The invention discloses a generation method and device for synthetic lesion image, equipment and a readable storage medium, and the method comprises the steps: generating a pathological descriptor according to a lesion reference map, a lesion detection network and a symmetrical activation network, the pathological descriptor containing lesion characteristics; and inputting a physiological featuremap and the random vector into a generation network model trained by the pathological descriptor to generate a synthetic lesion map. in the scheme, the pathological descriptors are added, deleted, modified and the like to manipulate the positions and the number of the lesions in the synthetic lesion map so as to generate diverse results. Therefore, the scheme is more targeted for lesion synthesis,synthesis in the aspect of physiological details is vivid enough, and the requirement for enlarging the scale of a deep learning detector training set can be met.