A fast and accurate extravasation correction method for Dynamic Susceptibility Contrast MRI
OHSU # 1934
Relative cerebral blood volume (rCBV) is an important biomarker for brain tumor diagnosis, prognosis, and therapy monitoring and Dynamic Susceptibility Contrast (DSC) MRI with low molecular weight Gadolinium contrast agent (Gd CR) is the current method of choice for rCBV quantification. However, Gd CR often leaks out of vascular space when the blood brain barrier is compromised in cancer patients. Current 510k cleared software packages for DSC MRI calculation could either overcorrect or fail to correct for this Gd CR leakage effect. This new software provides a more accurate and faster correction of Gd CR leakage in DSC-MRI.
A novel, robust, fast software that enables Dynamic Susceptibility Contrast (DSC) MRI quantification. Unlike existing approach that relies heavily on numerical manipulations, the OHSU software is based on pharmacokinetic principles, obtains the leakage rate properly and accurately, and reduces calculation time. In particular, the software performs a linearization transform based on fundamental equations. Consequently, Gd CR leakage is the slope of the linear portion of the transformed data, thereby identifying the underlining pseudo extravasation rate constant.
Current Limitations in DSC Gd CR extravasation analysis. Figure 1 shows example time-courses in DSC MRI and associated leakage rate quantification for extracellular or intravascular CRs –Gd and Fe. While the time-course shown in Fig. 1B exhibits a consistent signal reduction post Fe as expected in the predominately T2*-weighted DSC sequence, the signature “leakage effect” of the signal to increase above the baseline due to T1 shortening post Gd is evident in Fig. 1A. With a direct calculation of (described by a fundamental MRI equation), the change of time-course artificially dips below zero after CR first pass (Fig. 2C). The shaded area illustrates that numerical integration of incurs some cancellation, thus underestimates the rCBV value. In particular, integration of the time-course up to ~1.0 minute post injection resulted in an underestimation of rCBV values due to partial cancellation of the positive and negative shaded areas as well as underestimation of the positive peak, as illustrated in Fig. 2C. For the nano-sized intravascular Fe CR, > 0 at all times throughout the CR passage and this returned consistent rCBV estimation as illustrated in Fig. 1D.
More accurate results. Fig. 1E symbols show the same pixel data from FIG. 1A that underwent the OHSU software linearization transform. The solid line with a slope of KL is from a linear regression of the transformed data points reflecting a DSC time-course range from 40 s to70 s post-CR injection. From Fig. 1E, the plot can be divided into three periods: i) transient period of CR first pass; ii) pseudo equilibrium period where linearity obtains; and iii) intravasation/linearity deviation period, where data may deviate from period ii linear trend, most likely due to CR intravasation. In the data used in this example, it was extremely rare (< 0.1%) to observe pixels with appreciable linearity departure (period iii) in Gd CR data when using the Fig. 2E approach, even though normally post CR T1 shortening commonly causes signal increases above baseline (as in FIG. 1A).
Patent Profile: US Provisional Patent Application Filed.
Technology Currently Available for Licensing.
- Xin Li, AI.Advanced Imagaing Research Center
- William Rooney, AI.Advanced Imaging Research Center
- Edward Neuwelt, SM.Neurology
For more information, contact:
Technology Development Manager