Pharmacokinetic Mapping of the Breast: A New Method for Dynamic MR Mammography

Abstract

A dynamic contrast-enhanced MRI technique for whole breast examinations is presented. The fast kinetics of tissue response during and after constant-rate intravenous infusion of gadolinium diethylenetriaminopentaacetic acid was resolved using a strongly T1-weighted saturation recovery TurboFLASH sequence that makes it possible to acquire signal-time courses sequentially from 15 adjacent slices with a temporal sampling rate of 21 s. On the basis of the mathematically established and experimentally verified linear relationship between the measured saturation recovery TurboFLASH signal variation and the gadolinium diethylenetriaminopentaacetic acid concentration in the tissue, the signal-time courses were analyzed within the framework of pharmacokinetic modeling. In our study, the tissue response was parameterized adequately using an open linear two-compartment model. With this approach, the tissue specific information contained in the signal-time course can be described using only two parameters: an amplitude A, reflecting the degree of MR signal enhancement, and an exchange parameter k21, characterizing vascular permeability and perfusion of the tissue. A clearly arranged representation of the large amount of data (480 saturation recovery TurboFLASH breast images/examination) was accomplished by means of color coding of the computed parameters, resulting in one color-coded pharmacokinetic parameter map/cross-section.

DCE‐MRI pixel‐by‐pixel quantitative curve pattern analysis and its application to osteosarcoma

“…The methods for DCE-MRI data processing can be divided into three types: Heuristic models (9 -14), pharmacokinetic models (15)(16)(17)(18)(19), and pattern analysis (20,21). Within each type there is a variety of derived models.…”

Improvement in the reproducibility of region of interest using an auditory feedback loop: A pilot assessment using dynamic contrast‐enhanced (DCE) breast MR images

“…One is the shape of the curve and the other is how well the measured data fit the model curve. The shape of the curve is also conveyed by pharmacokinetic parameters (e.g., Kep Amp) (15)(16)(17), which can be color coded and overlaid on grayscale images, but the locally calculated values of the pharmacokinetic parameters may be irrelevant if the measured data do not fit the model curve well. Local GF information for the time-intensity curve is conveyed through auditory feedback while a reader draws a ROI.…”

“…SS error is the summation of squared differences between measured data points and mathematically calculated ideal data points from the best fitting pharmacokinetic parameters (Amp, Kep, Kel) from two compartment modeling algorithm (16,17) for each pixel point. The pharmacokinetic parameters are estimated using the Levenberg-Marquardt curve-fitting algorithm.…”

“…2e), pharmacokinetic parameters (Amp: amplitude, Kep: redistribution rate constant, Kel: elimination rate constant; Tleg: time log; Fig. 2f) (15)(16)(17), image-slice information (Fig. 2g), and adjustable sound parameters ( Fig.…”

Model-free visualization of suspicious lesions in breast MRI based on supervised and unsupervised learning

“…Pixel-mapping functions are used to map individual time-series signals to pseudo-colors which reflect dedicated features of the temporal signal. These signal features are derived from explicit mathematical models of the time-series signals, like in the threetime-points (3TP) method [14] illustrated in Figure 3, or from more sophisticated pharmacokinetic models [16][17][18] describing the exchange of CA molecules between tissue compartments over time. Beside these model-based approaches, an increasing number of applications apply pattern recognition methods to extract and to visually appreciate clinically relevant information [5][6][7][8][9][10][11]13].…”