# US 12,112,470 B2 · Probabilistic tree tracing and large vessel occlusion detection in medical imaging > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,112,470 B2 - Document ID: US12112470B2 - Status: Granted - Kind: B2 - Application number: 17/449,263 - Filed: 2021-09-29 - Published: 2023-03-30 - Granted: 2024-10-08 - Earliest claimed: 2021-09-29 - Assignees: Siemens Healthineers AG - Applicants: Siemens Healthineers AG - Inventors: Bogdan Georgescu; Eli Gibson; Thomas Re; Dorin Comaniciu; Florin-Cristian Ghesu; Vivek Singh - Web page: https://patentgrep.com/patent/US12112470B2 ## CPC - G06T 7/0012: {Biomedical image inspection} - G06T 7/143: involving probabilistic approaches, e.g. Markov random field [MRF] modelling - G06T 7/162: involving graph-based methods - G06T 2207/10081: Computed x-ray tomography [CT] - G06T 2207/10088: Magnetic resonance imaging [MRI] - G06T 2207/10116: X-ray image - G06T 2207/10132: Ultrasound image - G06T 2207/20016: Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform - G06T 2207/20072: Graph-based image processing - G06T 2207/20076: Probabilistic image processing - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30016: Brain - G06T 2207/30101: Blood vessel; Artery; Vein; Vascular - G06T 2207/30172: Centreline of tubular or elongated structure - G06T 2207/30204: Marker - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract Systems and methods for generating a probabilistic tree of vessels are provided. An input medical image of vessels of a patient is received. Anatomical landmarks are identified in the input medical image. A centerline of the vessels in the input medical image is determined based on the anatomical landmarks. A probabilistic tree of the vessels is generated based on a probability of fit of the anatomical landmarks and the centerline of the vessels. The probabilistic tree of the vessels is output. ## Claims ### Claim 1 A method comprising: receiving an input medical image of vessels of a patient; identifying anatomical landmarks in the input medical image; determining a centerline of the vessels in the input medical image based on the anatomical landmarks; generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels; and outputting the probabilistic tree of the vessels. ### Claim 2 (depends on 1) The method of claim 1, wherein determining a centerline of the vessels in the input medical image based on the anatomical landmarks comprises generating a centerline probability map of the centerline of the vessels and wherein generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels comprises: extracting centerline segments from the centerline probability map; generating candidate mappings between the centerline segments and anatomical landmarks to the probabilistic tree; calculating a score for each candidate mapping based on a probability of detection of the anatomical landmarks and the centerline segments and a probability of location of the anatomical landmarks and the centerline segments; and selecting one of the candidate mappings based on the scores. ### Claim 3 (depends on 1) The method of claim 1, wherein generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels comprises: determining paths between the anatomical landmarks using RL (reinforcement learning) agents; and generating the probabilistic tree based on a likelihood of the paths. ### Claim 4 (depends on 3) The method of claim 3, wherein each of the RL agents is trained to navigate to a respective anatomical landmark based on geometric features and local image context features of the input medical image and wherein one or more of the RL agents share the geometric features and the local image context features with connected RL agents. ### Claim 5 (depends on 4) The method of claim 4, further comprising: determining an order for running the RL agents using a meta-agent based on a configuration of the RL agents. ### Claim 6 (depends on 1) The method of claim 1, further comprising: segmenting the vessels from the input medical image based on the centerline of the vessels; and computing geometric indices of the vessels based on the segmented vessels. ### Claim 7 (depends on 1) The method of claim 1, further comprising: detecting a location of a large vessel occlusion by identifying discontinuities in the probabilistic tree that result in missing downstream segments. ### Claim 8 (depends on 1) The method of claim 1, wherein identifying anatomical landmarks in the input medical image comprises: identifying the anatomical landmarks in the input medical image using RL (reinforcement learning) agents, each of the RL agents is trained to navigate to a respective anatomical landmark based on geometric features and local image context features of the input medical image and wherein one or more of the RL agents share the geometric features and the local image context features with connected RL agents. ### Claim 9 (depends on 8) The method of claim 8, wherein the geometric features comprise one or more of relative position, relative distance, relative angle, current agent location confidence, or