Automatic design method of radiotherapy plan

Abstract

The invention discloses an automatic design method of a radiotherapy plan. The automatic design method comprises the steps that various kinds of same-type tumor treatment plans are acquired and randomly sorted into a first data set and a second data set; a neural network prediction model is built; the first data set and the second data set are optimized circularly; the above steps are repeated till the first data set or the second data set is more superior than or similar to the acquired prediction result to acquire the optimal first data set and the optimal second data set, and the optimal same-type tumor treatment plan is obtained; treatment information is extracted from the optimal same-type tumor treatment plan; a neural network prediction model is obtained based on treatment information training to acquire an optimized prediction model; and the treatment information of a target patient is acquired, the optimized prediction model is input, and the radiotherapy plan of the target patient is automatically acquired.

Description

technical field [0001] The invention belongs to the field of radiotherapy, in particular to an automatic design method for radiotherapy plans. Background technique [0002] In radiology research, how to increase the irradiation dose of the target area and reduce the toxicity of normal tissues is the focus of attention. The rapid development of three-dimensional conformal technology, beam intensity modulation technology and volume arc intensity modulation technology in recent years has made it possible to achieve a highly conformal dose distribution and better protect normal tissues. Intensity-modulated radiation therapy is the mainstream technology of tumor radiation therapy today, and its smooth development requires a dedicated radiation therapy plan design system and professional plan designers. The design of radiation therapy plan is a process of simulating the actual treatment environment of the patient, rationally arranging the radiation field, and calculating the radi...

AI Technical Summary

Problems solved by technology

Therefore, there are many parameters to be considered when using the above method to model, and it is difficult to select an accurate and reasonable parametric model. There are large differences in the predicted results between different models, which greatly reduces the accuracy and scope of application.

Moreover, the above methods are difficult to ensure the quality of the treatment plan database used for modeling, which ultimately leads to uncertainty in the prediction results

[0006] The above shortcomings of existing methods cannot meet the needs of clinical applications, so it is urgent to develop new automatic design methods for treatment plans

Images