The invention provides a mammary gland auxiliary diagnosis system and method based on fusion depth characteristics, and relates to the technical field of medical image post-processing. The system comprises a preprocessing unit, a lump detection unit, a fusion depth feature extraction unit and a lump diagnosis unit, and is characterized in that an original mammary gland image is preprocessed, and amammary gland region is divided into a plurality of non-overlapped sub-regions; Mammary gland sub-region depth features are extracted by using a convolutional neural network CNN, and US- ELM is usedto cluster the depth characteristics of each sub-region to obtain mammary gland lumps and non-lump regions; a convolutional neural network CNN is used to extract lump depth features, lump morphology and texture features are extracted, and the lump depth, morphology and texture features are fused into a fused depth feature; And the fusion depth features are learned by using an extreme learning machine (ELM) to finally obtain benign and malignant diagnosis results of the lumps. The method is applied to breast auxiliary diagnosis, and accurate diagnosis of breast diseases can be effectively assisted.