医疗图像分割是进行精确诊断、医学图像理解分析的重要技术，有效进行医疗图像分割为定位病变组织和组织清晰化呈现提供重要辅助信息，其准确性直接对医生诊断、患者治疗效果评估产生影响。现有医疗图像处理有人工分割、传统方法和基于深度学习方法，人工分割耗费人力资源，同时受到医生经验限制。传统的医疗图像分割方法存在效率低、准确率低等局限性。基于深度学习的分割方法存在梯度消失或爆炸、过拟合或欠拟合，容易出现漏分割和误分割，导致分割准确率低，本文对基于卷积神经网络的医疗图像分割方法进行深入研究。具体研究内容概括如下：

（1）针对复杂医学图像纹理不清晰导致分割准确率低的问题，本文从加强原始图像特征和增强模型图像感知两个方面进行提升，提出全卷积注意力网络。利用Gabor滤波器进行预处理，提取图像局部空间和频率域信息，加强图像纹理信息；提出全卷积注意力网络，在全卷积网络中融合注意力机制，来对图像深度特征进行提取，通过增加目标区域权重来抑制背景干扰，增强图像的特征。实验表明，该算法有效减少误分割和漏分割，提高对复杂医学图像分割准确率。

（2）针对医疗图像特征变化大图像特征提取困难导致分割准确率低的问题，本文从提高对医疗图像特征提取和对提取特征充分利用进行提升，提出基于密集空洞卷积的医疗图像分割方法。首先设计密集空洞卷积模块，结合Inception结构思想和空洞卷积，捕获图像中多尺度特征信息。其次设计密集残差池化模块，通过不同大小残差池化块来检测图像多种尺度特征。最后结合端到端U-Net网络，在解码部分引入注意力机制，突出多特征图像特征。实验表明，该算法对多特征图像有较高分割准确率和稳定性。

Medical image segmentation is an important technology for accurate diagnosis and medical image understanding and analysis. Effective medical image segmentation provides important auxiliary information for locating lesions and presenting tissues clearly, and its accuracy has a direct impact on doctors' diagnosis and assessment of patients' treatment effects. Existing medical image processing includes manual segmentation, traditional methods and deep learning-based methods. Manual segmentation is human resource-intensive and limited by physicians' experience. Traditional medical image segmentation methods have limitations such as low efficiency and low accuracy. Deep learning-based segmentation methods suffer from gradient disappearance or explosion, overfitting or underfitting, and are prone to missed segmentation and mis-segmentation, resulting in low segmentation accuracy. In this paper, we conduct an in-depth study on medical image segmentation methods based on convolutional neural networks. The specific research is summarized as follows:

(1) To address the low segmentation accuracy due to unclear texture of complex medical images, this paper proposes a full convolutional attention network by enhancing both the original image features and the enhanced model image perception. The Gabor filter is used for pre-processing to extract the image local space and frequency domain information and enhance the image texture information; the full convolutional attention network is proposed, and the attention mechanism is fused in the full convolutional network to extract the image depth features, and the background interference is suppressed and the image features are enhanced by increasing the target region weights. Experiments show that the algorithm effectively reduces mis-segmentation and missed segmentation, and improves the accuracy of segmentation of complex medical images.

(2) To address the difficulty of image feature extraction due to the large variation of medical image features, this paper proposes a medical image segmentation method based on dense cavity convolution by improving the extraction of medical image features and making full use of the extracted features. Firstly, the dense cavity convolution module is designed to capture the multi-scale feature information in the image by combining the Inception structure idea and cavity convolution. Secondly, the dense residual pooling module is designed to detect multi-scale features in images by pooling blocks of different residual sizes. Finally, combining with end-to-end U-Net network, an attention mechanism is introduced in the decoding part to highlight multi-feature image features. Experiments show that the algorithm has high segmentation accuracy and stability for multi-feature images.

[1] 滕熠, 曹毛毛, 陈万青. 中国癌症筛查的发展、现状与挑战[J]. 中国肿瘤, 2022, 31(07): 481-487.

[2] Ott J J, Ullrich A, Miller A B. The importance of early symptom recognition in the context of early detection and cancer survival[J]. European Journal of Cancer, 2009, 45(16): 2743-2748.

[3] 杨健程, 倪冰冰. 医学3D计算机视觉:研究进展和挑战[J]. 中国图象图形学报, 2020, 25(10): 2002-2012.

[4] Chandy A. A review on iot based medical imaging technology for healthcare applications[J]. Journal of Innovative Image Processing (JIIP), 2019, 1(01): 51-60.

[5] 郑光远, 刘峡壁, 韩光辉. 医学影像计算机辅助检测与诊断系统综述[J]. 软件学报, 2018, 29(5): 1471-1514.

