Active visual attention models for computer vision tasks
US-2018060691-A1 · Mar 1, 2018 · US
US11568970B2 · US · B2
| Field | Value |
|---|---|
| Publication number | US-11568970-B2 |
| Application number | US-202017092487-A |
| Country | US |
| Kind code | B2 |
| Filing date | Nov 9, 2020 |
| Priority date | Nov 21, 2018 |
| Publication date | Jan 31, 2023 |
| Grant date | Jan 31, 2023 |
A practical reading order for non-experts. Skip the full description unless you need deep technical detail.
What the patent document calls the invention.
A short plain-language summary of the technical disclosure.
Who owns or filed the patent and who is credited as inventor.
Filing, priority, publication, and grant dates set the timeline.
The legal scope of protection — read this for what is actually claimed.
Technology tags used to group this patent with similar filings.
Prior art links and similar publications in this corpus.
Official abstract text for this publication.
A medical picture archive integration system includes a de-identification system that includes a first memory designated for protected health information (PHI), operable to perform a de-identification function is on a DICOM image, received from a medical picture archive system, to identify at least one patient identifier and generate a de-identified medical scan that does not include the at least one patient identifier. The medical picture archive integration system further includes a de-identified image storage system that stores the de-identified medical scan in a second memory that is separate from the first memory, and an annotating system, operable to utilize model parameters received from a central server to perform an inference function on the de-identified medical scan, retrieved from the second memory to generate annotation data for transmission to the medical picture archive system as an annotated DICOM file.
Opening claim text (preview).
What is claimed is: 1. A medical picture archive integration system, comprising: a receiver operable to receive data from a medical picture archive system; a transmitter operable to transmit data to the medical picture archive system; a network interface operable to bidirectionally communicate with a central server via a network, wherein the network does not include the medical picture archive system; a de-identification system that includes a first memory designated for protected health information (PHI), operable to: receive, from the medical picture archive system via the receiver, a first medical image, wherein the first medical image includes at least one patient identifier; and perform a de-identification function on the first medical image to identify the at least one patient identifier and generate a first de-identified medical scan that does not include the at least one patient identifier; a de-identified image storage system that includes a second memory that is separate from the first memory, operable to: receive the first de-identified medical scan from the de-identification system; and store the first de-identified medical scan in the second memory; and an annotating system, operable to: receive first model parameters from the central server via the network interface; retrieve the first de-identified medical scan from the second memory; utilize the first model parameters to perform a first inference function on the first de-identified medical scan to generate first annotation data; and generate a first annotated medical image file for transmission to the medical picture archive system via the transmitter, wherein the first annotated medical image file includes the first annotation data and further includes an identifier that indicates the first medical image. 2. The medical picture archive integration system of claim 1 , wherein the annotating system is further operable to: receive updated model parameters from the central server via the network interface; retrieve a second de-identified medical scan from the de-identified image storage system, wherein the second de-identified medical scan was generated by the de-identification system by performing the de-identification function on a second medical image received from the medical picture archive system; utilize the updated model parameters to perform the first inference function on the second de-identified medical scan to generate second annotation data; and generate a second annotated medical image file for transmission to the medical picture archive system via the transmitter, wherein the second annotated medical image file includes the second annotation data and further includes an identifier that indicates the second medical image. 3. The medical picture archive integration system of claim 1 , wherein the second memory stores a plurality of de-identified medical scans generated by the de-identification system by performing the de-identification function on a corresponding plurality of medical images received from the medical picture archive system via the receiver, wherein the plurality of de-identified medical scans is transmitted to the central server via the network interface, and wherein the central server generates the first model parameters by performing a training function on training data that includes the plurality of de-identified medical scans. 4. The medical picture archive integration system of claim 1 , wherein the central server generates the first model parameters by performing a training function on training data that includes a plurality of de-identified medical scans received from a plurality of medical picture archive integration systems via the network, wherein each of the plurality of medical picture archive integration systems communicates bidirectionally with a corresponding one of a plurality of medical picture archive systems, and wherein the plurality of de-identified medical scans corresponds to a plurality of medical images stored by the plurality of medical picture archive integration systems. 5. The medical picture archive integration system of claim 1 , wherein the first de-identified medical scan indicates a scan category of the first medical image, and wherein the annotating system is further operable to select the first inference function from a set of inference functions based on the scan category, and wherein the set of inference functions corresponds to a set of unique scan categories that includes the scan category. 6. The medical picture archive integration system of claim 5 , wherein each unique scan category of the set of unique scan categories is characterized by one of a plurality of modalities and one of a plurality of anatomical regions. 7. The medical picture archive integration system of claim 5 , wherein the de-identification system is further operable to: receive a plurality of medical image data from the medical picture archive system via the receiver for storage in the first memory in response to a query transmitted to the medical picture archive system via the transmitter, wherein the query is generated by the medical picture archive integration system in response to a request indicating a new scan category received from the central server via the network, wherein the new scan category is not included in the set of unique scan categories, and wherein the plurality of medical image data corresponds to the new scan category; and perform the de-identification function on the plurality of medical image data to generate a plurality of de-identified medical scans for transmission to the central server via the network; wherein the annotating system is further operable to: receive second model parameters from the central server via the network for a new inference function corresponding to the new scan category; update the set of inference functions to include the new inference function; receive a second de-identified medical scan from the de-identified image storage system, wherein the second de-identified medical scan was generated by the de-identification system by performing the de-identification function on a second medical image received from the medical picture archive system; select the new inference function from the set of inference functions by determining the second de-identified medical scan indicates the second medical image corresponds to the new scan category; utilize the second model parameters to perform the new inference function on the second de-identified medical scan to generate second annotation data; and generate a second annotated medical image file for transmission to the medical picture archive system via the transmitter, wherein the second annotated medical image file includes the second annotation data and further includes an identifier that indicates the second medical image. 8. The medical picture archive integration system of claim 5 , wherein the medical picture archive integration system generates parameter data for transmission to the medical picture archive system that indicates the set of unique scan categories, and wherein the medical picture archive system automatically transmits the first medical image to the medical picture archive integration system in response to determining that the first medical image compares favorably to one of the set of unique scan categories. 9. The medical picture archive integration system of claim 1 , further comprising a second transmitter operable to transmit data to a report database associated with the medical picture archive system, wherein the annotating system is further operable to: generate natural language report data based on the first annotation data; and transmit, via the second transmitter, the natural langua
References adjustable by an adaptive method, e.g. learning · CPC title
Lung · CPC title
Physics · mapped topic
for calculating health indices; for individual health risk assessment · CPC title
Segmentation; Edge detection (motion-based segmentation G06T7/215) · CPC title
Related publications grouped by family.
Answers are generated from the same data shown on this page.