|
|
论文:2023,Vol:41,Issue(6):1198-1208 |
|
|
引用本文： |
|
|
何富运, 韦燕, 丰芳宇, 钱有为. 基于特征重构自愈残差网络的神经元形态分类[J]. 西北工业大学学报 |
|
|
HE Fuyun, WEI Yan, FENG Fangyu, QIAN Youwei. Classification of neuronal morphology based on feature reconstruction and self-cure residual networks[J]. Journal of Northwestern Polytechnical University |
|
|
|
|
|
|
|
| 基于特征重构自愈残差网络的神经元形态分类 |
|
| 何富运, 韦燕, 丰芳宇, 钱有为 |
|
| 广西师范大学 电子与信息工程学院, 广西 桂林 541004 |
| 摘要: |
| 针对不同类别神经元之间的形态相似度高、类内区别性大,容易导致神经元分类准确率不高的问题,提出了一种基于特征重构自愈残差网络的神经元形态分类方法。针对传统卷积造成边缘像素弱化和填充策略带来新像素侵蚀特征的问题,在基础网络后端构建特征重构模块来保留重要的中心特征并过滤受损的边缘特征。利用自注意力权重模块和排序正则化损失方法增强对神经元形态特征的关注。自注意力权重模块为每个样本重新分配权重,以此捕获样本重要性进行加权损失;排序正则化模块则将这些权重按降序重新排序,分为高低2组权重,同时通过在2组平均权重之间强制执行边距进行正则化处理。所提方法在大鼠神经元形态数据集上进行实验,实现了较为优良的分类效果,在Img_raw、Img_resample和Img_XYalign数据集上进行十二分类的准确率分别达到了96.7%,86.94%,85.84%。与其他分类方法相比,所提方法具有更高的神经元形态分类准确率,相较于基础网络ResNet18,有效地提升了神经元形态分类准确率。 |
| 关键词:
神经元形态分类
特征重构
自愈残差网络
自注意力权重
深度学习
|
|
| Classification of neuronal morphology based on feature reconstruction and self-cure residual networks |
|
| HE Fuyun, WEI Yan, FENG Fangyu, QIAN Youwei |
|
| School of Electronic and Information Engineering, Guangxi Normal University, Guilin 541004, China |
| Abstract: |
| Aiming at the problem of high morphological similarity between the different types of neurons and the large intra-class difference, which is easy to lead to low accuracy of neuron classification, a neural morphology classification method based on feature reconstruction and self-cure residual network is proposed. Firstly, to address the problems of edge pixel weakening and feature erosion by padding strategies that tend to occur during the convolution process of conventional convolution, a feature reconstruction module is constructed at the back end of the backbone network to retain important central features and filter damaged edge features. Then, the attention to neuronal morphological features is enhanced by using a self-attentive weight module and a rank regularization loss method, where the self-attention weight module assigns a weight to each sample to capture the sample importance for weighted loss. In addition, the rank regularization module re-ranked these weights in descending order, dividing them into two groups of high and low weights and regularizing the two groups by enforcing margins between the two average weights. The method achieved superior classification results on the NeuroMorpho-rat dataset, with twelve-way classification accuracies of 96.7%, 86.94% and 85.84% on the Img_raw, Img_resample and Img_XYalign datasets, separately. Comparing with the other methods, the present method has a higher classification accuracy of neurons. Comparing with the original ResNet18 network, it can effectively improve the neuron classification accuracy. |
| Key words:
neuronal morphological classification
self-cure residual networks
feature reconstruction
attention weight
deep learning
|
|
| 收稿日期: 2023-01-11
修回日期:
|
| DOI: 10.1051/jnwpu/20234161198 |
| 基金项目: 国家自然科学基金(62062014)、广西自然科学基金(2018GXNSFAA050024,2021JJA170004)与广西师范大学重点科研项目(2018ZD007)资助 |
| 通讯作者:
Email： |
| 作者简介: 何富运(1982-),广西师范大学副教授,主要从事生物图像信息处理与分析研究。e-mail:he_fuyun@gxnu.edu.cn
|
|
|
|
|
|
|
|
参考文献: |
|
|
[1] 蔺想红, 张玉平, 李志强, 等. 三维神经元几何形态生成算法研究进展[J]. 计算机工程, 2015, 41(2): 161-166 LIN Xianghong, ZHANG Yuping, LI Zhiqiang, et al. Research progress on geometric morphology generation algorithms for 3D neurons[J]. Computer Engineering, 2015, 41(2): 161-166 (in Chinese)
