Research On Brain Tumor Segmentation Algorithm Based On MRI Images

Brain tumor refers to a type of malignant or benign tumor that grows in brain tissue,posing a serious threat to human health and life.Magnetic Resonance Imaging(MRI)technology has been widely used in medical imaging due to its non-invasiveness,high resolution,and good soft tissue contrast.Therefore,to better diagnose and treat brain tumors at an early stage,it is crucial to use MRI imaging technology to understand the overall shape,size,and location of the tumor.Accurate segmentation of MRI brain tumor images can help doctors develop more precise and effective treatments.In this context,the research work of this paper is as follows:(1)In order to solve the problems of misidentification and uncertainty in the segmentation of the entire brain tumor region in MRI brain images,an improved method is proposed.The improved K-means algorithm is combined with a Hidden Markov Random Field(HMRF)model to achieve precise segmentation of brain tumor images.First,the improved K-means algorithm is used for initial parameter estimation and segmentation of the image,where the Manhattan-Chebyshev distance replaces the original Euclidean distance as the distance metric formula.Then,HMRF captures the global features of the image,which can better preserve the image details during segmentation.By combining the EM algorithm and iterative optimization of clustering centers,the clustering centers become more accurate.Experimental results show that the proposed method significantly improves the accuracy of brain tumor segmentation in MRI images.(2)Although the combination of HMRF and the improved K-means algorithm can segment the entire tumor area in MRI brain images,there are limitations in the core and enhanced areas of brain tumors.Therefore,this paper proposes a U-Net brain tumor segmentation algorithm that embeds a spatial convolutional pooling pyramid module with holes.The algorithm uses the U-Net encoder to extract features of different scales,and uses an improved ASPP module to capture context information at different scales.Finally,the decoder maps the feature maps back to the original image and uses skip connections to connect the feature maps of different layers in the encoder and decoder.The experiment shows that the proposed algorithm performs well in brain tumor segmentation and can more accurately extract various regions of the tumor than other deep learning image segmentation algorithms,providing strong support for doctors’ diagnosis and treatment.