[6] Deshpande N M, Gite S, Aluvalu R. A review of microscopic analysis of blood cells for disease detection with AI perspective[J]. PeerJ Computer Science, 2021, 7: 1-21.

[7] Ishiwata T, Hasegawa F, Michishita M, et al. Electron microscopic analysis of different cell types in human pancreatic cancer spheres[J]. Oncology letters, 2018, 15(2): 2485-2490.

[8] 郭宁, 柏正尧. 注意力机制下密集空洞卷积的肺部图像分割[J]. 中国图象图形学报, 2021, 26(09): 2146-2155.

[9] Bhandary A, Prabhu G A, Rajinikanth V, et al. Deep-learning framework to detect lung abnormality–A study with chest X-Ray and lung CT scan images[J]. Pattern Recognition Letters, 2020, 129: 271-278.

[10] 胡玉龙. 基于多尺度领域自适应的医疗图像分割算法研究[D]. 吉林大学, 2021.

[11] 江宗康, 吕晓钢, 张建新, 张强, 魏小鹏. MRI脑肿瘤图像分割的深度学习方法综述[J]. 中国图象图形学报, 2020, 25(02): 215-228.

[12] 谢李鹏. 基于全卷积神经网络的心脏磁共振图像分割与量化分析[D]. 电子科技大学, 2021.

[13] Li X, Yu L, Chen H, et al. Transformation-consistent self-ensembling model for semisupervised medical image segmentation[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 32(2): 523-534.

[14] Zhu H, Liu Y, Gao X, et al. Combined CNN and Pixel Feature Image for Fatty Liver Ultrasound Image Classification[J]. Computational and Mathematical Methods in Medicine, 2022, 2022: 1-10.

[15] Abdul Gafoor S, Sampathila N, KS S. Deep learning model for detection of COVID-19 utilizing the chest X-ray images[J]. Cogent Engineering, 2022, 9(1): 2079221.

[16] 李峰, 孙艳丽, 郑长宏, 等. X 射线影像在正常骨组织及骨疾病诊断中的应用分析[J]. 中国组织工程研究, 2012, 16(35): 6591-6598.

[17] Mikołajczyk A, Grochowski M. Data augmentation for improving deep learning in image classification problem[C]//2018 international interdisciplinary PhD workshop (IIPhDW). IEEE, 2018: 117-122.

[18] Chan S C. Image-based rendering[M]//Computer Vision: A Reference Guide. Cham: Springer International Publishing, 2021: 656-664.

[19] 王雪. 基于神经网络的医学图像分割算法研究[D].吉林大学,2022.

[20] Adjabi I, Ouahabi A, Benzaoui A, et al. Past, present, and future of face recognition: A review[J]. Electronics, 2020, 9(8): 1-52.

[21] 付光华. 基于三维边缘检测和三维区域生长方法的CT图像脊椎区域的分割[D]. 扬州大学, 2019.

[22] Pujar J H, Gurjal P S, Kunnur K S. Medical image segmentation based on vigorous smoothing and edge detection ideology[J]. International Journal of Electrical and Computer Engineering, 2010, 4(8): 1143-1149.

[23] Bhargavi K, Jyothi S. A survey on threshold based segmentation technique in image processing[J]. International Journal of Innovative Research and Development, 2014, 3(12): 234-239.

[24] Kapur J N, Sahoo P K, Wong A K C. A new method for gray-level picture thresholding using the entropy of the histogram[J]. Computer vision, graphics, and image processing, 1985, 29(3): 273-285.

[25] Otsu N. A threshold selection method from gray-level histograms[J]. IEEE transactions on systems, man, and cybernetics, 1979, 9(1): 62-66.

[26] Beveridge J R, Griffith J, Kohler R R, et al. Segmenting images using localized histograms and region merging[J]. International Journal of Computer Vision, 1989, 2: 311-347.

[27] 吕鑫, 慕晓冬, 张钧. 基于改进麻雀搜索算法的多阈值图像分割[J]. 系统工程与电子技术, 2021, 43(02): 318-327.

[28] Huang Y L, Chen D R. Watershed segmentation for breast tumor in 2-D sonography[J]. Ultrasound in medicine & biology, 2004, 30(5): 625-632.

[29] Jemimma T A, Vetharaj Y J. Watershed algorithm based DAPP features for brain tumor segmentation and classification[C]//2018 International Conference on Smart Systems and Inventive Technology (ICSSIT). IEEE, 2018: 155-158.

[30] Soltani-Nabipour J, Khorshidi A, Noorian B. Lung tumor segmentation using improved region growing algorithm[J]. Nuclear Engineering and Technology, 2020, 52(10): 2313-2319.

[31] 郑洲, 张学昌, 郑四鸣, 施岳定. 基于区域增长与统一化水平集的CT肝脏图像分割[J]. 浙江大学学报(工学版), 2018, 52(12): 2382-2396.

