# US 12,599,285 B2 · Analysis of in-vivo images using connected graph components > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,599,285 B2 - Document ID: US12599285B2 - Status: Granted - Kind: B2 - Application number: 18/206,868 - Filed: 2023-06-07 - Published: 2023-12-14 - Granted: 2026-04-14 - Earliest claimed: 2022-06-14 - Assignees: Given Imaging Ltd. - Applicants: Given Imaging LTD - Inventors: Dorit Baras; Tomer Goldstein; Gal Shenker - Web page: https://patentgrep.com/patent/US12599285B2 ## CPC - A61B 1/000096: {using artificial intelligence} - A61B 1/000094: {extracting biological structures} - A61B 1/041: {Capsule endoscopes for imaging} - A61B 1/31: for the rectum, e.g. proctoscopes or sigmoidoscopes - G06T 7/0014: {using an image reference approach} - G06T 7/0016: {involving temporal comparison} - G06T 2200/24: involving graphical user interfaces [GUIs] - G06T 2207/10068: Endoscopic image - G06T 2207/20072: Graph-based image processing - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30032: Colon polyp - G06T 2207/30092: Stomach; Gastric - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A system includes at least one processor, and at least one memory storing instructions. The instructions, when executed by the processor(s), cause the system to: access a in-vivo images captured by a capsule endoscope where the in-vivo images are designated as containing a pathology indicator; form a connected graph component based on the in-vivo images where the connected graph component includes at least two images of the in-vivo images connected based on at least one indication that the at least two images contain the same occurrence of the pathology indicator; process the connected graph component with a machine learning model to provide a score indicative of whether the at least two images contain the same occurrence of the pathology indicator; and associate the score with the at least two images for presentation to a reviewer of the at least two images. ## Claims ### Claim 1 A system for analyzing images, comprising: at least one processor; and at least one memory storing instructions which, when executed by the at least one processor, cause the system to: access a plurality of in-vivo images of at least a portion of a gastrointestinal tract, the plurality of in-vivo images designated as including a pathology indicator, form a connected graph component based on the plurality of in-vivo images, the connected graph component comprising at least two images of the plurality of in-vivo images connected based on at least one indication that the at least two images contain a same occurrence of the pathology indicator, process the connected graph component with a machine learning model to provide a score indicative of whether the at least two images contain the same occurrence of the pathology indicator, and associate the score with the at least two images for a presentation to a reviewer of the at least two images. ### Claim 2 (depends on 1) The system of claim 1, wherein in forming the connected graph component based on the plurality of in-vivo images, the instructions, when executed by the at least one processor, cause the system to: for each image pair of the plurality of in-vivo images: analyze whether both images of the respective image pair contain a same occurrence of the pathology indicator, and based on the analysis indicating that both images of the respective image pair contain the same occurrence of the pathology indicator, designate both images of the respective image pair as graph nodes and connect the graph nodes by an edge; and select a group of graph nodes which are connected by one or more edges as the connected graph component. ### Claim 3 (depends on 2) The system of claim 2, wherein in analyzing whether both images of the respective image pair contain the same occurrence of the pathology indicator, the instructions, when executed by the at least one processor, cause the system to: process both images of the respective image pair by a triplet-loss network to provide an output score; and provide an indication of whether both images of the respective image pair contain the same occurrence of the pathology indicator based on comparing the output score to a predetermined score threshold. ### Claim 4 (depends on 1) The system of claim 1, wherein the instructions, when executed by the at least one processor, further cause the system to: determine at least one graph characteristic of the connected graph component, wherein processing the connected graph component with the machine learning model comprises inputting the at least one graph characteristic to the machine learning model. ### Claim 5 (depends on 4) The system of claim 4, wherein the at least one graph characteristic comprises at least one of: number of nodes of the connected graph component, number of edges of the connected graph component, ratio of nodes to edges in the connected graph component, or ratio of edges to nodes in the connected graph component. ### Claim 6 (depends on 1) The system of claim 1, wherein the instructions, when executed by the at least one processor, further cause the system to: communicate a user interface to a device of the reviewer, the user interface comprising a presentation of the at least two images and of the score associated with the at least two images. ### Claim 7 (depends on 6) The system of claim 6, wherein the user interface further comprises: based on a value of the score associated with the at least two images, one of: an indication that the at least two images may contain the same occurrence of the pathology indicator, or an indication that the at least two images may contain different occurrences of the pathology indicator. ### Claim 8 A method for analyzing images, comprising: accessing a plurality of in-vivo images of at least a portion of a gastrointestinal tract, the plurality of in-vivo images designated as including a pathology indicator; forming a connected graph component based on the plurality of in-vivo images, the connected graph component comprising at least two images of the plurality of in-vivo images connected based on at least one indication that the at least two images contain a same occurrence of the pathology indicator; processing