Adaptive SENSE reconstruction for parallel imaging with massive array coils

This work presents an adaptive SENSE reconstruction method for parallel magnetic resonance imaging with a large number of localized coils. This method uses a Gaussian model to obtain improved coil sensitivity estimate. For image reconstruction, it dynamically selects a subset of receiver channels, in a pixel-by-pixel fashion, to improve computational efficiency and the signal-to-noise ratio (SNR). Computer simulations and real experiments show that the proposed method reconstructs images with reduced artifacts and higher SNR than the SENSE method.