Fig. 10From: macJNet: weakly-supervised multimodal image deformable registration using joint learning framework and multi-sampling cascaded MINDFeature map visualization of macMIND and MIND. macMIND shows its advantage for representing complex anatomical structures. The multi-channel feature map is translated to a single-channel image by calculating the average value of channel-dimension for visualization. In this figure, macMIND is calculated with K = 2, α1 = 0.7, α2 = 0.3Back to article page