A HYBRID SUPERVISED FUSION DEEP LEARNING FRAMEWORK FOR MICROSCOPE MULTI-FOCUS IMAGES" won the Best Paper Award in the Special Session "3D Medical Image Processing, Quality Enhancement and Analysis". Award. The quality of multifocal microscope image fusion depends on the accuracy of image alignment, but there is a lack of alignment algorithms specifically for microscope images. Due to the existence of blurred regions and weak textures in multifocal microscope images, the image alignment results are not satisfactory. To address these issues, the paper proposes a hybrid supervised model for microscope multifocus image fusion. By introducing an artificial deformation field, the generalisation ability of the model to the actual deformation field is enhanced. The blurring of the multifocus microscope images is performed by simulating the lens motion, so as to simulate different blurring regions in the images to be aligned and enhance the stability of the model. Compared with some traditional alignment algorithms, the model extracts richer and more stable features of the multifocus image. Experimental results show that our model effectively improves the accuracy of microscope multifocal image alignment and fusion. This research is of great significance to the medical industry and is expected to provide doctors with more accurate and clear and complete panoramic images.