基于空间信息关注和纹理增强的短小染色体分类方法

doi:10.11996/JG.j.2095-302X.2024051017

摘要/Abstract

摘要：

染色体分类是核型分析中的重要任务之一。尽管残差神经网络已经在染色体分类领域取得了显著成就，但由于部分染色体具有长度较短、分类特征难以识别以及形态相似度较高的特点，使得其分类仍然具有挑战性。提出了基于空间信息关注和纹理增强的染色体分类模型(SIATE-Net)，以Inception_ResNetV2模型作为骨干网络提取染色体的深层特征，自注意力机制和深度可分离卷积的引入能够更好地关注和保留短小染色体的空间信息。染色体长度较短易造成显带信息混淆，模型融入了纹理增强机制以扩大染色体间的差异性，为分类任务增加更多的判定依据。SIATE-Net模型分别在私人数据集与公开数据集上进行验证，分类性能明显优于其他方法，尤其是短小染色体。在私人数据集上，SIATE-Net模型表现出了最佳的总体分类准确率98.05%，短小染色体分类精度高达97.42%。在公开数据集上，SIATE-Net模型的总体分类准确率为98.95%，而短小染色体也达到了98.51%。实验结果表明，具有较强针对性的自注意力模块、深度可分离卷积和纹理增强模块在不降低整体分类准确性的前提下，能够有效地解决短小染色体分类任务。

关键词: 医学图像处理, 短小染色体分类, Inception_ResNetV2模型, 自注意力机制, 深度可分离卷积, 纹理增强

Abstract:

Chromosome classification is a crucial task in karyotype analysis. Despite the significant achievements made by residual neural networks in chromosome classification, the classification still presents challenges due to the short length, difficult-to-identify classification features, and high morphological similarity of certain chromosomes. To address this issue, a spatial information attention and texture enhancement network (SIATE-Net) model was proposed for chromosome classification. The SIATE-Net model utilized the Inception_ResNetV2 model as its backbone network to extract deep features of chromosomes. By introducing self-attention mechanisms and depth-wise separable convolution, the model could better focus on and retain the spatial information of short chromosomes. The short length of certain chromosomes often leads to confusion in banding information. To mitigate this issue, the model integrated a texture enhancement mechanism to amplify the differences between chromosomes, providing the model with more discriminative features for classification. The SIATE-Net model was validated on both private and public datasets, demonstrating superior classification performance compared to other methods, especially in classifying short chromosomes. On the private dataset, the SIATE-Net model achieved the best overall classification accuracy of 98.05%, with a high accuracy of 97.42% for short chromosomes. On the public dataset, the overall classification accuracy of the SIATE-Net model was 98.95%, with short chromosomes reaching an accuracy of 98.51%. Experimental results demonstrated that the targeted self-attention module, depth-wise separable convolution, and texture enhancement module effectively addressed the classification of short chromosomes without compromising overall classification accuracy.

Key words: medical image processing, short chromosome classification, Inception_ResNetV2 model, self-attention mechanism, depth-wise separable convolution, texture enhancement

中图分类号:

彭文, 林金炜. 基于空间信息关注和纹理增强的短小染色体分类方法[J]. 图学学报, 2024, 45(5): 1017-1029.

PENG Wen, LIN Jinwei. A short chromosome classification method based on spatial attention and texture enhancement[J]. Journal of Graphics, 2024, 45(5): 1017-1029.