[32] Bezdek J C, Ehrlich R, Full W. FCM: The fuzzy c-means clustering algorithm[J]. Computers & geosciences, 1984, 10(2-3): 191-203.

[33] Verma H, Verma D, Tiwari P K. A population based hybrid FCM-PSO algorithm for clustering analysis and segmentation of brain image[J]. Expert Systems with Applications, 2021, 167: 1-22.

[34] Zheng X, Lei Q, Yao R, et al. Image segmentation based on adaptive K-means algorithm[J]. EURASIP Journal on Image and Video Processing, 2018, 2018(1): 1-10.

[35] Madhukumar S, Santhiyakumari N. Evaluation of k-Means and fuzzy C-means segmentation on MR images of brain[J]. The Egyptian Journal of Radiology and Nuclear Medicine, 2015, 46(2): 475-479.

[36] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2015: 3431-3440.

[37] Li H , Fan J , Hua Q , et al. Biomedical sensor image segmentation algorithm based on improved fully convolutional network[J]. Measurement, 2022, 197: 1-9.

[38] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]//Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18. Springer International Publishing, 2015: 234-241

[39] Oktay O, Schlemper J, Folgoc L L, et al. Attention u-net: Learning where to look for the pancreas[J]. arXiv preprint arXiv:1804.03999, 2018.

[40] Zhou Z, Siddiquee M M R, Tajbakhsh N, et al. Unet++: Redesigning skip connections to exploit multiscale features in image segmentation[J]. IEEE transactions on medical imaging, 2019, 39(6): 1856-1867.

[41] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[42] Alom M Z, Hasan M, Yakopcic C, et al. Recurrent residual convolutional neural network based on u-net (r2u-net) for medical image segmentation[J]. arXiv preprint arXiv:1802.06955, 2018.

[43] Gu Z, Cheng J, Fu H, et al. Ce-net: Context encoder network for 2d medical image segmentation[J]. IEEE transactions on medical imaging, 2019, 38(10): 2281-2292.

[44] Goodfellow I J, Pouget-Abadie J, Mirza M, et al. Generative adversarial networks[J]. Commun. Acm, 2020, 63(11): 139-144.

[45] Zhang C, Song Y, Liu S, et al. MS-GAN: GAN-based semantic segmentation of multiple sclerosis lesions in brain magnetic resonance imaging[C]//2018 Digital Image Computing: Techniques and Applications (DICTA). IEEE, 2018: 1-8.

[46] Tomar N K, Jha D, Riegler M A, et al. Fanet: A feedback attention network for improved biomedical image segmentation[J]. IEEE Transactions on Neural Networks and Learning Systems, 2022: 1-14.

[47] Liao X, Li W, Xu Q, et al. Iteratively-refined interactive 3D medical image segmentation with multi-agent reinforcement learning[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 9394-9402.

[48] Cheplygina V, de Bruijne M, Pluim J P W. Not-so-supervised: a survey of semi-supervised, multi-instance, and transfer learning in medical image analysis[J]. Medical image analysis, 2019, 54: 280-296.

[49] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30: 1-11.

[50] 王金祥, 付立军, 尹鹏滨, 等. 基于 CNN 与 Transformer 的医学图像分割[J]. 计算机系统应用, 2022: 1-8.

[51] Zhang Y D, Satapathy S C, Guttery D S, et al. Improved breast cancer classification through combining graph convolutional network and convolutional neural network[J]. Information Processing & Management, 2021, 58(2): 1-25.

[52] Weng Y, Zhou T, Li Y, et al. Nas-unet: Neural architecture search for medical image segmentation[J]. IEEE access, 2019, 7: 44247-44257.

[53] LeCun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.

[54] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556, 2014.

[55] Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2015: 1-9.

[56] Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90.

[57] Huang G, Liu Z, Van Der Maaten L, et al. Densely connected convolutional networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 4700-4708.

[58] 盛文, 夏斌. 基于Gabor环滤波的纹理分割方法[J]. 红外与激光工程, 2003(05): 484-488.

[59] Noh H, Hong S, Han B. Learning deconvolution network for semantic segmentation[C]//Proceedings of the IEEE international conference on computer vision. 2015: 1520-1528.

[60] Cheng H, Lu J, Luo M, et al. PTANet: Triple Attention Network for point cloud semantic segmentation[J]. Engineering Applications of Artificial Intelligence, 2021, 102: 1-12.

[61] Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift[C]//International conference on machine learning. pmlr, 2015: 448-456.

[62] Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 2818-2826.

[63] Szegedy C, Ioffe S, Vanhoucke V, et al. Inception-v4, inception-resnet and the impact of residual connections on learning[C]//Proceedings of the AAAI conference on artificial intelligence. 2017, 31(1): 1-9.