ContiLNN: Mitigating Slice Sampling Discontinuity with Liquid Neural Networks for Medical Image Restoration
ContiLNN enhances 2D medical image restoration by introducing bidirectional closed-form continuous-time (Bi-CfC) modules, which model cross-slice context using liquid neural dynamics. This approach addresses discontinuities caused by varying slice sampling, improving fidelity across CT denoising, MRI super-resolution, and PET restoration compared to existing methods.