Introduction

The use of contrast agent (CA) for the enhancement of different structures and tissues is a standard method in clinical routine MRI. In oncology, usually one initial scan before injection and a second scan after are performed. However, the two scans are acquired at different times and thus movement of the patient can occur and suboptimal subtraction, potentially leading to false positive or negative diagnosis. Until now, virtual non-contrast enhanced (VNC) imaging has only been used in dual energy computed tomography (CT) [1]. Applying this technique to MRI would allow for contrast enhanced and non-contrast enhanced images without spatial shift and time saving. The aim of this study is to investigate whether it is possible to subtracted the contrast enhanced regions. Using the variable flip angle (VFA) method, two contrast enhanced images are recorded at the same time [2].

Materials and Methods

VFA scans with two flip angles have been performed both on a phantom and two human subjects (Figure 1). The phantom contained different sample tubes casted in gelatin (Figure 3). The sample tubes contained Dotarem as CA (concentration of 0.5 mmol/ml, diluted 1/8 with water), pure water and oil, a mix of two materials each and one mix of all three materials (Figure 3A). The two human subjects (glioblastoma with CA uptake, Fig. 4 and glioma without CA uptake, Fig. 5) underwent scanning to test the postprocessing algorithm in vivo. All scans were acquired on a Siemens Prisma 3T Scanner (Siemens Healthcare, Erlangen, Germany) using a 64-channel head coil. The VFA scan took 4 minutes and 58 seconds. Parameters: 5° and 26° flip angle,TR/TE 15 ms/2.31, 240*240 mm field of view, 0.31*0.31 in-plane resolution and 22 slices of 3 mm slice thickness with 20 % gap. The high FA value was chosen to be optimized for signal of CA (Figure 2) Postprocessing was performed using Matlab R2018a (The Mathworks, Natick, MA) [3]. The system of three unknowns therefore looks like the following:
$$ F_{CA}S_{CA,FA1}+F_{TIS}S_{TIS,FA1}+F_{BG}S_{BG,FA1} = S_{FA1} $$
$$ F_{CA}S_{CA,FA2}+F_{TIS}S_{TIS,FA2}+F_{BG}S_{BG,FA2} = S_{FA2} $$
One assumption of the equation is that the signal value of the pixel of interest only consists of the three constituent materials, i.e. the sum of the fraction of the three materials equals 1.
$$ F_{CA}+F_{TIS}+F_{BG} = 1 $$
The individual fractions can then be found by multiplying both images with the inverse of the sensitivity matrix:
$$ \begin{bmatrix} F_{CA} \\ F_{TIS} \\ F_{BG} \end{bmatrix} = \begin{bmatrix} S_{CA,FA1} & S_{TIS,FA1} & S_{BG,FA1} \\ S_{CA,FA2} & S_{TIS,FA2} & S_{BG,FA2} \\ 1 & 1 & 1 \end{bmatrix}^{-1} \begin{bmatrix} S_{FA1} \\ S_{FA2} \\ 1 \end{bmatrix} $$

BG = background, TIS = Tissue, CA = Contrast Agent, F = Fraction, S = Signal, FA = flip angle
Using six fixed values (Fig. 4) for three different materials, material decomposition is possible by using the inverse of the sensitivity matrix to calculate the material fractions [4]. This leads to a determination of the contrast enhanced areas and an elimination in the resulting image

Results and Discussion

Results of the phantom scans show that three materials can be distinguished correctly (Fig. 3). Postprocessing of the brain scans generates two output images, one for each flip angle (Fig. 4 and 5). All four output images are free of regions enhanced by the contrast agent except for the area between the two hemispheres (falx cerebri) where a slightly brighter residual is still visible (Fig. 5D). Furthermore, all VNC images show the same contrast as the corresponding VFA image. All structural borders are clearly identifiable, gray and white matter can be discriminated throughout the whole image and the visual impression of the TNC images is met.

Conclusion

Using the decomposition algorithm based on six values derived from two images acquired with VFA allows for the generation of VNC images presenting the same information as the TNC image and the subsequent contrast enhanced image.

Acknowledgements

No acknowledgement found.