Classification and Detection of Gastrointestinal Cancer Diagnosis Using Advanced Deep Learning Technique: An Explainable AI Approach
Keywords:
Gastrointestinal, Deep learning, Cancer diagnosis, Artificial intelligence, particle swarm optimization, Autoencoder.Abstract
Globally, gastric cancer ranks third in terms of cancer-related deaths. The histological interpretation of Gastric specimens is essential for clinical management, and this requires skilled pathologists. Deep learning systems have surpassed human pathologists in several domains. Current models are trained in a way that reduces the amount of knowledge acquired for the model and eliminates the need for extra stages to improve performance. Enhancements are necessary to strengthen the dependability of current methods when dealing with applications demanding accurate classification outcomes. In this work, we have proposed a gastrointestinal cancer diagnosis system with an explainable AI approach. The autoencoder has been employed for feature extraction, while Particle Swarm Optimization (PSO) selects the optimal features. Then, the DenseNet-169 architecture model has been employed for effective cancer diagnosis and classification. Finally, the Grad CAM is proposed to implement the explainable AI approach. The proposed system has been implemented and the performance is verified.