[2] Oshio K, Yamada S, Nakashima M. Neuron classification based on temporal firing patterns by the dynamical analysis with changing time resolution(DCT) method[J]. Biological Cybernetics, 2003, 88(6): 438-449
[3] GOUWENS N W, BERG J, FENG D, et al. Systematic generation of biophysically detailed models for diverse cortical neuron types[J]. Nature Communications, 2018, 9(1): 710
[4] ARENDT D, MUSSER J M, BAKER C V H, et al. The origin and evolution of cell types[J]. Nature Reviews Genetics, 2016, 17(12): 744-757
[5] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90
[6] SHI J, ZHU S, WANG D, et al. ARM: a lightweight module to amend facial expression representation[J]. Signal, Image and Video Processing, 2022, 17(4): 1315-1323
[7] 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, 2015: 770-778
[8] SHI W, CABALLERO J, HUSZÁR F, et al. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network[C]//Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2016: 1874-1883
[9] WANG T, LIAO D. Neuronal morphology classification based on SVM[C]//2011 International Conference on Computer Science and Service System, 2011
[10] ALAVI A, CAVANAGH B, TUXWORTH G, et al. Automated classification of dopaminergic neurons in the rodent brain[C]//2009 International Joint Conference on Neural Networks, 2009: 81-88
[11] VASQUES X, VANEL L, VILLETTE G, et al. Morphological neuron classification using machine learning[J]. Frontiers in Neuroanatomy, 2016, 10(102): 102
[12] FOGO G M, ANZELL A R, MAHERAS K J, et al. Machine learning-based classification of mitochondrial morphology in primary neurons and brain[J]. Scientific Reports, 2021, 11(1): 5133
[13] LIN X, ZHENG J. A Neuronal morphology classification approach based on locally cumulative connected deep neural networks[J]. Applied Sciences, 2019, 9(18): 3876
[14] LIN X, ZHENG J, WANG X, et al. A neuronal morphology classification approach based on deep residual neural networks[J]. Neural Information Processing, 2018, 11304: 336-348
[15] 蔺想红, 郑鉴洋, 王向文, 等. 基于深度学习网络的神经元自适应投影分类方法[J]. 电子学报, 2020, 48(7): 1321-1329 LIN Xianghong, ZHENG Jianyang, WANG Xiangwen, et al. Neuronal adaptive projection classification method based on deep learning network[J]. Acta Electronica Sinica, 2020, 48(7): 1321-1329 (in Chinese)
[16] ZHANG T, ZENG Y, ZHANG Y, et al. Neuron type classification in rat brain based on integrative convolutional and tree-based recurrent neural networks[J]. Scientific Reports, 2021, 11(1): 7291
[17] YAMASHIRO K, LIU J, MATSUMOTO N, et al. Deep learning-based classification of GAD67-positive neurons without the immunosignal[J]. Frontiers in Neuroanatomy, 2021, 15: 643067
[18] SHI Jiawei, ZHU Songhao, LIANG Zhiwei. Amending facial expression representation via de-albino[C]//2022 41st Chinese Control Conference, 2022
[19] HU W, HUANG Y, ZHANG F, et al. Noise-tolerant paradigm for training face recognition CNNs[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 11879-11888
[20] VAN DER MAATEN L, HOMTON G. Visualizing data using t-SNE[J]. Journal of Machine Learning Research, 2008, 9: 2579-2605
[21] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 1409-1556
[22] XIE S, GIRSHICK R, DOLLÁR P, et al. Aggregated residual transformations for deep neural networks[C]//30th IEEE Conference on Computer Vision and Pattern Recognition, 2017: 5987-5995 |
|
|
|
|
|
|
|
