SNR and resolution are two important parameters for the quantitative assessment of MR images. In k-space, the central part contributes SNR, whereas the edges contribute details. For hyperpolarized GRE sequences, the constant flip angle (CFA) scheme achieves sine-decaying signals due to the non-renewability of the polarization. The homogeneous variable flip angle (H-VFA) scheme proposed by Lei Zhao et al.[1] can yield constant signal from the 1st until the last RF pulse. Thus the edge part of k-space uses the same amount of magnetization excited as the central part. However, this situation is sub-optimal and leads to low SNR images. For diffusion-weighted MRI, the accuracy of the apparent diffusion coefficient (ADC) calibration was significantly affected by SNR[2]. Thus a higher SNR image is more important than one containing more details. The compressed sensing method[3] can be used to obtain higher SNR images with fewer excitations. However, complex arithmetic and priori knowledge for calibration are needed. Here we propose a simple method termed inhomogeneous variable flip angle (I-VFA) to derive ADC of hyperpolarized gases. Higher SNR images and more stable results can be achieved by this simple method.Images were acquired on a Bruker Biospec 7 T animal scanner. An emphysema rat was instilled with 75 IU of porcine elastase stock (Elastin Products Company) and experiments were performed 12 weeks post-instillation. Enriched (86%) ¹²⁹Xe gas was polarized by a home-built polarizer, then thawed from a spiral Teflon container, and finally delivered to rat lung by ventilator. The normal breaths are ventilated with pure oxygen. Three xenon gas pre-washes were prepared before the breath for sampling, and a 4-second breath hold was needed for the sampling. Diffusion weighted images were obtained by a 2-D gradient echo sequence. Four interleaved images (TR/TE = 9/4.2 ms, matrix = 96 x 96, FOV = 5 x 5 cm², echo position = 30%, centric encoding, without slice selection) were acquired with different b-values (rise and fall time of 0.123 ms, Δ = 1.0 ms, ordinally b-values = 0, 20, 32, 46 s/cm²) in a single breath hold. The flip angle in CFA scheme was ~ 5°. The flip angles in H-VFA scheme were followed the formula $$$\theta_{n}=\tan^{-1}\left[\exp\left(-TR/T_{1}\right)\cdot\sin\left(\theta_{n+1}\right)\right]$$$, where n=1,2,3,…,N-1, with N = 384. The flip angles in I-VFA scheme contained two parts: the centric parts were the forward 96 angles within the N = 104 variable angles, and the edge parts were N = 244 variable angles. Simulations and image post processing were performed in Matlab.Fig. 1 shows the contrast of simulated signals obtained by I-VFA, CFA, and H-VFA. The forward signals by I-VFA are higher than those of CFA and H-VFA, while the remaining signals are smaller than those of CFA and H-VFA. Fig. 2 shows the contrast of images acquired by I-VFA, CFA, H-VFA and zero-filled with I-VFA. The central edge in k-space of I-VFA is significant, but the corresponding image with b = 0 is distinct. On the contrary, the corresponding image with b = 0 obtained by zero-filled is indistinct. The ADC map achieved by I-VFA shows obvious enhancement compared to those of CFA and H-VFA, as shown by the visibly higher ADC values on the apex of emphysema rat lung (red top arrows). The mean ADC values of parenchyma are 0.0286±0.0093 cm²/s for zero-fill map, 0.0284±0.0094 cm²/s for I-VFA map. 0.0291±0.0096 cm²/s for CFA map, and 0.0281±0.0099 cm²/s for H-VFA map. The higher value in the CFA scheme may due to the affect of sine-decaying signals.The I-VFA scheme is a potential useful method for hyperpolarized ¹²⁹Xe DWI. The lack of signal in the outer parts of k-space does not affect the ADC measurement significantly.