A Novel Meta-Learning Ensemble Framework for Cancer Classification Using Convolution Neural Networks

Detecting cancer at an early stage is vital because it is a complicated global health crisis that killed 10 million people in 2018. Creating more accurate technology that provides information about the patient's cancer through early diagnosis helps clinicians make better-informed treatment decisions. This research delves deeply into a number of malignancies. A solid overview of machine-learning(ML) and deep-learning(DL) strategies employed in cancer research is also presented in this work. The paper also suggested a multi-neural ensemble learning strategy that relies on stacking to improve the anticipation performance on 8datasets. These datasets include prostate, non-small cell lung cancer survival, cervical, mesothelioma, and Wisconsin breast cancer. The 3real-time cancer datasets from the Jammu and Kashmir region-lung, ovarian, and leukemia-are also examined in this work. Our study's methodology achieved the highest level of anticipation accuracy across all types of cancer datasets, according to the simulation findings. Also, statistical validation of the proposed approach has been done. The researchers set out to create and test a model that could integrate anthropometric, clinical, imaging, and genetic data to make predictions about cancer risk. When compared to homogenous model-multi-modal and uni-modal frameworks, the simulation results show that the suggested framework is more accurate.