Optimising Breast Cancer Identification and Categorisation by Hyperparameter Adjustment

Breast cancer is curable if caught early enough to stop it from spreading. In spite of this, misdiagnosis of breast cancer contributes to its continued status as the world's largest cause of mortality for women. Videos of breast cytology are analysed using deep learning frameworks to enhance early detection and categorisation. Layers based on convolution and pooling are used in these frameworks to extract features, and then dense layers are used for classification. To distinguish between malignant and benign cells, models such as neural networks using convolution and DenseNet undergo training, validation, and testing. In terms of both diagnosis and classification of breast tumours, the suggested approach performs better than CNN (Convolutional neural network) and DenseNet, despite different size of batches and learning rates. By increasing the effectiveness of these models through hyperparameter tuning, histological pictures can be used to make diagnoses that are more accurate.