embedded image intensity features. ### Claim 10 An apparatus comprising: means for receiving an input medical image of vessels of a patient; means for identifying anatomical landmarks in the input medical image; means for determining a centerline of the vessels in the input medical image based on the anatomical landmarks; means for generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels; and means for outputting the probabilistic tree of the vessels. ### Claim 11 (depends on 10) The apparatus of claim 10, wherein the means for determining a centerline of the vessels in the input medical image based on the anatomical landmarks comprises generating a centerline probability map of the centerline of the vessels and wherein the means for generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels comprises: means for extracting centerline segments from the centerline probability map; means for generating candidate mappings between the centerline segments and anatomical landmarks to the probabilistic tree; means for calculating a score for each candidate mapping based on a probability of detection of the anatomical landmarks and the centerline segments and a probability of location of the anatomical landmarks and the centerline segments; and means for selecting one of the candidate mappings based on the scores. ### Claim 12 (depends on 10) The apparatus of claim 10, wherein the means for generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels comprises: means for determining paths between the anatomical landmarks using RL (reinforcement learning) agents; and means for generating the probabilistic tree based on a likelihood of the paths. ### Claim 13 (depends on 12) The apparatus of claim 12, wherein each of the RL agents is trained to navigate to a respective anatomical landmark based on geometric features and local image context features of the input medical image and wherein one or more of the RL agents share the geometric features and the local image context features with connected RL agents. ### Claim 14 (depends on 13) The apparatus of claim 13, further comprising: means for determining an order for running the RL agents using a meta-agent based on a configuration of the RL agents. ### Claim 15 A non-transitory computer readable medium storing computer program instructions, the computer program instructions when executed by a processor cause the processor to perform operations comprising: receiving an input medical image of vessels of a patient; identifying anatomical landmarks in the input medical image; determining a centerline of the vessels in the input medical image based on the anatomical landmarks; generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels; and outputting the probabilistic tree of the vessels. ### Claim 16 (depends on 15) The non-transitory computer readable medium of claim 15, wherein determining a centerline of the vessels in the input medical image based on the anatomical landmarks comprises generating a centerline probability map of the centerline of the vessels and wherein generating a probabilistic tree of the vessels based on a probability of fit of the anatomical landmarks and the centerline of the vessels comprises: extracting centerline segments from the centerline probability map; generating candidate mappings between the centerline segments and anatomical landmarks to the probabilistic tree; calculating a score for each candidate mapping based on a probability of detection of the anatomical landmarks and the centerline segments and a probability of location of the anatomical landmarks and the centerline segments; and selecting one of the candidate mappings based on the scores. ### Claim 17 (depends on 15) The non-transitory computer readable medium of claim 15, the operations further comprising: segmenting the vessels from the input medical image based on the centerline of the vessels; and computing geometric indices of the vessels based on the segmented vessels. ### Claim 18 (depends on 15) The non-transitory computer readable medium of claim 15, the operations further comprising: detecting a location of a large vessel occlusion by identifying discontinuities in the probabilistic tree that result in missing downstream segments. ### Claim 19 (depends on 15) The non-transitory computer readable medium of claim 15, wherein identifying anatomical landmarks in the input medical image comprises: identifying the anatomical landmarks in the input medical image using RL (reinforcement learning) agents, each of the RL agents is trained to navigate to a respective anatomical landmark based on geometric features and local image context features of the input medical image and wherein one or more of the RL agents share the geometric features and the local image context features with connected RL agents. ### Claim 20 (depends on 19) The non-transitory computer readable medium of claim 19, wherein the geometric features comprise one or more of relative position, relative distance, relative angle, current agent location confidence, or embedded image intensity features.