Deep Learning of ADC Maps from Under-sampled Diffusion-Weighted Radially Sampled MRI
Yuemeng Li1, Hee Kwon Song1, Miguel Romanello Giroud Joaquim1, Stephen Pickup1, Rong Zhou1, and Yong Fan1
1Radiology, University of Pennsylvania, Philadelphia, PA, United States
Comparisons with standard ADC extraction and acceleration methods are made to support this concept.
Figure-2: A densely connected Encoder-Decoder network
of CNNs (top and middle), enhanced by spatial-attention and channel-attention
layers (bottom), for computing ADC maps from undersampled complex image-space
data.
Figure-1
: Representative diffusion-weighted images and ADC map. The tumor is indicated
by the red arrow.