基于PSO-FCM的标记控制分水岭PCOS超声图像分割

PCOS Ultrasound Image Segmentation Based on Marker-Controled Watershed Optimized by PSO-FCM

多嚢卵巢综合征(PCOS)是一种严重危害女性健康的疾病。针对PCOS超声图像存在目标区域大小不一、对比度低,轮廓特征模糊等问题,提出了一种基于PSO-FCM的标记控制分水岭PCOS超声图像分割方法(PSO-FCM-Watershed)。首先对PCOS图像进行形态学开操作滤波和直方图均衡化预处理,然后使用经过粒子群优化算法(PSO)优化后的模糊C均值(FCM)方法聚类相邻像素,最后采用标记控制分水岭算法处理得到最终分割结果。实验结果表明,使用优化后的标记控制分水岭方法相比于单一的标记控制分水岭方法,加强了囊泡的轮廓特征,在平均交并比(MIOU)和平均准确率指标上,分别提升了4%和14%,达到63.5%和80.74%。可视化结果进一步说明该方法对囊泡区域轮廓的检测更灵敏,对边缘的分割更精准,有较理想的分割效果。

Polycystic ovary syndrome(PCOS)is a disease that seriously endangers women’s health.Aiming at the problems of PCOS ultrasound images,such as having different target areas,low contrast,and fuzzy contour features,a marker-controlled watershed PCOS ultrasound image segmentation method is proposed based on PSO-FCM(PSO-FCM-Watershed).Firstly,the PCOS image is preprocessed through morphological open filter denoising and histogram equalization,then the fuzzy C-means(FCM)method optimized by the particle swarm optimization(PSO)algorithm is used to cluster adjacent pixels,and finally the marker-controlled watershed algorithm is used to get the final segmentation result.The experimental results show that the use of the optimized marker-controlled watershed method enhances the contour features of vesicles compared with the single marker-controlled watershed method,and improves the mean intersection ratio(MIOU)and mean accuracy indicators by 4%and 14%.%,reaching 63.5%and 80.74%.The visualization results further show that the method is more sensitive to the detection of the contour of the vesicle region,more accurate to the edge segmentation,and has a better segmentation effect.