MimickNet, Matching Clinical Post-Processing under Realistic Black-Box Constraints
Image post-processing is used in clinical-grade ultrasound scanners to improve image quality (e.g., reduce speckle noise and enhance contrast). These post-processing techniques vary across manufacturers and are generally kept proprietary, which presents a challenge for researchers looking to match current clinical-grade workflows. We introduce a deep learning framework, MimickNet, that transforms conventional delay-and-summed (DAS) beamformed images into the approximate post-processed images found on clinical-grade scanners. Training MimickNet only requires post-processed image samples from a scanner of interest without the need for explicit pairing to DAS data. Unpaired image flexibility allows MimckNet to hypothetically approximate any manufacturer's post-processing without hacking into commercial machines for pre-processed data. MimickNet generates images with an average similarity index measurement (SSIM) of 0.930±0.0892 on a 300 cineloop test set, and it generalizes to cardiac cineloops achieving an SSIM of 0.967±0.002 despite using no cardiac data in the training process. To our knowledge, this is the first work to approximate current clinical-grade ultrasound post-processing under realistic black-box constraints where before and after post-processing data is unavailable. MimickNet can be used out of the box or retrained to serve as a clinical post-processing baseline to compare against for future works in ultrasound image formation. To this end, we have made the MimickNet software open source at https://github.com/ouwen/mimicknet.
