Ensuring data privacy and security is a critical challenge in cancer diagnosis, driving the adoption of Federated Learning as a collaborative and privacy-preserving solution for tumour classification. Tumour classification plays a significant role in enabling precise diagnoses and treatments. However, centrally aggregating medical data raises concerns regarding the privacy and security of an individual. To tackle these challenges, we use Federated Learning, which offers a privacy-preserving approach that involves collaborative model training on decentralized data sources. This paper showcases the research contributions and gaps in the field of healthcare data, focusing on the significance of ethics. This survey introduces the concepts of federated learning, emphasizing healthcare studies highlighting the importance of tumor classification in healthcare decision-making. Moreover, this paper discusses the recent advancements in Federated Learning concerning tumor classification, including a wide variety of data such as medical imaging, genetics, and clinical data.