A Robust Ensemble Learning Approach for Medical Image Segmentation with Transfer Learning

This paper proposes a robust ensemble mastering approach for clinical picture segmentation. The proposed technique combines a convolution neural community (CNN) with a switch studying-based totally ensemble model. The CNN is pre-educated with a dataset containing clinical photo modalities, particularly Magnetic Resonance Imaging (MRI) and computed tomography (CT)., a weight-averaged ensemble model is acquired between the two scientific photograph modalities. This version is an initial way to the clinical image segmentation problem. Eventually, the ensemble model is great-tuned with additional imaging statistics from an unmarried modality to improve the segmentation accuracy. The proposed technique is evaluated on a range of datasets, and the effects display that it achieves competitive performance compared to the country of the artwork. Moreover, the experiments reveal that combining the transfer gaining knowledge of-based totally ensemble mastering approach with extra imaging information enhances the accuracy of medical picture segmentation.