Abstract: The present disclosure relates to image reconstruction with favorable properties in terms of noise reduction, spatial resolution, detail preservation and computational complexity. The disclosed techniques may include some or all of: a first-pass reconstruction, a simplified datafit term, and/or a deep learning denoiser. In various implementations, the disclosed technique is portable to different CT platforms, such as by incorporating a first-pass reconstruction step.

Abstract: A computer-implemented method for image processing is provided. The method includes acquiring multiple multi-energy spectral scan datasets and computing basis material images representative of multiple basis materials from the multi-energy spectral scan datasets, wherein the multiple basis material images include correlated noise. The method also includes jointly denoising the multiple basis material images in at least a spectral domain utilizing a deep learning-based denoising network to generate multiple de-noised basis material images.

Abstract: A signal processing method is disclosed, which includes detecting a total intensity of X-rays passing through an object comprising multiple materials; obtaining at least one set of basis information of basis material information of the multiple materials and basis component information of photon-electric absorption basis component and Compton scattering basis component of the object; estimating a scatter intensity component of the detected X-rays based on the at least one set of basis information and the detected total intensity; and obtaining an intensity estimate of primary X-rays incident on a detector based on the detected total intensity and the estimated scatter intensity component. An imaging system adopting the above signal processing method is also disclosed.

Abstract: The present disclosure relates to the use of X-ray detector cassettes that may be abutted or overlapped to form a detector assembly suitable for imaging objects that are too large to image using a single X-ray detector cassette. Such a detector assembly may be customized in terms of the size and/or shape of the field-of-view (FOV). In certain embodiments the radiation-sensitive electronics (e.g., readout electronics) are positioned to the side of the X-ray detecting components (e.g., scintillator, TFT array, and so forth), allowing the cassette to be thin relative to other detector devices and allowing the electronics to remain outside the X-ray beam path.

Abstract: Various embodiments discussed herein utilize a C-shaped imager to provide images with a minimal footprint, such as may be suitable in a surgical context. In addition the systems and methods described herein allow for suitable angular (i.e., azimuthal) scan coverage about the patient. To provide real-time 3D imaging, multiple X-ray tubes or a distributed X-ray source may be employed, coupled with an extended detector or multiple detectors. To reconstruct high-quality volumes, in some implementations reconstruction techniques may be employed that utilize pre-operative (pre-op) computed tomography (CT), magnetic resonance imaging (MRI), ultrasound (U/S), or other suitable modality images or data as prior information.

Abstract: A method relates to the use of deep learning techniques, which may be implemented using trained neural networks (50), to estimate various types of missing projection or other unreconstructed data. Similarly, the method may also be employed to replace or correct corrupted or erroneous projection data as opposed to estimating missing projection data.

Abstract: The present disclosure relates to performing an unrolled iterative reconstruction of image data. Such approaches may include use of neural networks used for one or both of data fidelity and/or image update steps, resulting in improved image quality and accelerated reconstruction for various imaging modalities including CT, PET and MR.

Abstract: The present disclosure relates to performing an unrolled iterative reconstruction of image data. Such approaches may include use of neural networks used for one or both of data fidelity and/or image update steps, resulting in improved image quality and accelerated reconstruction for various imaging modalities including CT, PET and MR.

Abstract: A computer-implemented method for image processing is provided. The method includes acquiring multiple multi-energy spectral scan datasets and computing basis material images representative of multiple basis materials from the multi-energy spectral scan datasets, wherein the multiple basis material images include correlated noise. The method also includes jointly denoising the multiple basis material images in at least a spectral domain utilizing a deep learning-based denoising network to generate multiple de-noised basis material images.

Abstract: In accordance with the present disclosure, the present technique finds a diagnostic scan timing for a non-static object (e.g., a heart or other dynamic object undergoing motion) from raw scan data, as opposed to reconstructed image data. To find the scan timing, a monitoring scan of a patient's heart is performed. In the monitoring scan, the patient dose may be limited or minimized. As the projection data is acquired during such a monitoring scan, the projection data may be subjected to sinogram analysis in a concurrent or real-time manner to determine when to start (or trigger) the diagnostic scan.

