts from vena cava and pulmonary artery have an artificially increased mean BF.

Conclusions

The different hemodynamic behaviour of AC and SCC should be considered as a biomarker supporting treatment planning to select the patients, mainly with AC, that would most benefit from antiangiogenic therapies. The significance of results was achieved by automatically detecting and excluding artefactual BF values.

Variations of quantitative perfusion measurement on dynamic contrast enhanced CT for colorectal cancer: implication of standardized image protocol

Tumor angiogenesis is considered an important prognostic factor. With an increasing emphasis on imaging evaluation of the tumor microenvironment, dynamic contrast enhanced-computed tomography (DCE-CT) has evolved as an important functional technique in this setting. Yet many questions remain as to how and when these functional measurements should be performed for each agent and tumor type, and what quantitative models should be used in the fitting process. In this study, we evaluated the variations of perfusion measurement on DCE-CT for rectal cancer patients from (1) different tracer kinetic models, (2) different scan acquisition lengths, and (3) different scan intervals. A total of seven commonly used models were studied: the adiabatic approximation to the tissue homogeneity (AATH) model, adiabatic approximation to the homogeneity tissue with fixed transit time (AATHFT) model, the Tofts model (TM), the extended Tofts model (ETM), Patlak model, Logan model, and the model-free deconvolution method. Akaike's information criterion was used to identify the best fitting model. The interchangeability of different models was further evaluated using Bland-Altman analysis. All models gave comparable blood volume (BV) measurements except the Patlak method. While for the volume transfer constant (K_trans) estimation, AATHFT, AATH, and ETM generated reasonable agreement among each other but not for the other models. Regarding the blood flow (BF) measurement, no two models were interchangeable. In addition, the perfusion parameters were compared with four acquisition times (45, 65, 85, and 105 s) and four temporal intervals (1, 2, 3, and 4 s). No significant difference was observed in the volume transfer constant (K_trans), BV, and BF measurements when comparing data acquired over 65 s with data acquired over 105 s using any of the DCE models in this study. Yet increasing the temporal interval led to a significant overestimation of BF in the deconvolution method. In conclusion, the perfusion measurement is indeed model dependent and the image acquisition/processing technique is dependent. The radiation dose of DCE-CT was an average of 1.5-2 times an abdomen/pelvic CT, which is not insubstantial. To take the DCE-CT forward as a biomarker in oncology, prospective studies should be carefully designed with the optimal image acquisition and analysis technique.

Use of patient outcome endpoints to identify the best functional CT imaging parameters in metastatic renal cell carcinoma patients

OBJECTIVE

To use the patient outcome endpoints overall survival and progression-free survival to evaluate functional parameters derived from dynamic contrast-enhanced CT.

METHODS

69 patients with metastatic renal cell carcinoma had dynamic contrast-enhanced CT scans at baseline and after 5 and 10 weeks of treatment. Blood volume, blood flow and standardized perfusion values were calculated using deconvolution (BV_deconv, BF_deconv and SPV_deconv), blood flow and standardized perfusion values using maximum slope (BF_max and SPV_max) and blood volume and permeability surface area product using the Patlak model (BV_patlak and PS). Histogram data for each were extracted and associated to patient outcomes. Correlations and agreements were also assessed.

RESULTS

The strongest associations were observed between patient outcome and medians and modes for BV_deconv, BV_patlak and BF_deconv at baseline and during the early ontreatment period (p < 0.05 for all). For the relative changes in median and mode between baseline and weeks 5 and 10, PS seemed to have opposite associations dependent on treatment. Interobserver correlations were excellent (r ≥ 0.9, p < 0.001) with good agreement for BF_deconv, BF_max, SPV_deconv and SPV_max and moderate to good (0.5 < r < 0.7, p < 0.001) for BV_deconv and BV_patlak. Medians had a better reproducibility than modes.

CONCLUSION

Patient outcome was used to identify the best functional imaging parameters in patients with metastatic renal cell carcinoma. Taking patient outcome and reproducibility into account, BV_deconv, BV_patlak and BF_deconv provide the most clinically meaningful information, whereas PS seems to be treatment dependent. Standardization of acquisition protocols and post-processing software is necessary for future clinical utilization. Advances in knowledge: Taking patient outcome and reproducibility into account, BV_deconv, BV_patlak and BF_deconv provide the most clinically meaningful information. PS seems to be treatment dependent.