Reconstruction From Projection TRIZ Cases

Partitioning image and projection domains into high- and low-resolution regions reduces truncation artifacts while maintaining fast reconstruction speeds.

A transformer network predicts missing radial k-space spokes to reconstruct high-fidelity MRI images from undersampled data.

Iterative weight refinement bridges missing data between kVp views, resolving aliasing artifacts while preserving energy separation for cardiac imaging.

Fan blades segment radiation into narrow bands to estimate scattered radiation, reducing image artifacts in cone-beam CT.

Graph convolutional neural networks reconstruct full volumetric images from undersampled slice groups, reducing acquisition time and motion artifacts.

Reflective particle-beam imaging system generates high-resolution output images by combining diffraction-plane data with phase information from sample-plane scans.

Calculating effective length from multiple energy bins reduces noise influence on material decomposition accuracy in X-ray CT systems.

A signal processing system uses Compton scattering as a surrogate for electron density to reduce fitting parameters in iterative reconstruction.

Penalized maximum-likelihood reconstruction uses feature space priors to enhance PET image spatial resolution and signal-to-noise ratio.

A 3D image reconstruction operator incorporates scattered radiation modeling to improve dataset fidelity.

Resampling detector data onto a virtual flat array enables tangential filtering, reducing cone-beam artifacts while maintaining computational efficiency.

Energy-resolved detection and k-edge imaging enable motion compensation in computed tomography, reducing cardiac artifacts without slowing scan speeds.

A tomography system weights pixel values by traversing length to produce volume images.

A reconstruction stabilizer analyzes reliability metrics to define dynamic modifications for radioactive emission density distribution.

A deep learning method processes artifact-contaminated projection sequences to produce high-quality images from incomplete X-ray data.

Iterative neural correction of scatter radiation improves measurement precision in PET imaging, resolving the trade-off with processing time.

Preliminary focal plane tracking limits scanning to relevant regions, reducing photobleaching and improving signal-to-noise ratio.

First processing circuitry groups pre-compression data to generate grouped data and restoration data for efficient image reconstruction.

A generative model reconstructs complete 3D medical images by optimizing latent variables to minimize distance between image subsets.

Maximum intensity projection selects densest pixel values from cine CT frames to build attenuation correction images that remove motion artifacts in PET scans.

A fused ultrasound image displays a circular icon indicating the possible positions of an interventional device feature relative to the imaging plane.

A trained convolutional neural network predicts material mass densities from multi-energy computed tomography data to generate virtual monoenergetic images.

Segmenting the detector array into spectral and non-spectral regions reduces system cost while maintaining photon energy separation.

A motion tracking system combines medical imaging and sensors to capture physiological data for radiation therapy.

A magnetic resonance imaging method acquires data in opposite gradient directions to determine phase differences for correcting signal curves.

Estimate attenuation sinogram gradients from TOF PET emission data to reconstruct accurate maps, eliminating artifacts caused by CT-PET mismatches.

Synthetic projections augment sparse X-ray acquisitions to reduce view aliasing artifacts and patient radiation exposure.

Iterative CT reconstruction applies differential projection data with a non-quadratic correction operator to refine image details.

MLAA and DCC algorithms generate accurate mu-maps to resolve truncation artifacts in PET imaging.

Segmenting target anatomical structures into a synopsis volume reduces information density and reading time.

Depth segmentation in photon counting detectors extracts energy information from Compton events to enhance image quality.

A computed tomography reconstruction unit offsets pixel central positions from detector element centers to achieve uniform interpolation across the imaging area.

Energy-resolving detectors segment X-ray spectra to eliminate beam hardening artifacts and improve material composition analysis accuracy.

Sequential independent stage training reduces computation time and overfitting risk while improving MR image reconstruction quality.

Iterative image reconstruction converts projection data into high-fidelity images using regularization and inconsistency correction.

X-ray computed tomography processes projection data to determine blood flow dynamics including velocity and volume.

Iterative HYPR reconstruction updates a composite image to guide undersampled data processing.

A spectral imaging system generates estimated native image data by applying correction factors to account for displaced materials during acquisition.

Combining gradient information from multiple photopeaks during iterative reconstruction improves activity concentration accuracy and signal-to-noise ratio.

Generative adversarial networks create synthetic baggage scan images to train deep learning object detection algorithms.

Phase-contrast cone-beam CT achieves 25 lp/mm resolution without increasing radiation dose, reducing false positives in breast cancer detection.

Image signal processing converts photon counting projection data into path lengths using non-injective calibration curves.

A viewer application modifies a view-to-image transformation matrix to convert resolution-based coordinates into resolution-independent coordinates.

Machine learning replaces linear convolution kernels for undersampled k-space reconstruction, eliminating calibration scans and reducing noise amplification.

Multi-directional 2D dictionaries reduce computational cost while maintaining image quality in low-dose CT reconstruction.

A medical image processing apparatus separates projection data into basis material components to reconstruct accurate monochromatic X-ray images.

A distance-based weighting subsystem enhances near-skin structures in volume rendering by computing spatial proximity to detected features.

Selective resampling of coincidence events reduces noise amplification in low-count PET scans while maintaining image resolution.

Partitioning aperture positions into distinct regions allows simultaneous computation, reducing iteration time from 14 minutes to 5 seconds.

Hardware-triggered digital clocks timestamp movable component positions to eliminate data transmission latency and improve image reconstruction accuracy.