the connected graph component with a machine learning model to provide a score indicative of whether the at least two images contain the same occurrence of the pathology indicator; and associating the score with the at least two images for a presentation to a reviewer of the at least two images. ### Claim 9 (depends on 8) The method of claim 8, wherein forming the connected graph component based on the plurality of in-vivo images comprises: for each image pair of the plurality of in-vivo images: analyzing whether both images of the respective image pair contain a same occurrence of the pathology indicator, and based on the analysis indicating that both images of the respective image pair contain the same occurrence of the pathology indicator, designating both images of the respective image pair as graph nodes and connecting the graph nodes by an edge; and selecting a group of graph nodes which are connected by one or more edges as the connected graph component. ### Claim 10 (depends on 9) The method of claim 9, wherein analyzing whether both images of the respective image pair contain the same occurrence of the pathology indicator comprises: processing both images of the respective image pair by a triplet-loss network to provide an output score; and providing an indication of whether both images of the respective image pair contain the same occurrence of the pathology indicator based on comparing the output score to a predetermined score threshold. ### Claim 11 (depends on 8) The method of claim 8, further comprising: determining at least one graph characteristic of the connected graph component, wherein processing the connected graph component with the machine learning model comprises inputting the at least one graph characteristic to the machine learning model. ### Claim 12 (depends on 11) The method of claim 11, wherein the at least one graph characteristic comprises at least one of: number of nodes of the connected graph component, number of edges of the connected graph component, ratio of nodes to edges in the connected graph component, or ratio of edges to nodes in the connected graph component. ### Claim 13 (depends on 8) The method of claim 8, further comprising: communicating a user interface to a device of the reviewer, the user interface comprising a presentation of the at least two images and of the score associated with the at least two images. ### Claim 14 (depends on 13) The method of claim 13, wherein the user interface further comprises: based on a value of the score associated with the at least two images, one of: an indication that the at least two images may contain the same occurrence of the pathology indicator, or an indication that the at least two images may contain different occurrences of the pathology indicator. ### Claim 15 A non-transitory processor-readable medium storing instructions which, when executed by at least one processor of a system, cause the system to: access a plurality of in-vivo images of at least a portion of a gastrointestinal tract, the plurality of in-vivo images designated as containing a pathology indicator, form a connected graph component based on the plurality of in-vivo images, the connected graph component comprising at least two images of the plurality of in-vivo images connected based on at least one indication that the at least two images contain a same occurrence of the pathology indicator, process the connected graph component with a machine learning model to provide a score indicative of whether the at least two images contain the same occurrence of the pathology indicator, and associate the score with the at least two images for a presentation to a reviewer of the at least two images. ### Claim 16 (depends on 15) The non-transitory processor-readable medium of claim 15, wherein in forming the connected graph component based on the plurality of in-vivo images, the instructions, when executed by the at least one processor, cause the system to: for each image pair of the plurality of in-vivo images: analyze whether both images of the respective image pair contain a same occurrence of the pathology indicator, and based on the analysis indicating that both images of the respective image pair contain the same occurrence of the pathology indicator, designate both images of the respective image pair as graph nodes and connect the graph nodes by an edge; and select a group of graph nodes which are connected by one or more edges as the connected graph component. ### Claim 17 (depends on 16) The non-transitory processor-readable medium of claim 16, wherein in analyzing whether both images of the respective image pair contain the same occurrence of the pathology indicator, the instructions, when executed by the at least one processor, cause the system to: process both images of the respective image pair by a triplet-loss network to provide an output score; and provide an indication of whether both images of the respective image pair contain the same occurrence of the pathology indicator based on comparing the output score to a predetermined score threshold. ### Claim 18 (depends on 15) The non-transitory processor-readable medium of claim 15, wherein the instructions, when executed by the at least one processor, further cause the system to: determine at least one graph characteristic of the connected graph component, wherein processing the connected graph component with the machine learning model comprises inputting the at least one graph characteristic to the machine learning model. ### Claim 19 (depends on 18) The non-transitory processor-readable medium of claim 18, wherein the at least one graph characteristic comprises at least one of: number of nodes of the connected graph component, number of edges of the connected graph component, ratio of nodes to edges in the connected graph component, or ratio of edges to nodes in the connected graph component. ### Claim 20 (depends on 15) The non-transitory processor-readable medium of claim 15, wherein the instructions, when executed by the at least one processor, further cause the system to: communicate a user interface to a device of the reviewer, the user interface comprising a presentation of the at least two images and of the score associated with the at least two images.