癫痫是一种由于大脑神经元突发性异常放电造成大脑暂时性功能障碍的慢性脑部疾病。其发作期具体表现为尖叫、全身抽搐、意识丧失和口吐白沫等，发作间期癫痫样放电（Interictal epileptiform discharges，IEDs），表现为棘波、尖波、棘慢波、尖慢波和以上的混杂波，是在癫痫患者发作之间的某些时段观察到的间歇性电生理事件。癫痫发作间期癫痫样放电可以用于定位癫痫灶，对癫痫手术的术前评估至关重要。

在脑磁图的癫痫定位研究中，其利用癫痫样放电超同步特性，使用了张量分解的无监督方法对癫痫患者的脑网络进行分解从而实现定位；使用了86个癫痫患者癫痫发作间期的静息态脑磁图数据，其中每个病人的数据都包含癫痫样放电；对其进行预处理至源重建，再使用AAL脑解剖图谱划分脑区得到每个脑区的时间序列，通过滑动窗口的方法提取信号样本，对每个信号样本作分析频带内的小波变换再计算正交化希尔伯特包络为动态功能连接来构建动态脑网络，再利用张量分解的方法分解脑网络以得到脑网络模式，将所得的脑网络模式可视化与偶极子对比以实现对病人癫痫定位。实验结果得出在脑叶级别脑网络与偶极子有匹配的患者数比总患者数为56/86，且在更精细的AAL级别区域脑网络与偶极子有匹配的患者数比总患者数为42/86。实验结果表明此方法的有效性。

对基于传统手工提取特征的机器学习癫痫样放电检测方法进行了对比。在脑磁图的癫痫样放电检测研究中，根据医师在脑磁图上的癫痫样放电标注，从中提取正负样本得到检测的数据集。对基于传统手工提取特征的机器学习方法进行了实验，其特征种类分为了时域统计特征和非线性特征，将这些特征分别输入SVM进行训练，实验结果得出模糊熵特征的方法是这些机器学习手工提取特征的方法中各项评价指标最高的方法。

本研究提出了一种时域、功能连接、频域融合的多尺度卷积空间注意力网络用于检测癫痫样放电。首先实验了深度学习中常用的卷积神经网络在时域上和时频域上进行癫痫样放电检测，基于时域信号进行癫痫样放电检测的网络实验了CNN、ShallowConvNet、DeepConvNet和EEGNet，基于时频域信号设计了一个tfCNN用于时频域信号的癫痫样放电检测；本研究提出的网络融合不同的特征信息，采用多个卷积核进行多路卷积以能够找出最适宜的卷积核，引入了空间注意力机制对特征图位置权重进行学习使得能够获得空间特征，在与所有对比方法比较中实现了92.47%的最高准确率以及90.03%的最高F1-score，以及在PR曲线和ROC曲线中都展现出了最好的效果，最后对提出的网络进行消融实验，验证了网络各个模块的有效性。

总体而言，本研究探究了利用了张量分解脑网络的无监督方法进行癫痫定位，实验结果表明了该方法的有效性，且脑网络与癫痫活动紧密相关，癫痫样放电脑网络特征一直是癫痫领域的重要研究内容，通过对该发作间期癫痫样放电脑网络的研究，深化了我们对癫痫发作间期脑网络的认识，为基于脑网络的无监督定位方法提供了见解，为癫痫手术术前评估起到了较好的辅助作用。在癫痫样放电检测中，本研究对比了基于机器学习方法的不同特征的优劣，实验了常用的卷积神经网络检测癫痫样放电，提出了一种时域、功能连接、频域融合的多尺度卷积空间注意力网络，在所有对比方法中获得了最高的准确率与F1-score。该网络的提出为后续癫痫样放电的检测系统提供了良好的基础，能很大地减轻医师标注负担。

Epilepsy is a chronic brain disorder characterized by temporary functional impairment of the brain due to sudden abnormal discharges of brain neurons. During seizures, specific symptoms such as screaming, convulsions, loss of consciousness, and foaming at the mouth may occur. Interictal epileptiform discharges (IEDs) are intermittent electrophysiological events observed in some periods between seizures in epilepsy patients. They are characterized by spike waves, sharp waves, slow waves, and mixed waves. IEDs can be used to locate the epileptic focus, which is crucial for preoperative evaluation of epilepsy surgery.

