为了解决超声图像斑点噪声、伪影、低图像对比度和图像亮度不均匀等问题，提出了一种改进的简化脉冲耦合神经网络(SPCNN)结合模糊互信息量的方法来自动检测乳腺肿瘤超声图像的感兴趣区域(ROI)。首先，对超声图像进行模糊增强预处理；然后，通过改进SPCNN对超声图像进行点火，以最大模糊互信息量作为最优判决准则，获得相应的分类结果；最后，对分类后的二值图像进行形态学等处理，从而得到乳腺超声图像的ROI。对包含118幅乳腺肿瘤超声图像的数据库进行处理，结果表明，该方法自动识别ROI准确率达到87.3%，处理每一幅图像的平均时间为4.68 s。本算法能有效快速地检测乳腺肿瘤超声图像的ROI，有望用于基于超声图像的乳腺肿瘤CAD中。

To solve the problems of the speckle noise, pseudo image, low contrast and luminous inhomogeneity in an ultrasound image, a method based on the improved Simplified Pulse Coupled Neural Network (SPCNN) combined with the fuzzy mutual information model was proposed to detect the Region of Interest(ROI) of the breast tumor ultrasound image. The ultrasound image was firstly mapped to the fuzzy sets to enhance the contrast, then the SPCNN model was used to pulse the ultrasound image, and the fuzzy mutual information was used as the optimization criterion to obtain the relative classification results. The ROI of the breast tumor ultrasound image was finally obtained by applying the morphologic processing on the corresponding classified results. The proposed segmentation method was performed on 118 breast tumor ultrasound images,and the obtained results show that the ROI accuracy is 87.3% and average processing time per image is 4.68 s. In conclusion, the proposed meth-od can be used to detect ROIs of breast tumor ultrasound images effectively and may have the potential applications in the breast tumor Computer Aided Diagnose(CAD) based on ultrasound images.