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

Abstract

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 (Ktrans), blood volume fraction (vp), and volume fraction of extracellular extravascular space (ve), 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 Ktrans, greater than 0.965 for vp, and greater than 0.948 for ve. 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

Dynamic Contrast Enhanced (DCE) MRI Estimation of Vascular Parameters using Knowledge-Based Adaptive Models

“…Unlike other similar works [24][25][26][27][28][29][30] that have used deep learning and CNNs for adaptive modeling that require a large number of samples for their hyper-parameter tuning, training, and validation, the proposed work uses shallow…”

“…Networks (CNNs) to generate more accurate and stable estimates of PK vascular parameters by extracting timedependent features from DCE-MRI [24][25][26][27][28][29][30] 62 ), or that R1 is proportional. As to the first assumption, in vasculature that leaks, it is clear from a vast literature that competing R1 and R2* effects strongly affect the relationship between [CA] and the R2* MR contrast.…”

“…The problem this paper addresses is one of optimizing the speed, accuracy, and stability of PK parameters through the use of knowledge-based adaptive models. In recent years, adaptive methods have been investigated by multiple groups in an effort to generate more accurate and stable estimates of PK vascular parameters by extracting time-dependent features from DCE-MRI [24][25][26][27][28][29][30] . However, less attention has been paid to the important matter of the tracer kinetic 'model selection' 1,12,14,[31][32][33][34] in processing that physiologically associates estimable vascular parameters with relevant information available in the DCE-MRI signal.…”

“…time trace of CA concentration in DCE MR experiments. Compared to recent studies[24][25][26][27][28][29][30] , one of the novel components of this study is the incorporation of the nested model selection concept into adaptive models for predicting PK parameters.The fundamental question in MRI procedures that aim to estimate cerebrovascular parameters is the relationship between MR contrast and CA concentration (notated as [CA]). Two common approaches have been to assume that R2* is proportional to [CA] (e.g, seeCao et al …”

Radiomics Characterization of Tissues in an Animal Brain Tumor Model Imaged Using Dynamic Contrast Enhanced (DCE) MRI

“…This would result in a more precise physiological-based characterization of tumor heterogeneity and normal tissue. In recent years, many studies have investigated the development of various deep learning and Convolutional Neural Networks (CNNs) to generate more accurate and stable estimates of PK vascular parameters by extracting time-dependent features from DCE-MRI 63,[138][139][140][141][142][143] . Compared to the recent studies 63,[138][139][140][141][142][143] , one of the novel components of this study is the incorporation of the nested model selection concept into the radiomics analysis.…”