Transfer Learning with Deep Convolutional Neural Networks for Esophageal Cancer Classification

Esophageal cancer is a critical health concern requiring accurate and timely diagnosis. This study investigates the use of transfer learning with pretrained convolutional neural networks (CNNs) for classifying endoscopic images. Six state-of-the-art models EfficientNetB7, VGG19, DenseNet121, ResNet50, InceptionV3, and InceptionResNetV2 were fine-tuned and evaluated on key performance metrics. EfficientNetB7 achieved the highest accuracy of $99.7 \%$, followed by VGG19 (99.7%), DenseNet121 (99.7%), ResNet50 (99.5%), InceptionV3 (98.7%), and InceptionResNetV2 (98.3%). This performance underscores the efficacy of transfer learning in leveraging feature-rich pretrained models for medical image classification. The proposed approach demonstrates its potential as a robust and efficient diagnostic tool for automated esophageal cancer detection, supporting clinical decision-making with high precision.