图/表 18

参考文献 33

[1]	XIAO L, LUO C L. DEEPACC: automate chromosome classification based on metaphase images using deep learning framework fused with Priori knowledge[C]// 2021 18th International Symposium on Biomedical Imaging. New York: IEEE Press, 2021: 607-610.
[2]	LIN C C, ZHAO G S, YIN A H, et al. A novel chromosome cluster types identification method using ResNeXt WSL model[J]. Medical Image Analysis, 2021, 69: 101943.
[3]	HU X, YI W L, JIANG L, et al. Classification of metaphase chromosomes using deep convolutional neural network[J]. Journal of Computational Biology, 2019, 26(5): 473-484. DOI PMID
[4]	张林, 易先鹏, 王广杰, 等. 基于网格重构学习的染色体分类模型[EB/OL]. [2024-01-23]. http://www.aas.net.cn/cn/article/doi/10.16383/j.aas.c210303.
	ZHANG L, YI X P,. WANG G J, et al. A grid reconstruction learning model for chromosome classification[EB/OL]. [2024-01-23]. http://www.aas.net.cn/cn/article/doi/10.16383/j.aas.c210303 (in Chinese).
[5]	QIN Y L, WEN J, ZHENG H, et al. Varifocal-Net: a chromosome classification approach using deep convolutional networks[J]. IEEE Transactions on Medical Imaging, 2019, 38(11): 2569-2581. DOI PMID
[6]	SHARMA M, SAHA O, SRIRAMAN A, et al. Crowdsourcing for chromosome segmentation and deep classification[C]// IEEE Conference on Computer Vision and Pattern Recognition Workshops. New York: IEEE Press, 2017: 786-793.
[7]	ZHANG W B, SONG S F, BAI T M, et al. Chromosome classification with convolutional neural network based deep learning[C]// 2018 11th International Congress on Image and Signal Processing, Biomedical Engineering and Informatics. New York: IEEE Press, 2018: 1-5.
[8]	REMYA R S, HARIHARAN S, VINOD V, et al. A comprehensive study on convolutional neural networks for chromosome classification[C]// 2020 Advanced Computing and Communication Technologies for High Performance Applications. New York: IEEE Press, 2020: 287-292.
[9]	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 770-778.
[10]	翟进有, 代冀阳, 王嘉琦, 等. 深度残差网络的无人机多目标识别[J]. 图学学报, 2019, 40(1): 158-164. DOI
	ZHAI J Y, DAI J Y, WANG J Q, et al. Multi-objective identification of UAV based on deep residual network[J]. Journal of Graphics, 2019, 40(1): 158-164 (in Chinese).
[11]	易序晟, 尹爱华, 黄杰晟, 等. 深度学习下主流染色体分类算法的性能评估[J]. 中国图象图形学报, 2023, 28(2): 570-588.
	YI X S, YIN A H, HUANG J S, et al. Performance evaluation of mainstream chromosome recognition algorithms under deep learning[J]. Journal of Image and Graphics, 2023, 28(2): 570-588 (in Chinese).
[12]	张成成, 宋婕萍, 徐淑琴, 等. 基于深度卷积神经网络对中期染色体分类的应用研究[J]. 中国临床新医学, 2020, 13(2): 123-126.
	ZHANG C C, SONG J P, XU S Q, et al. Application research on classification of metaphase chromosomes based on deep convolutional neural networks[J]. Chinese Journal of New Clinical Medicine, 2020, 13(2): 123-126 (in Chinese).
[13]	SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 2818-2826.
[14]	李妮妮, 王夏黎, 付阳阳, 等. 一种优化YOLO模型的交通警察目标检测方法[J]. 图学学报, 2022, 43(2): 296-305.
	LI N N, WANG X L, FU Y Y, et al. A traffic police object detection method based on optimized YOLO model[J]. Journal of Graphics, 2022, 43(2): 296-305 (in Chinese). DOI
[15]	LIN C C, ZHAO G S, YANG Z R, et al. CIR-Net: automatic classification of human chromosome based on inception-resnet architecture[J]. IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2022, 19(3): 1285-1293.
[16]	LIU X B, FU L J, LIN J C W, et al. SRAS-Net: low-resolution chromosome image classification based on deep learning[J]. IET Systems Biology, 2022, 16(3/4): 85-97.
[17]	MCGOWAN-JORDAN J, SIMONS A, SCHMID M. ISCN 2016: an international system for human cytogenomic nomenclature[M]. Basel: Karger, 2016: 8-10.
[18]	HERNÁNDEZ-MIER Y, NUÑO-MAGANDA M A, POLANCO-MARTAGÓN S, et al. Machine learning classifiers evaluation for automatic karyogram generation from G-banded metaphase images[J]. Applied Sciences, 2020, 10(8): 2758.
