Deep-learning-based scatter estimation and correction for X-ray projection data and computer tomography (CT)

An apparatus, comprising:

• processing circuitry configured to train a neural network to estimate, based on a plurality of energy-resolved projection images of respective material components, an X-ray scatter flux corresponding to the plurality of energy-resolved projection images, the neural network being trained by
• obtaining a training dataset including input images that are paired with respective target images, the respective target images being a desired result of applying the input images to the neural network, the target images representing an X-ray scatter flux of the input images, and the input images representing a plurality of energy-resolved projection images that are resolved according to a plurality of energy bins and that are separated according material component,
• initializing the neural network, the neural network including neuronal nodes connected by connections having weighting coefficients, and the initializing of the neural network including initializing the weighting coefficients,
• applying the input images to the neural network to generate an output image,
• calculating a loss function representing a difference or disagreement between the output image and the target images,
• updating the weighting coefficients in the neural network to optimize the cost function, and
• stopping, upon satisfying a predefined stopping criteria, the updating of the weighting coefficients, and then outputting the neural network as a trained neural network.

A method, comprising:

• obtaining, via processing circuitry, projection data representing an intensity of X-ray radiation detected at a plurality of detector elements during a computed tomography (CT) scan;
• generating, via the processing circuitry and based on the projection data, (i) a first plurality of energy-resolved projection images corresponding to a first material component of the material components and (ii) a second plurality of energy-resolved projection images corresponding to a second material component of the material components;
• acquiring, via the processing circuitry, a neural network including weighting coefficients of connections between neuronal nodes of respective layers of a plurality of layers between an input layer and an output layer of the neural network, the neural network being a trained neural network;
• applying, via the processing circuitry, the first plurality of energy-resolved projection images together with the second plurality of energy-resolved projection images to the trained neural network to estimate an X-ray scatter flux of the projection data;
• and
• removing, via the processing circuitry and using the estimated X-ray scatter flux, a scatter component from the projection data to generate corrected projection data representing an intensity of a primary X-ray beam in an absence of X-ray scatter.

A non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by a computer, cause the computer to perform a method of performing a scatter estimation and correction, comprising:

• obtain projection data representing an intensity of X-ray radiation detected at a plurality of detector elements during a computed tomography (CT) scan;
• generate, based on the projection data, (i) a first plurality of energy-resolved projection images corresponding to a first material component of the material components and (ii) a second plurality of energy-resolved projection images corresponding to a second material component of the material components;
• acquire a neural network including weighting coefficients of connections between neuronal nodes of respective layers of a plurality of layers between an input layer and an output layer of the neural network, the neural network being a trained neural network;
• apply the first plurality of energy-resolved projection images together with the second plurality of energy-resolved projection images to the trained neural network to estimate an X-ray scatter flux of the projection data;
• and
• correct, using the estimated X-ray scatter flux, the projection data to generate corrected projection data representing an intensity of a primary X-ray beam in an absence of X-ray scatter.