FIRE-Breast: A Firefly Optimization-Driven EfficientNetB0 CNN Model for Optimized Breast Cancer Detection

Breast cancer (BC) is one of the most common causes of death in women, necessitating rapid and accurate diagnostic tests. This study offers a new Breast Cancer Detection (BCD) framework called FIRE-Breast and uses a Firefly Optimization Algorithm (FOA) with an EfficientNetB0 Convolutional Neural Network (CNN) model. The sole reason behind this is to obtain better results in detecting breast cancer and optimize the hyperparameters such as learning rates (LR), batch sizes (BS), and layer configuration for maximum accuracy with computational efficiency. The FIRE-Breast model was tested on histopathology images in breast cancer, reporting an accuracy of $\mathbf{8 5. 8 4 \%}$, outperforming baseline models such as EfficientNetB0. This showed a sensitivity and specificity of AUC of 0.86 for both benign and malignant classifications. In addition, the precisionrecall analysis showed an average precision of 0.81 and 0.80 for benign and malignant cases, respectively. These results confirm that the model can deliver reliable and effective detection of breast cancer, which is feasible for real-time clinical applications. FIREBreast reduces the computational overhead by including the Firefly Optimization Algorithm while sustaining diagnostic performance, opening venues for new research in medical image analysis and diagnostic support systems. Future work will extend the model to other medical imaging tasks and optimize its performance for broader clinical use.