 

In epilepsy localization studies using magnetoencephalography (MEG), an unsupervised method of tensor decomposition was used to decompose the brain networks of epilepsy patients by utilizing the super-synchronous characteristics of IEDs. A total of 86 resting-state MEG data sets from epilepsy patients with IEDs were used, and the preprocessed data were reconstructed to the source level. The time series of each brain area was obtained by dividing the brain into regions using the automated anatomical labeling (AAL) atlas. Signal samples were extracted using a sliding window method, and for each sample, the wavelet transform was used to analyze the frequency band and calculate the orthogonalized Hilbert envelope as the dynamic functional connection to construct the dynamic brain network. The brain network was then decomposed using tensor decomposition to obtain the brain network patterns, and the obtained brain network patterns were visualized and compared with dipoles to achieve localization. The experimental results showed that 56 out of 86 patients had a match between the brain network and the dipole at the lobe level, and 42 out of 86 patients had a match at the more refined AAL region level, indicating the effectiveness of the method.

 

A comparison was made between the machine learning method based on traditional handcrafted feature extraction and the proposed method for detecting IEDs in MEG data. The dataset used for machine learning was obtained by extracting positive and negative samples based on physician-labeled IEDs in MEG data. Traditional handcrafted feature extraction methods were divided into time-domain statistical features and nonlinear features, and these features were input into the support vector machine (SVM) for training. The experimental results showed that fuzzy entropy feature extraction method has the highest evaluation indicators among all the machine learning handcrafted feature extraction methods.

This study proposes a multi-scale convolutional spatial attention network that integrates time-domain, functional connectivity, and frequency-domain fusion for detecting epileptic discharges. First, used common convolutional neural networks in deep learning were experimented for detecting epileptic discharges in the time-domain and time-frequency domain. CNN, ShallowConvNet, DeepConvNet, EEGNet were experimented for detecting epileptic discharges based on time-domain signals, while tfCNN was designed for detecting epileptic discharges based on time-frequency domain signals. The proposed network fuses different feature information and uses multiple convolution kernels for multiple convolution pathways to find the most suitable kernel. The spatial attention mechanism is introduced to learn the weight of feature map positions, which enables the acquisition of spatial features. The proposed network achieved the highest accuracy of 92.47% and the highest F1-score of 90.03% compared with all the comparison methods. The network also demonstrated the best performance in both PR curves and ROC curves. Finally, ablation experiments were conducted to verify the effectiveness of each module of the proposed network.

 

Overall, this study investigated the unsupervised method of using tensor decomposition brain networks for epileptic localization. The experimental results demonstrated the effectiveness of this method, and the brain network was closely related to epileptic activity. The brain network feature of epileptic discharges has always been an important research topic in the field of epilepsy. By studying the brain network of epileptic discharges during the interictal period, this study deepened our understanding of the brain network during the interictal period of epilepsy and provided insights for unsupervised localization methods based on brain networks. It also played a good auxiliary role in preoperative evaluation for epilepsy surgery. In the detection of epileptic discharges, this study compared the advantages and disadvantages of different features based on machine learning methods and experimented with using common convolutional neural networks for detecting epileptic discharges. A multi-scale convolutional spatial attention network that integrates time-domain, functional connectivity, and frequency-domain fusion was proposed, which achieved the highest accuracy and F1-score among all the comparison methods. The proposal of this network provides a good foundation for subsequent epileptic discharge detection systems and greatly reduces the burden of physician annotation.

