A ConvNeXt Backbone-Based Medical Image Segmentation Model for Brain Glioma

Medical imaging is a very important information carrier, which can help clinicians understand the patient's condition and assist physicians in planning treatment plans. Medical image segmentation technology also plays a crucial role in clinical applications, such as lesion size measurement, organ and lesion localization, lesion shape, radiotherapy planning, assisted surgery, and anatomical structure research. At present, image segmentation techniques based on traditional image processing and deep learning are widely used in the automatic segmentation of medical images. The Deeplabv3+ algorithm framework is one of the algorithms with higher segmentation accuracy at present. However, the Deeplabv3+ algorithm framework is easy to lose local details when obtaining the global information of the image. Aiming at this problem of the Deeplabv3+ algorithm framework, this paper proposes a new image semantic segmentation algorithm, which integrates the ConvNeXt backbone with the Deeplabv3+ algorithm framework. The ConvNeXt backbone network architecture has fewer activation functions and normalization layers and separates the downsampling layer. The throughput advantage and robustness make ConvNeXt more valuable for industrial deployment. There is a significant improvement in the overall effect of the paper experiment.