System and method for radiation therapy treatment planning
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-10864382-B2
Publication Date
2020-12-15
Expiration Date
Abstract
Radiation therapy treatment planning methods facilitate predicting achievable dose distributions by training a prediction model for clinical DVH curves based on simplified DVH curves obtained using standardized methods. Machine learning applied on pairs of clinical and simplified DVH curves enables predicting actual clinical DVH curves based on simplified DVH curves.
Core Innovation
A radiation therapy treatment planning method obtains a protocol plan created by applying a plan generation protocol to patient data that includes at least one planning image. From the protocol plan, it calculates at least one protocol dose volume histogram (DVH) curve, where each protocol DVH curve relates a dose distribution for the protocol plan to a volume of a structure in a structure set associated with the at least one planning image. The method then uses the protocol DVH curves as a basis to estimate clinical outcomes for the current patient.
The method obtains estimated clinical DVH curves by sending the at least one protocol DVH curve as a query to a prediction model. The prediction model determines the estimated clinical DVH curves based on a relationship between previous protocol DVH curves and previous clinical DVH curves for a collection of previous patients. The estimated clinical DVH curves are determined using clinical DVH curves from previous patients whose protocol DVH curves have a distance to the at least one protocol DVH curve of the current patient below a specified threshold value.
At least one of the estimated clinical DVH curves is used in treatment planning for the current patient. A corresponding training method trains the prediction model by obtaining protocol plan DVH curves and associated clinical DVH curves for previous patients, forming pairs of protocol DVH curves and clinical DVH curves, selecting a subset of the pairs, and training a machine learning model to map protocol DVH curves of a current patient to estimated clinical DVH curves. The training involves adjusting system parameters of the machine learning model toward minimization of a deviation between estimated clinical DVH curves and true clinical DVH curves for the previous patients.
Claims Coverage
The document provides two independent claims, centered on learning and using a relationship mapping protocol DVH curves to estimated clinical DVH curves with distance-threshold selection of similar prior cases.
Protocol plan to protocol DVH for current patient
Obtain a protocol plan created by applying a plan generation protocol to patient data for a current patient comprising at least one planning image, and calculate at least one protocol DVH curve relating a dose distribution for the protocol plan to a volume of a structure in a structure set associated with the at least one planning image.
Prediction model query using distance-threshold similar prior DVHs
Obtain estimated clinical DVH curves by sending the at least one protocol DVH curve as a query to a prediction model arranged to determine estimated clinical DVH curves based on a relationship between previous protocol DVH curves and previous clinical DVH curves for a collection of previous patients, wherein the estimated clinical DVH curves are determined based on clinical DVH curves from previous patients that have protocol DVH curves having a distance to the at least one protocol DVH curve of the current patient below a specified threshold value.
Using estimated clinical DVHs in treatment planning
Using at least one of the estimated clinical DVH curves in treatment planning for the current patient.
Training prediction model from protocol/clinical DVH pairs and deviation minimization
Train a machine learning model using pairs including a protocol DVH curve and an associated clinical DVH curve of previous patients, select a subset of the pairs of DVH curves, and train the machine learning model configured to take protocol DVH curves of a current patient as input and return estimated clinical DVH curves, the process of training involving adjusting system parameters of the machine learning model toward minimization of a deviation between estimated clinical DVH curves and true clinical DVH curves for the previous patients.
Overall claim coverage centers on generating protocol DVH curves from a protocol plan, querying a prediction model that uses a relationship to previous protocol/clinical DVH pairs selected by a distance threshold, and using the resulting estimated clinical DVH curves in treatment planning; the second independent claim covers training that prediction model from paired DVHs with subset selection and deviation minimization.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?