The invention relates to the field of radiotherapy plan design, and discloses a radiotherapy parameter intelligent recommendation
system based on
disease species, target area volume and
dose characteristics, and the
system comprises the steps: constructing a response surface model of
machine physical parameters and plan quality indexes, generating a
Pareto optimal solution set through a non-dominated sorting
genetic algorithm, and building an optimization rule base; extracting anatomical and geometric features of historical cases, and training a
CART decision tree to establish a mapping relation between the features and clinical scenes; analyzing to-be-planned case features, matching a target scene through the
decision tree, and retrieving an optimal
machine physical parameter combination in the rule base according to the user
preference weight. According to the method, the width, the screw
pitch and the
modulation factor of the
collimator can be automatically recommended, the problem that manual
trial and error efficiency is low and depends on experience is solved, and balance between
treatment plan quality and
machine execution efficiency is achieved.