Lung and colon cancers are among the most prevalent and fatal tumors worldwide; prompt detection and precise categorization are vital. Recent developments in artificial intelligence (AI), especially in the areas of deep learning (DL), transfer learning (TL), explainable AI (XAI), vision transformers (ViT), and federated learning (FL), have shown great promise for enhancing the accuracy and openness of the interpretation of histopathological, CT Scan images. This review thoroughly examines 60 peer-reviewed papers published between 2022 and 2025, emphasizing AI in analyzing histopathological and CT Scan images to classify lung and colon malignancies. We explore various methods such as distributed Federated Learning (FL) frameworks, Vision Transformer (ViT) architectures, multiscale feature fusion, and ensemble approaches. The study systematically examines datasets, network designs, performance assessment metrics, and visual explanation tools like SHAP, Grad-CAM, and LIME to improve model interpretability. The review identifies key barriers, such as the need for AI systems that are both clinically relevant and interpretable, the lack of universal benchmarking standards, difficulties guaranteeing model adaptability to various datasets, and insufficient validation across several clinical facilities. This paper is intended to serve as a comprehensive reference for researchers and medical practitioners by consolidating the current landscape of AI-enabled histopathological, CT Scan cancer diagnosis and directing future investigations toward creating transparent, scalable, and resilient AI solutions in the field of oncology.