Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12260507-B2
Publication Date
2025-03-25
Expiration Date
Abstract
A method for generating data representing the volume of part of a body, the method comprising generating a point distribution model “PDM” based on an input dataset comprising data representing at least one surface of part of a body, the PDM defining a surface model dataset based on an average dataset and one or more weight-eigenvector pairs, generating a first surface model dataset based on the PDM by modifying at least one weight of the one or more weight-eigenvector pairs, wherein the first surface model dataset is different from the average dataset, and generating an output volume dataset based on the first surface model dataset and a first reference dataset, the first reference dataset comprising data representing the volume of a corresponding part of a body, the output volume dataset comprising data representing a deformed volume of the corresponding part of the body.
Core Innovation
The invention relates to generating data representing the volume of part of a body by using a point distribution model "PDM" based on an input dataset comprising data representing at least one surface of part of a body. The PDM defines a surface model dataset for the part based on an average dataset and one or more weight-eigenvector pairs. By modifying at least one weight of the one or more weight-eigenvector pairs, the method generates different surface model datasets that each correspond to reference datasets and are different from the average dataset.
A mapping is then determined between the first surface model dataset and a second surface model dataset. The mapping is applied to a reference volume dataset comprising volume data representing the volume of a corresponding part of a body to produce an output volume dataset comprising data representing a deformed volume of the corresponding part of the body, where the first reference dataset is the same as the second reference dataset or is different from the second reference dataset.
The disclosed approach supports generating deformed body-part volume datasets that correspond to reference ROIs by varying the PDM weights to produce surface model datasets and by computing a mapping between surface models. The resulting deformed volume outputs are used to augment machine/deep learning training sets, including image segmentation training, to improve realism versus prior random deformations and to enable more robust artificial intelligence performance.
Claims Coverage
Two independent claims are provided. Each claim centers on the same core workflow: generate a PDM, create different surface model datasets by modifying PDM weights, determine a mapping between the surface model datasets, and apply the mapping to a reference volume dataset to generate an output volume dataset representing a deformed corresponding body-part volume.
Point distribution model from reference-weighted modes
Generating a point distribution model "PDM" based on an input dataset comprising data representing at least one surface of part of a body, wherein the PDM defines a surface model dataset for the part based on an average dataset and one or more weight-eigenvector pairs.
Reference-corresponding surface datasets by modifying PDM weights
Generating a first surface model dataset based on the PDM by modifying at least one weight of the one or more weight-eigenvector pairs such that the first surface model dataset corresponds to a first reference dataset, wherein the first surface model dataset is different from the average dataset; and modifying at least one weight of the one or more weight-eigenvector pairs to provide a second surface model dataset different from the first surface model dataset.
Mapping between surface model datasets
Determining a mapping between the first surface model dataset and the second surface model dataset.
Deformed output volume by applying the mapping to a reference volume dataset
Generating an output volume dataset for the part of the body by applying the mapping to a second reference dataset, the second reference dataset comprising volume data representing the volume of a corresponding part of a body, wherein the output volume dataset comprises data representing a deformed volume of the corresponding part of the body, and wherein the first reference dataset is the same as the second reference dataset or is different from the second reference dataset.
System implementation with instructions stored in memory and executed by a processor
A system comprising a memory that stores instructions and a processor that runs the instructions causing the processor to perform the method for generating the output volume dataset using the PDM, weight-modified surface model datasets, the mapping between the surface model datasets, and application of the mapping to a second reference dataset to generate a deformed output volume dataset.
Across both independent claims, the inventive contribution is a PDM-driven generation of deformed body-part volume data: construct a PDM from average and weight-eigenvector pairs, generate multiple surface model datasets by modifying weights to correspond to reference datasets, determine a mapping between those surface model datasets, and apply the mapping to a reference volume dataset to generate an output volume dataset representing a deformed corresponding volume.
Stated Advantages
Improves realism versus prior random deformations.
Enables more robust AI performance by augmenting machine/deep learning training sets, including for image segmentation.
Documented Applications
Used to augment machine/deep learning training sets, including for image segmentation.
Output volume datasets are used as training data for a deep learning process in an artificial intelligence system.
Interested in licensing this patent?