Techniques for generating radiotherapy treatment plans and establishing machine learning models for the generation and optimization of radiotherapy dose data are disclosed. An example method for generating a radiotherapy dose distribution using a generative model, trained in a generative adversarial network, includes: receiving anatomical data of a human subject that indicates a mapping of an anatomical area for radiotherapy treatment; generating radiotherapy dose data corresponding to the mapping with use of the trained generative model, as the generative model processes the anatomical data as an input and provides the dose data as output; and identifying the radiotherapy dose distribution for the radiotherapy treatment of the human subject based on the dose data. Another example method for training of the generative model includes establishing values of the generative model and a discriminative model of the generative adversarial network using adversarial training, including in a conditional generative adversarial network arrangement.

A computing system comprising a central processing unit (CPU), and memory coupled to the CPU and having stored therein instructions that, when executed by the computing system, cause the computing system to execute operations to generate a radiation treatment plan. The operations include accessing a minimum prescribed dose to be delivered into and across the target, determining a number of beams and directions of the beams, and determining a beam energy for each of the beams, wherein the number of beams, the directions of the beams, and the beam energy for each of the beams are determined such that the entire target receives the minimum prescribed dose. The operations further include prescribing a dose rate and optimizing dose rate constraints for FLASH therapy, and displaying a dose rate map of the FLASH therapy.

A method of generating a treatment plan for treating a patient with radiotherapy, the method includes obtaining a plurality of sample plans, which are generated by use of a knowledge base comprising historical treatment plans and patient data. The method also includes performing a multi-criteria optimization based on the plurality of sample plans to construct a Pareto frontier, where the plurality of sample plans are evaluated with at least two objectives measuring qualities of the plurality of sample plans such that treatment plans on the constructed Pareto frontier are Pareto optimal with respect to the objectives. The method further includes identifying a treatment plan by use of the constructed Pareto frontier.

A computing system comprising a central processing unit (CPU), and memory coupled to the CPU and having stored therein instructions that, when executed by the computing system, cause the computing system to execute operations to generate a radiation treatment plan. The operations include accessing a minimum prescribed dose to be delivered into and across the target, determining a number of beams and directions of the beams, and determining a beam energy for each of the beams, wherein the number of beams, the directions of the beams, and the beam energy for each of the beams are determined such that the entire target receives the minimum prescribed dose. The operations further include prescribing a dose rate and optimizing dose rate constraints for FLASH therapy, and displaying a dose rate map of the FLASH therapy.

Embodiments described herein provide for revising radiation therapy treatment plans, and in particular, revising beam angles used during radiation therapy treatment. A computer may receive a radiation therapy treatment plan based on a particular patient's diagnosis. The computer may use a machine learning model to revise radiation therapy treatment parameters such as a beam angle indicating a direction of radiation into the patient. The machine learning model may use reinforcement learning to optimize an initial beam angle from the radiation therapy treatment plan, revising the beam angle. The performance of the machine learning model is measured against metrics including fulfilling dosimetric clinical goals. The machine learning model may present the revised beam angle for display to a medical professional, or transmit the revised beam angle to downstream applications to further revise the radiation therapy treatment plan.

Nominal values of parameters, and perturbations of the nominal values, that are associated with previously defined radiation treatment plans are accessed. For each treatment field of the treatment plans, a field-specific planning target volume (fsPTV) is determined based on those perturbations. At least one clinical target volume (CTV) and at least one organ-at-risk (OAR) volume are also delineated. Each OAR includes at least one sub-volume that is delineated based on spatial relationships between each OAR and the CTV and the fsPTV for each treatment field. Dose distributions for the sub-volumes are determined based on the nominal values and the perturbations. One or more dose prediction models are generated for each sub-volume. The dose prediction model(s) are trained using the dose distributions.

Systems and methods are disclosed for performing operations comprising: receiving a plurality of training images representing different phases of a periodic motion of a target region in a patient; applying a model to the plurality of training images to generate a lower-dimensional feature space representation of the plurality of training images; clustering the lower-dimensional feature space representation of the plurality of training images into a plurality of groups corresponding to the different phases of the periodic motion; and classifying a motion phase associated with a new image of the target region in the patient based on the plurality of groups of the clustered lower-dimensional feature space representation of the plurality of training images.

Systems and methods can include a method for training a deep convolutional neural network to provide a patient radiation treatment plan, the method comprising collecting patient data from a group of patients, the patient data including at least one image of patient anatomy and a prior treatment plan, wherein the treatment plan includes predetermined machine parameters, and training a deep convolution neural network for regression by using the prior treatment plans and the corresponding collected patient data to determine a new treatment plan. Systems and methods can also include a method of using a deep convolutional neural network to provide a radiation treatment plan, the method comprising retrieving a trained deep convolution neural network previously trained on patient data from a group of patients, collecting new patient data, wherein the new patient data includes at least one image of patient anatomy, and determining a new treatment plan for the new patient using the trained deep convolutional neural network for regression, wherein the new treatment plan has a new set of machine parameters.

A method for delivering radiation treatment may include defining a preliminary trajectory including a plurality of control points. Each control point may be associated with position parameters of a gantry and a couch. The method may also include generating a treatment plan based on the preliminary trajectory by optimizing an intensity and position parameters of a collimator and MLC leaves for each control point. The method may also include decomposing the treatment plan into a delivery trajectory including the plurality of control points. Each of the plurality of control points may be further associated with the optimized intensity, the optimized position parameters of the collimator and the MLC leaves, an output rate, and a motion parameter of each of the gantry, the couch, the collimator, and the MLC leaves. The method may further include instructing a radiation delivery device to deliver the treatment plan according to the delivery trajectory.

A method, system and computer readable medium of: providing feature data of at least one organ at risk or target volume of said patient from a database of non-transitory data stored on a data storage device of prior patients data; generating, using a data processor, a distribution of dose points of the at least one organ at risk or target volume of said patient based on said feature data; calculating, using the data processor, at least one of (i) a probability of toxicity for the at least one organ at risk or (ii) a probability of treatment failure for the at least one target volume, based on said distribution of dose points; assessing, using the data processor, a dosimetric-outcome relationship based on the calculated probability; and automatically formulating, using the data processor, a treatment plan using the dosimetric-outcome relationship to minimize the at least one treatment-related risk.