Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11847721-B2
Publication Date
2023-12-19
Expiration Date
Abstract
Much of the image processing that is applied to medical images is a form of “inverse problem”. This is a class of mathematical problems in which a “forward” model by which a signal is converted into dataset is known, to at least some degree, but where the aim is to reconstruct the signal given the resulting dataset. Thus, an inverse problem is essentially seeking to discover x given knowledge of A(x)+noise by finding an appropriate reconstruction operator A† such that A† (A(x)+noise)≈x, thereby enabling us to obtain x (or a close approximation) given knowledge of an output dataset consisting of A(x)+noise. Generally, several such processes (or their equivalents) are applied to the image dataset. If the first process (for example, noise reduction) is expressed via a first reconstruction operator A1† characterised by a parameter set Θ1 and the second process (for example, segmentation) is expressed via a second reconstruction operator A2† characterised by a parameter set Θ2, then the result of the two steps applied consecutively is A2† (A1†(y)). This can be expressed as an overall reconstruction operator P+, characterised by a parameter set Φ. If we then allow a machine learning process to optimise P+, then the steps previously carried out separately can be combined into a single optimisation. This yields advantages in terms of computational load and in the accuracy of the end result.
Core Innovation
The invention relates to medical image processing that is formulated as successive inverse problems. Reconstruction, enhancement, noise reduction, segmentation, and treatment planning are described as usually solved separately by separate reconstruction operators, and this separate treatment is contrasted with a proposed approach that combines multiple inverse problems into a single optimisation using a unified inverse operator P+.
The proposed single inverse function is arranged to output a data set that can include cleaned or reconstructed volumes, segmented images, and/or radiotherapy treatment plans. Machine learning optimisation is used to optimise the unified inverse operator P+ so that the single inverse function concurrently solves multiple inverse problems rather than using separate operators for each task.
The document describes integrated process routes in which different combinations of inverse problems are consolidated, including joint reconstruction and noise reduction, noise reduction and segmentation, and segmentation and treatment planning. It further describes integrating raw image output to produce a deliverable treatment plan before radiation therapy dose delivery, including computation of radiotherapy-related outputs prior to radiation therapy dose administration.
Claims Coverage
The document contains two independent claims. Each independent claim covers a medical image analysis apparatus that computes a single inverse function before delivery of a radiation therapy dose to concurrently solve at least two inverse problems, with different input data types and different pairs of inverse problems.
Single inverse function for concurrent segmentation and treatment planning before dose delivery
A data store for receiving unsegmented output data of a medical imaging apparatus used for imaging a patient, wherein the output data comprises a patient structure prior to delivery of a radiation therapy dose to the patient; and a processing unit programmed to compute, before delivery of the radiation therapy dose, a single inverse function that is arranged to, when applied to the output data, output a data set comprising a treatment plan for administering the radiation therapy dose to the patient by using the single inverse function to concurrently solve at least two inverse problems including image segmentation and treatment planning.
Single inverse function for concurrent volume reconstruction and noise or artefact reduction before dose delivery
A data store for receiving raw unreconstructed output data of a medical imaging apparatus used for imaging a patient, wherein the output data is obtained prior to delivery of a radiation therapy dose to the patient; and a processing unit programmed to compute, before delivery of the radiation therapy dose, a single inverse function that, when applied to the raw unreconstructed output data, is arranged to produce a data set comprising a reconstructed volume image of a patient structure prior to delivery of the radiation therapy dose by using the single inverse function to concurrently solve at least two inverse problems including volume reconstruction and image noise and or artefact reduction.
Across both independent claims, the core coverage is the computation of a single inverse function before radiation therapy dose delivery that concurrently solves at least two inverse problems. One claim focuses on concurrently solving image segmentation and treatment planning to output a treatment plan, while the other claim focuses on concurrently solving volume reconstruction and noise or artefact reduction to output a reconstructed volume image.
Stated Advantages
Reducing computation by combining multiple inverse problems into a single optimisation using a unified inverse operator P+.
Improving end-result accuracy.
Documented Applications
Computing, before radiation therapy dose delivery, a treatment plan for administering the radiation therapy dose to a patient using outputs produced by the single inverse function.
Producing, before radiation therapy dose delivery, a reconstructed volume image of a patient structure prior to delivering the radiation therapy dose.
Interested in licensing this patent?