LRR-CED: low-resolution reconstruction-aware convolutional encoder--decoder network for direct sparse-view CT image reconstruction