This paper presents a novel deep learning method for breast cancer classification using Convolutional Neural Networks (CNNs) and Particle Swarm Optimization (PSO). The primary challenge addressed is the accurate determination of hyperparameters and architectures for CNNs, which is crucial for effective image classification. The proposed method, called PSOCNN, uses PSO to optimize the hyperparameters of the CNN model, including kernel size, stride, and filter number. The effectiveness of the PSOCNN approach is demonstrated through experiments on two datasets: the Digital Database for Screening Mammography (DDSM) and the Mammographic Image Analysis Society (MIAS). The results show that the PSOCNN model achieved high accuracy rates of 98.23% and 97.98% on the DDSM and MIAS datasets, respectively, outperforming other methods such as CNN, YOLO-based CAD, and DBN-based CAD systems. The study highlights the potential of PSOCNN in automating the creation of CNN models for mammography classification, making it a powerful tool for breast cancer prediction. However, the research also acknowledges limitations, such as the specific datasets used and the need for further validation with additional pre-trained models and metaheuristic techniques.This paper presents a novel deep learning method for breast cancer classification using Convolutional Neural Networks (CNNs) and Particle Swarm Optimization (PSO). The primary challenge addressed is the accurate determination of hyperparameters and architectures for CNNs, which is crucial for effective image classification. The proposed method, called PSOCNN, uses PSO to optimize the hyperparameters of the CNN model, including kernel size, stride, and filter number. The effectiveness of the PSOCNN approach is demonstrated through experiments on two datasets: the Digital Database for Screening Mammography (DDSM) and the Mammographic Image Analysis Society (MIAS). The results show that the PSOCNN model achieved high accuracy rates of 98.23% and 97.98% on the DDSM and MIAS datasets, respectively, outperforming other methods such as CNN, YOLO-based CAD, and DBN-based CAD systems. The study highlights the potential of PSOCNN in automating the creation of CNN models for mammography classification, making it a powerful tool for breast cancer prediction. However, the research also acknowledges limitations, such as the specific datasets used and the need for further validation with additional pre-trained models and metaheuristic techniques.