本文使用 variavle-density possion-disc 采样的多通道膝盖数据进行并行重建和压缩感知重建。

0 数据欠采样sampling pattern

1 计算ESPIRiT maps

% A visualization of k-space data

knee = readcfl('data/knee');
ksp_rss = bart('rss 8', knee);

ksp_rss = squeeze(ksp_rss);
figure, imshow(abs(ksp_rss).^0.125, []); title('k-space')


% Root-of-sum-of-squares image

knee_imgs = bart('fft -i 6', knee);
knee_rss = bart('rss 8', knee_imgs);


% ESPIRiT calibration (one map)

knee_maps = bart('ecalib -c0. -m1', knee);

2 使用小波变换做 L1 sparsity  正则

% l1-regularized reconstruction (wavelet basis)

knee_l1 = bart('pics -l1 -r0.01', knee, knee_maps);

<