CiNet Monthly Seminar
August 6, 2018
16:00 ~ 17:00
CiNet 1F Conference Room
“Exploration of Machine Learning Approaches for Automating Medical Image Reconstruction and Acquisition”
Harvard Medical School
Host : Kaoru Amano (PI)
Image reconstruction plays a critical role in the implementation of all contemporary imaging modalities across the physical and life sciences including optical, radar, magnetic resonance imaging (MRI), X-ray computed tomography (CT), positron emission tomography (PET), ultrasound, and radio astronomy. Image reconstruction is challenging because analytic knowledge of the exact inverse transform may not exist a priori, especially in the presence of sensor non-idealities and noise.
Thus, the standard reconstruction approach involves approximating the inverse function with multiple ad hoc stages in a signal processing chain whose composition depends on the details of each acquisition strategy, and often requires expert parameter tuning to optimize reconstruction performance. We present a unified framework for image reconstruction, AUtomated TransfOrm by Manifold APproximation (AUTOMAP), which recasts image reconstruction as a data-driven, supervised learning task that allows a mapping between sensor and image domain to emerge from an appropriate corpus of training data. We implement AUTOMAP with a deep neural network and exhibit its flexibility in learning reconstruction transforms for a variety of MRI acquisition strategies, using the same network architecture and hyperparameters. We further demonstrate its efficiency in sparsely representing transforms along low-dimensional manifolds, resulting in superior immunity to noise and a reduction in reconstruction artifacts compared with conventional handcrafted reconstruction methods. In this talk I also describe work in progress on automated pulse sequence generation (AUTOSEQ), wherein we recast the general problem of MR pulse sequence development as a model-free problem optimized with a Bayesian derivative of reinforcement learning within a MRI physics simulation environment. We show preliminary proof-of-principle experiments and demonstrate our agent learning a canonical pulse sequence (the gradient echo) and also generating non-intuitive pulse sequences that can produce signals approximating Fourier spatial encoding.