Colon Disease Classification Using Semi-Supervised GAN

The diagnosis of colon diseases is essential due to the rising prevalence of gastrointestinal disorders. However, challenges such as high image similarity, limited labeled data, and the complexity of distinguishing between conditions hinder effective classification. This study introduces a semi-supervised deep generative model that employs an Inception architecture in the generator to enhance diversity and improve the quality of feature representations. Additionally, we utilize an attention mechanism in the discriminator to strengthen important features and reduce classification errors. This method improves image generation quality and classification accuracy, achieving a testing accuracy of 95.66%. Furthermore, the low generator loss value of 0.08 indicates efficient learning and high-quality output.