Fig. 5 | Scientific Reports

Fig. 5

From: Performance of a GPU- and time-efficient pseudo-3D network for magnetic resonance image super-resolution and motion artifact reduction

Fig. 5

Qualitative comparison of 3D MRI SRR to the state-of-the methods with the scale factor of \(2\times 2\times 2\). The SR images of SRCNN, FSRCNN and DCSRN exhibited noticeable blurriness in both the sagittal and axial views, leading to an almost complete loss of gray matter visibility in the hand knob region (indicated by the arrow). In contrast, ReCNN, MINet, and TS-RCAN demonstrated comparable performance with reduced errors and improved differentiation of small anatomical structures (arrow).

Back to article page