Apparatus and process for medical imaging
Inventors
ABBOSH, Amin • AFASRI, Arman • Zamani, Ali • Bialkowski, Alina • ZHU, Guohun • NGUYEN, Thanh Phong • Guo, Lei • Wang, Yifan
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12625086-B2
Publication Date
2026-05-12
Expiration Date
Abstract
A process for medical imaging, the process including: (i) receiving scattering data representing mono-static or multi-static measurements of scattering of electromagnetic signals from tissues of a body part of a subject at a plurality of different signal frequencies, wherein electromagnetic signals are emitted from one or more antennas and the corresponding scattered signals are measured by the one or more antennas;(ii) processing the scattering data to calculate electric field power values at each of a plurality of scattering locations of the subject's tissues within the body part and for each of the plurality of frequencies;(iii) for each of the scattering locations, summing the calculated electric field power values at the scattering location over the plurality of frequencies and the plurality of antennas to generate an image of the tissues within the body part; and(iv) iteratively updating a model of the tissues within the body part based on a comparison of the model with the generated image until a termination criterion is satisfied, wherein the updated model is output as an image of the subject's tissues within the body part.
Core Innovation
A process is described for medical imaging that receives scattering data representing mono-static or multi-static measurements of scattering of electromagnetic signals from tissues of a body part of a subject at a plurality of different signal frequencies. The electromagnetic signals are emitted from one or more antennas and corresponding scattered electromagnetic signals are measured by the one or more antennas. The process processes the scattering data to calculate electric field power values at a plurality of scattering locations within the body part for each signal frequency, and then sums electric field power over the plurality of signal frequencies and the one or more antennas to generate an image of the tissues.
An iterative model update is performed by comparing the model with the generated image until a termination criterion is satisfied, and the updated model is provided as an output image of the subject’s tissues. The approach further includes processing the scattering data to calculate electric field power values at scattering locations and generating the image by summation across frequencies and antennas. Additional aspects described include pre-processing such as normalization and clutter removal, and calibration of scattering data using a matching-medium material or an average body part phantom in a reference imaging domain.
The described system also includes subject-specific template generation using machine learning on subject biodata and a 3D head surface, with spatial warping or registration to generate template data representing a tissue model. It further includes stroke classification/localization using a complex network representation of time series mapped to weighted/visibility graphs, computing graph degree mutual information features, and training a support vector machine to distinguish haemorrhagic from ischemic tissue, with localization based on weighted graph strength across opposed antenna pairs. Additional verification/refinement is described using local-area and global microwave tomography, including contrast source optimization and multi-frequency extensions, and combining outputs via multi-modal image fusion.
Claims Coverage
The independent claims cover an end-to-end medical imaging process and a corresponding apparatus. The claims include multiple inventive features, including mono-static or multi-static multi-frequency scattering acquisition, frequency-wise electric field power calculation at scattering locations, summation across frequencies and antennas to generate an image, iterative tissue model updating until a termination criterion is met, and optional refinements such as calibration, clutter removal, template generation with machine learning, abnormal-tissue classification, and opposed-antenna hemispheric comparison.
Mono-static or multi-static multi-frequency scattering imaging workflow
Receiving scattering data representing mono-static or multi-static measurements of scattering of electromagnetic signals from tissues at a plurality of different signal frequencies, where electromagnetic signals are emitted from one or more antennas and corresponding scattered electromagnetic signals are measured by the one or more antennas.
Frequency-wise electric field power calculation and multi-frequency/antenna summation image formation
Processing the scattering data to calculate electric field power values at each of a plurality of scattering locations and for each of the plurality of different signal frequencies, and for each scattering location summing the calculated electric field power values over the plurality of different signal frequencies and the one or more antennas to generate an image.
Iterative tissue model updating with termination criterion
Iteratively updating a model of the tissues based on a comparison of the model with the generated image until a termination criterion is satisfied, where the updated model is provided as an output image.
Multi-static selective antenna operation
Performing multi-static electromagnetic measurements using a plurality of antennas that selectively emit electromagnetic signals and measure corresponding scattered electromagnetic signals from the body part.
Template generation with machine learning selection and spatial warping/registration
Using machine learning on subject biodata to select a base template from a library of tissue models and geometrically transforming spatial coordinates using measurements of outer dimensions and/or shape to generate template data representing the subject’s tissue model.
Clutter removal by averaging and per-frequency subtraction
Normalising and removing clutter from scattering data, including averaging measured scattered electromagnetic signals and subtracting the average from each measurement at each frequency to remove strong reflections and clutter.
Calibration by dividing scattering parameters using a matching-medium or phantom reference
Calibrating scattering data by dividing measured scattering parameters from the body part by corresponding measured scattering parameters from an imaging domain without the body part, where the imaging domain is filled with a material having dielectric properties of a matching medium or with an average body part phantom.
Abnormal tissue classification via time-to-graph conversion and graph degree mutual information
Classifying abnormal tissues in the body part as haemorrhagic or ischemic by converting frequency-domain scattering data to time-domain data, mapping it to a graph, computing graph node degree and degree-sequence properties, calculating graph degree mutual information for similarity, training a classifier on graph degree mutual information feature vectors with class labels, and applying the classifier to graphs calculated for the subject’s tissues.
Hemispheric abnormality indication using opposed antenna pair comparison
Comparing scattering data from paired opposing antennas using multiple antennas to find significant hemispheric differences in the subject’s brain, indicating an abnormality in one hemisphere.
Overall, the independent claims establish a mono-static or multi-static, multi-frequency electromagnetic scattering imaging process (and apparatus) that computes electric field power at tissue scattering locations, forms an image by summing over frequencies and antennas, and refines a tissue model iteratively until a termination criterion is met. Dependent claim coverage further specifies multi-static operation, clutter removal, calibration against a matching-medium or phantom reference, subject-specific template generation with machine-learning selection and geometric transformation, and abnormal-tissue classification/localization including haemorrhagic versus ischemic discrimination and opposed-antenna hemispheric comparison.
Stated Advantages
Documented Applications
Medical imaging for generating images of tissues within a body part of a subject using mono-static or multi-static multi-frequency electromagnetic scattering measurements.
Stroke diagnosis, including haemorrhagic (ICH) versus ischemic (IS) classification and localization in the subject’s brain.
Verification/refinement of microwave tomography using local-area and global microwave tomography approaches, including contrast source inversion (CSI) and multi-frequency tomography methods.
Multi-modal image fusion that combines template-based output, beamography output, and tomography output.
Interested in licensing this patent?