This research addresses the VGG model to test the healthcare dataset. The traditional VGG model for evaluating image data has not consistently provided good performance, as indicated by various tests on the dataset. Thus, this system proposes an advanced VGG model with specific approaches to enhance performance across various image datasets. The number of layers of the VGG model is tuned to improve the performance using freezing and unfreezing approaches. In this model, the proposed work incorporated a convolutional block with a backbone layer, batch normalization, and other enhancements to improve the unfreezing approach in the VGG model. These two approaches were demonstrated with the breast cancer dataset and evaluated through training and validation datasets. The experiment is conducted over multiple epochs, which helps improve performance during evaluation. Among the freezing and unfreezing approaches, we got better performance in the unfreezing approach. The accuracy of the training dataset under the unfreezing technique is 98.88%, which is improved by 7% with the freezing approach..