Image quality improvement in cone beam computed tomography images using deep convolutional neural networks

A method of using a deep convolutional neural network to reduce one or more artifacts in a projection space image obtained from a patient, the method comprising:

• collecting a three-dimensional artifact contaminated cone beam computed tomography (CBCT) projection volume of a patient, the artifact contaminated CBCT volume including at least one projection space image;
• and
• using a trained deep convolutional neural network to reduce at least one artifact in the at least one CBCT projection space image of the artifact contaminated CBCT volume, the deep convolutional neural network including a model trained by using a three-dimensional, artifact contaminated CBCT projection volume and a corresponding artifact reduced, CBCT projection volume collected from respective patients in a group of patients.

A method for training a deep convolutional neural network to reduce one or more artifacts in a cone beam computed tomography (CBCT) anatomical image, the method comprising:

• obtaining a plurality of CBCT anatomical images from an artifact contaminated CBCT anatomical volume, and a plurality of computed tomography (CT) anatomical images obtained from at least one corresponding artifact reduced CT anatomical volume for respective patients in a group of patients;
• and
• using the plurality of CBCT anatomical images and the corresponding CT anatomical images to train a deep convolutional neural network to reduce at least one artifact in at least one evaluated CBCT anatomical image.

A method of using a deep convolutional neural network to reduce one or more artifacts in an anatomical image obtained of a patient, the method comprising:

• collecting at least one artifact contaminated cone beam computed tomography (CBCT) anatomical image produced from imaging of the patient;
• and
• using a trained deep convolutional neural network to reduce at least one artifact in the at least one CBCT anatomical image collected from the patient, the deep convolutional neural network including a model trained by using anatomical images from at least one artifact contaminated CBCT anatomical volume, and at least one corresponding artifact reduced computed tomography (CT) anatomical volume for respective patients in a group of patients.

At least one non-transitory machine-readable storage medium comprising instructions, wherein the instructions, when executed by a processing circuitry of a computing device, cause the processing circuitry to perform operations to reduce one or more artifacts in an anatomical image obtained of a patient, the operations comprising:

• collecting a three-dimensional artifact contaminated cone beam computed tomography (CBCT) projection volume of a patient, the artifact contaminated CBCT volume including at least one projection space image;
• and
• using a trained deep convolutional neural network to reduce at least one artifact in the at least one CBCT projection space image of the artifact contaminated CBCT volume, the deep convolutional neural network including a model trained by using a three-dimensional, artifact contaminated CBCT projection volume and a corresponding artifact reduced, CBCT projection volume collected from respective patients in a group of patients.

At least one non-transitory machine-readable storage medium comprising instructions, wherein the instructions, when executed by processing circuitry of a computing device, cause the processing circuitry to perform operations to reduce one or more artifacts in an anatomical image obtained of a patient, the operations comprising:

• collecting at least one artifact contaminated cone beam computed tomography (CBCT) anatomical image produced from imaging of the patient;
• and
• using a trained deep convolutional neural network to reduce at least one artifact in the at least one CBCT anatomical image collected from the patient, the deep convolutional neural network including a model trained by using anatomical images from at least one artifact contaminated CBCT anatomical volume, and at least one corresponding artifact reduced computed tomography (CT) anatomical volume for respective patients in a group of patients.