[19]	GU Z W, CHENG J, FU H Z, et al. CE-Net: context encoder network for 2D medical image segmentation[J]. IEEE Transactions on Medical Imaging, 2019, 38(10): 2281-2292. DOI PMID
[20]	WANG G J, LIU H, YI X P, et al. ARMS Net: overlapping chromosome segmentation based on adaptive receptive field multi-scale network[J]. Biomedical Signal Processing and Control, 2021, 68: 102811.
[21]	DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An image is worth 16x16 words: transformers for image recognition at scale[EB/OL]. [2024-03-07]. https://dblp.uni-trier.de/rec/conf/iclr/DosovitskiyB0WZ21.html?view=bibtex.
[22]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]// The 31st International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2017: 6000-6010.
[23]	杨陈成, 董秀成, 侯兵, 等. 基于参考的Transformer纹理迁移深度图像超分辨率重建[J]. 图学学报, 2023, 44(5): 861-867. DOI
	YANG C C, DONG X C, HOU B, et al. Reference based transformer texture migrates depth images super resolution reconstruction[J]. Journal of Graphics, 2023, 44(5): 861-867 (in Chinese).
[24]	陈静, 杨学志, 陈鲸, 等. 采用自注意力抗干扰网络的视频房颤检测[J]. 图学学报, 2023, 44(2): 313-323. DOI
	CHEN J, YANG X Z, CHEN J,, et al. Video atrial fibrillation detection using self-attentional anti-interference network[J]. Journal of Graphics, 2023, 44(2): 313-323 (in Chinese). DOI
[25]	朱光辉, 缪君, 胡宏利, 等. 基于自增强注意力机制的室内单图像分段平面三维重建[J]. 图学学报, 2024, 45(3): 464-471. DOI
	ZHU G H, MIAO J, HU H L, et al. 3D piece-wise planar reconstruction from a single indoor image based on self-augmented -attention mechanism[J]. Journal of Image and Graphics, 2024, 45(3): 464-471 (in Chinese).
[26]	RAHIMZADEH M, PARVIN S, ASKARI A, et al. Wise-SrNet: a novel architecture for enhancing image classification by learning spatial resolution of feature maps[J]. Pattern Analysis and Applications, 2024, 27(2): 30.
[27]	张运波, 易鹏飞, 周东生, 等. 深度可分离卷积和标准卷积相结合的高效行人检测器[J]. 图学学报, 2022, 43(2): 230-238.
	ZHANG Y B, YI P F, ZHOU D S, et al. Efficient pedestrian detector combining depthwise separable convolution and standard convolution[J]. Journal of Graphics, 2022, 43(2): 230-238 (in Chinese).
[28]	刘南杉, 裴云强, 蒋皓, 等. 基于VD-MobileNet网络的WebAR生活垃圾分类信息可视化方法[J]. 图学学报, 2022, 43(4): 667-676.
	LIU N S, PEI Y Q, JIANG H, et al. WebAR garbage classification information visualization method based on VD-MobileNet network[J]. Journal of Graphics, 2022, 43(4): 667-676 (in Chinese).
[29]	毛爱坤, 刘昕明, 陈文壮, 等. 改进YOLOv5算法的变电站仪表目标检测方法[J]. 图学学报, 2023, 44(3): 448-455. DOI
	MAO A K, LIU X M, CHEN W Z, et al. Improved substation instrument target detection method for YOLOv5 algorithm[J]. Journal of Graphics, 2023, 44(3): 448-455 (in Chinese).
[30]	ZHU L Y, JI D Y, ZHU S P, et al. Learning statistical texture for semantic segmentation[C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2021: 12532-12541.
[31]	SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2015: 1-9.
[32]	FAN D P, JI G P, SUN G L, et al. Camouflaged object detection[C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2020: 2774-2784.
[33]	ZHAO H Q, ZHOU W B, CHEN D D, et al. Multi-attentional deepfake detection[C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2021: 2185-2194.

数据增强	训练集	验证集	测试集	总数
增强前	3 220	460	920	4 600
增强后	45 080	460	920	46 460

数据增强	训练集	验证集	测试集	总数
增强前	3 220	460	920	4 600
增强后	45 080	460	920	46 460

图像	Acc	P	R	F1-score	L-F1	S-F1
原始图像	94.67	94.83	94.46	94.59	95.20	91.98
放大图像	98.04	98.08	97.81	97.90	98.33	97.42

图像	Acc	P	R	F1-score	L-F1	S-F1
原始图像	94.67	94.83	94.46	94.59	95.20	91.98
放大图像	98.04	98.08	97.81	97.90	98.33	97.42

纹理增强模块	Acc	P	R	F1-score	L-F1	S-F1
文献[32]	96.88	96.95	97.08	96.94	97.70	95.25
文献[33]	97.15	96.87	96.91	97.03	97.86	95.47
STL	97.32	96.98	97.10	97.11	98.02	95.98