Purpose

The 3D FLASH sequence is frequently used in structural imaging of the brain. Tissue contrast inhomogeneity, resulting from inhomogeneous transmit field and receiver sensitivity, significantly affects quantitative structural brain analyses such as classification and quantification of brain tissues in voxel-based morphometry and detection of pathological brain changes in clinical studies [1]. It is important to optimize the sequence to jointly improve contrast efficiency and image homogeneity. In this work, we propose optimal imaging parameters and present methods to improve contrast efficiency and reduce or eliminate image inhomogeneity.

METHODS

Simulation: GM-WM contrast efficiency was simulated using Bloch’s equation based on T₁, T₂, and proton density of the GM, WM, and CSF of the adult brain at 3.0 T: 1400/850/3500 ms, 100/90/300 ms, and 0.75/0.65/1.0, respectively [2,3]. Experiment: MRI scans of ten normal volunteers were acquired on a 3.0 T Siemens Trio-Tim system with a Siemens 32 channel head coil. T1-weighted images were acquired using a 3D FLASH sequence with the resolution of 1 mm3 at TR/FA=13 ms/17^o. A segmented EPI sequence (FOV 240x218 mm2, matrix 128x116, slice thickness 5 mm, TR/TE = 3000/13 ms, flip angles (FA) 120^o and 60^o) was used to acquire EPI images from the subjects. The minimum contrast image of the brain was obtained at TR/TE =1600/13 ms with a flip angle of 90^o and used to estimate receiver sensitivity [4].

RESULTS AND DISCUSSIONS

Fig. 1 shows simulated GM-WM contrast efficiency as a function of FA and TR. The results demonstrate that high GM-WM contrast efficiencies can be achieved with a wide range of FAs and TRs. Since multiple settings of TR and FA can be used to achieve high GM-WM contrast efficiency, to reduce the total scan time and patient burden, a short TR is preferred. We suggest that the optimal TR is 13 ms, and the optimal FA is between 16^o to 18^o. The optimal FA is finally chosen to be 17^o to achieve maximum contrast efficiency and reduce variations of GM-WM contrast caused by non-uniform transmit field (Fig. 2a). When FA is 17^o, image contrast is insensitive to the inhomogeneity of the transmit field, but can still be affected by inhomogeneous receiver sensitivity (Fig. 3a). We further corrected effects of inhomogeneous receiver sensitivity by taking the ratio of the raw image (Fig. 3a) and the measured receiver sensitivity (Fig. 2b). The result is shown in Fig. 3b. After the correction, tissue intensities became more uniform across the whole brain. The intensity histogram of the corrected image is shown in Fig. 3d. The histograms of the GM and WM are more separated, with distinct peaks. It has been shown that contrast inhomogeneity cannot be fully corrected by general post-processing methods, such as the N3 algorithm [1]. We propose to select the optimal FA to not only maximize contrast efficiency but also reduce or remove the effect of transmit field on contrast inhomogeneity. With our optimal parameters, contrast inhomogeneity only results from non-uniform receiver sensitivity, and can be corrected using the estimated receiver sensitivity. Therefore, our method can correct contrast inhomogeneity caused by both transmit field and receiver sensitivity, and improve tissue histograms and brain tissue classification.

CONCLUSIONS

We present a parameter optimization scheme that incorporates contrast inhomogeneity correction to improve the quality of structural brain images acquired with the 3D FLASH sequence. The optimal imaging parameters to maximize GM-WM contrast efficiency and minimize contrast inhomogeneity caused by the transmit field are: TR/TE = 13 /minimized ms, FA 17^o. Contrast inhomogeneity caused by receiver sensitivity is corrected with estimated receiver sensitivity. Our method can be used to reduce tissue contrast variation across the brain, and improve the quality of input images for voxel-based morphometry analyses.