The embodiment of the invention provides a tumor clinical target area automatic sketching method and 
system based on a conditional random 
vector field, and the method comprises the following steps: S1, carrying out data discrete operation of an 
image sequence of a template CT, and generating a 
binary image of a tumor 
primary lesion region; s2, preprocessing a read 
primary lesion tumor data, defining a transfer characteristic function and a state characteristic function in a given 
data set, training a model, and determining parameters; and S3, reading 
primary tumor data of a new patient to obtain a corresponding 
binary image, and continuously iteratively calculating the 
state transition probability between adjacent voxels from the general 
tumor target region to the outside according to a set 
voxel state transition formula, thereby finally obtaining the global tumor probability invasion probability. According to the method, the conditional random 
vector field is used, the invasion probability of the tumor can be predicted, the radiotherapy clinical target area is automatically determined, and a prediction result is more reasonable and more suitable for the actual clinical situation.