Abstract: The present disclosure relates to the design of phantoms configurable using one or more inserts and to their use in generating images that may be used to compare image quality between different imaging systems. Such phantoms may have a modular design with inserts that may be exchanged one for another within a phantom body.

Abstract: The present approach provides a non-invasive methodology for estimation of coronary flow and/or fractional flow reserve. In certain implementations, various approaches for personalizing blood flow models of the coronary vasculature are described. The described personalization approaches involve patient-specific measurements and do not assume or rely on the resting coronary flow being proportional to myocardial mass. Consequently, there are fewer limitations in using these approaches to obtain coronary flow and/or fractional flow reserve estimates non-invasively.

Abstract: A signal processing method is disclosed, which includes detecting a total intensity of X-rays passing through an object comprising multiple materials; obtaining at least one set of basis information of basis material information of the multiple materials and basis component information of photon-electric absorption basis component and Compton scattering basis component of the object; estimating a scatter intensity component of the detected X-rays based on the at least one set of basis information and the detected total intensity; and obtaining an intensity estimate of primary X-rays incident on a detector based on the detected total intensity and the estimated scatter intensity component. An imaging system adopting the above signal processing method is also disclosed.

Abstract: In accordance with the present disclosure, the present technique finds a diagnostic scan timing for a non-static object (e.g., a heart or other dynamic object undergoing motion) from raw scan data, as opposed to reconstructed image data. To find the scan timing, a monitoring scan of a patient's heart is performed. In the monitoring scan, the patient dose may be limited or minimized. As the projection data is acquired during such a monitoring scan, the projection data may be subjected to sinogram analysis in a concurrent or real-time manner to determine when to start (or trigger) the diagnostic scan.

Abstract: The present disclosure relates to the use of a hierarchical tomographic reconstruction approach that employs data representations in intermediate steps are between a full line integral and a voxel (e.g., an intermediate line integral). Each of the steps is progressively more local in nature and therefore has computational advantages and is also amenable to a deep learning solution using trained neural networks. The proposed hierarchical structure provides a mechanism to divide a large-scale inverse problem into a series of smaller-scale problems.

Abstract: The present approach relates to avoiding azimuthal blur in a computed tomography context, such as in dual energy imaging with fast kV switching. In accordance with certain aspects, the focal spot position is held stationary in the patient coordinate system within each respective view and the detector signals within the view are summed. In one embodiment, this results in the low and high energy views within the signal being collected from the same position within the patient coordinate system.

Abstract: A signal processing method is disclosed, which includes detecting a total intensity of X-rays passing through an object comprising multiple materials; obtaining at least one set of basis information of basis material information of the multiple materials and basis component information of photon-electric absorption basis component and Compton scattering basis component of the object; estimating a scatter intensity component of the detected X-rays based on the at least one set of basis information and the detected total intensity; and obtaining an intensity estimate of primary X-rays incident on a detector based on the detected total intensity and the estimated scatter intensity component. An imaging system adopting the above signal processing method is also disclosed.

Abstract: The present disclosure relates to image reconstruction with favorable properties in terms of noise reduction, spatial resolution, detail preservation and computational complexity. The disclosed techniques may include some or all of: a first-pass reconstruction, a simplified datafit term, and/or a deep learning denoiser. In various implementations, the disclosed technique is portable to different CT platforms, such as by incorporating a first-pass reconstruction step.

Abstract: The present approach relates to the use of machine learning and deep learning systems suitable for solving large-scale, space-variant tomographic reconstruction and/or correction problems. In certain embodiments, a tomographic transform of measured data obtained from a tomography scanner is used as an input to a neural network. In accordance with certain aspects of the present approach, the tomographic transform operation(s) is performed separate from or outside the neural network such that the result of the tomographic transform operation is instead provided as an input to the neural network. In addition, in certain embodiments, one or more layers of the neural network may be provided as wavelet filter banks.

Abstract: A method relates to the use of deep learning techniques, which may be implemented using trained neural networks (50), to estimate various types of missing projection or other unreconstructed data. Similarly, the method may also be employed to replace or correct corrupted or erroneous projection data as opposed to estimating missing projection data.