Network for medical image analysis, decision support system, and related graphical user interface (GUI) applications
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12224067-B1
Publication Date
2025-02-11
Expiration Date
Abstract
Described herein is a platform and supported graphical user interface (GUI) decision-making tools for use by medical practitioners and/or their patients, e.g., to aide in the process of making decisions about a course of cancer treatment and/or to track treatment and/or the progress of a disease.
Core Innovation
The invention provides a network-based decision support system and a corresponding method that receive and store a plurality of medical images in a database, each medical image associated with a corresponding patient. For a first patient, the system accesses one or more medical images comprising a computed tomography (CT) scan and a positron emission tomography (PET) scan, and automatically segments the CT scan using a convolutional neural network (CNN) to identify 3D boundaries of one or more regions of imaged tissue.
The system transfers the identified 3D boundaries to the PET scan, identifies one or more hotspots within the PET scan, and computes one or more risk index values using the PET scan and the identified 3D boundaries. Each risk index value is indicative of a cancer state or progression in the first patient, and a first risk index value is based on a total volume of identified hotspots within one or more of the identified 3D boundaries, or on a total number of identified hotspots within the one or more identified 3D boundaries.
In another aspect, the system computes a risk map comprising graphical denotations marking regions of risk of cancer or risk of recurrence of cancer using the PET scan and the identified 3D boundaries. In further embodiments, the PET scan is obtained following administration of an imaging agent comprising a radionuclide labelled prostate specific membrane antigen (PSMA) binding agent, where the radionuclide labelled PSMA binding agent comprises [18F]DCFPyL.
Claims Coverage
The independent claims define a network-based decision support system and method for CT-to-PET 3D boundary transfer, PET hotspot identification, and computation of cancer-state/progression risk indices. The claims include four inventive feature sets tied to hotspot-volume risk indices, hotspot-count risk indices, PSMA-targeted PET imaging with [18F]DCFPyL and a graphical risk map, and combined risk map plus risk index output.
CT-to-PET 3D boundary transfer for hotspot-based risk indices
Automatically segment the CT scan using a convolutional neural network (CNN) to identify 3D boundaries of one or more regions of imaged tissue; transfer the identified 3D boundaries to the PET scan; identify one or more hotspots within the PET scan; and compute, using the PET scan and the identified 3D boundaries, one or more risk index values indicative of a cancer state or progression in the first patient, wherein the first risk index value is based on a total volume of identified hotspots within one or more of the identified 3D boundaries.
CT-to-PET 3D boundary transfer for risk indices based on hotspot count
Automatically segment the CT scan using a convolutional neural network (CNN) to identify 3D boundaries of one or more regions of imaged tissue; transfer the identified 3D boundaries to the PET scan; identify one or more hotspots within the PET scan; and compute, using the PET scan and the identified 3D boundaries, one or more risk index values indicative of a cancer state or progression in the first patient, wherein the first risk index value is based on a total number of identified hotspots within one or more of the identified 3D boundaries.
PSMA radionuclide PET imaging with [18F]DCFPyL and graphical risk map
Access medical images associated with a first patient, including a CT scan and a PET scan obtained following administration of an imaging agent comprising a radionuclide labelled PSMA binding agent comprising [18F]DCFPyL; automatically segment the CT scan using a convolutional neural network (CNN) to identify 3D boundaries of one or more regions of imaged tissue; and compute, using the PET scan and the identified 3D boundaries, a risk map comprising graphical denotations marking regions of risk of cancer or risk of recurrence of cancer.
Risk map and risk index values produced together from CT-segmented 3D boundaries
Automatically segment the CT scan using a convolutional neural network (CNN) to identify 3D boundaries of one or more regions of imaged tissue; transfer the identified 3D boundaries to the PET scan; identify one or more hotspots within the PET scan; and compute, using the PET scan and the identified 3D boundaries, both a risk map comprising graphical denotations marking regions of risk of cancer or risk of recurrence of cancer and one or more risk index values indicative of a cancer state or progression in the first patient, wherein the first risk index value is based on a total volume of identified hotspots within one or more of the identified 3D boundaries.
The independent claims cover a unified computational framework in which CT is segmented with a CNN to define 3D boundaries, those boundaries are transferred to PET, PET hotspots are identified, and risk indices are computed as indicators of cancer state or progression. The claims further specify that the first risk index value can be based on a total volume or a total number of identified hotspots within the identified 3D boundaries, and they include embodiments that generate a graphical risk map of regions at risk of cancer or cancer recurrence, including a PSMA-targeted PET acquisition using a radionuclide labelled PSMA binding agent comprising [18F]DCFPyL.
Stated Advantages
Provides computed outputs including one or more risk index values indicative of a cancer state or progression.
Provides a risk map comprising graphical denotations marking regions of risk of cancer or risk of recurrence of cancer.
Enables generating a radiologist report based on one or more medical images and/or the risk map.
Documented Applications
Imaging and decision support for a first patient using CT and PET scans to compute cancer state/progression risk indices and, in some cases, a risk map marking regions at risk of cancer or risk of recurrence.
Time-series tracking of risk index values over a series of medical images taken over time.
Prostate imaging context, where the imaged tissue regions include a prostate region and the PET imaging uses a PSMA-targeted imaging agent.
Skeletal region imaging context, where the imaged tissue regions include a skeletal region.
Interested in licensing this patent?