Transfer Learning-Based Convolution Neural Network for Differentiation between Benign and Malignant Cancer Cells Using MobileNetV2
Goutam Sarker · International Journal of Computer Science and Engineering · 2025
The accurate classification of medical images plays a pivotal role in early cancer diagnosis and treatment planning. This study presents a robust image classification framework leveraging transfer learning with MobileNetV2 for binary classification of histopathological images as benign or malignant. The proposed model incorporates data augmentation, dropout regularization, and batch normalization to address overfitting and enhance generalization on limited datasets. The model is evaluated using a confusion matrix and performance metrics. Furthermore, the model is extended to predict unknown samples from a dedicated prediction folder. The results demonstrate high classification accuracy of 80% with low validation loss of 0.6501, indicating the effectiveness of the transfer learning strategy in medical image diagnostics.