[1] Ramzan Q, Shidlovskiy S. Evolution of the Brain Computing Interface (BCI) and Proposed Electroencephalography (EEG) Signals Based Authentication Model[C]//MATEC Web of Conferences. EDP Sciences, 2018, 155: 01006.
[2] Acharya U R, Hagiwara Y, Deshpande S N, et al. Characterization of focal EEG signals: a review[J]. Future Generation Computer Systems, 2019, 91: 290-299.
[3] Saxena S, Li S. Defeating epilepsy: a global public health commitment[J]. Epilepsia Open, 2017, 2(2): 153-155.
[4] Van Mierlo P, Vorderwülbecke B J, Staljanssens W, et al. Ictal EEG source localization in focal epilepsy: Review and future perspectives[J]. Clinical Neurophysiology, 2020, 131(11): 2600-2616.
[5] Duez L, Tankisi H, Hansen P O, et al. Electromagnetic source imaging in presurgical workup of patients with epilepsy: a prospective study[J]. Neurology, 2019, 92(6): e576-e586.
[6] Blanco J A, Stead M, Krieger A, et al. Unsupervised classification of high-frequency oscillations in human neocortical epilepsy and control patients[J]. Journal of Neurophysiology, 2010, 104(5): 2900-2912.
[7] 张青, 丁成赟, 王潇慧, 等. 癫痫耐药机制的研究进展[J]. 中华临床医师杂志: 电子版, 2015, 9(4): 115-119.
[8] Jacobs J, Vogt C, LeVan P, et al. The identification of distinct high-frequency oscillations during spikes delineates the seizure onset zone better than high-frequency spectral power changes[J]. Clinical Neurophysiology, 2016, 127(1): 129-142.
[9] 万婷. 实现癫痫始发区精确定位的高频振荡节律自动检测算法[D]. 中国地质大学, 2018.
[10] 张馨月. 高频振荡信号自动检测算法及其在定位致痫灶中的应用[D]. 电子科技大学, 2020.
[11] Burnos S, Hilfiker P, Sürücü O, et al. Human intracranial high frequency oscillations (HFOs) detected by automatic time-frequency analysis[J]. PloS One, 2014, 9(4): e94381.
[12] Cho J R, Koo D L, Joo E Y, et al. Resection of individually identified high‐rate high‐frequency oscillations region is associated with favorable outcome in neocortical epilepsy[J]. Epilepsia, 2014, 55(11): 1872-1883.
[13] Gliske S V, Irwin Z T, Davis K A, et al. Universal automated high frequency oscillation detector for real-time, long term EEG[J]. Clinical Neurophysiology, 2016, 127(2): 1057-1066.
[14] Cotic M, Chinvarun Y, del Campo M, et al. Spatial coherence profiles of ictal high-frequency oscillations correspond to those of interictal low-frequency oscillations in the ECoG of epileptic patients[J]. IEEE Transactions on Biomedical Engineering, 2014, 63(1): 76-85.
[15] Cotic M, Zalay O C, Chinvarun Y, et al. Mapping the coherence of ictal high frequency oscillations in human extratemporal lobe epilepsy[J]. Epilepsia, 2015, 56(3): 393-402.
[16] 杜玉晓, 陈崇毅. 基于脑电高频振荡节律的癫痫始发区快速定位算法研究[J]. 广东工业大学学报, 2015, 32(4): 60-66.
[17] 郑霄, 张丹, 石岩芳, 等. 基于发作间期颅内脑电高频振荡的癫痫病灶定位[J]. 北京生物医学工程, 2014 (3): 253-257.
[18] Bartolomei F, Chauvel P, Wendling F. Epileptogenicity of brain structures in human temporal lobe epilepsy: a quantified study from intracerebral EEG[J]. Brain, 2008, 131(7): 1818-1830.
[19] Sharma R, Pachori R B, Acharya U R. An integrated index for the identification of focal electroencephalogram signals using discrete wavelet transform and entropy measures[J]. Entropy, 2015, 17(8): 5218-5240.
[20] Shiraishi H, Watanabe Y, Watanabe M, et al. Interictal and ictal magnetoencephalographic study in patients with medial frontal lobe epilepsy[J]. Epilepsia, 2001, 42(7): 875-882.
[21] Hämäläinen M S, Ilmoniemi R J. Interpreting magnetic fields of the brain: minimum norm estimates[J]. Medical & Biological Engineering & Computing, 1994, 32: 35-42.
[22] Pascual-Marqui R D. Standardized low-resolution brain electromagnetic tomography (sLORETA): technical details[J]. Methods Find Exp Clin Pharmacol, 2002, 24(Suppl D): 5-12.
[23] Dale A M, Liu A K, Fischl B R, et al. Dynamic statistical parametric mapping: combining fMRI and MEG for high-resolution imaging of cortical activity[J]. Neuron, 2000, 26(1): 55-67.
[24] Van Veen B D, Van Drongelen W, Yuchtman M, et al. Localization of brain electrical activity via linearly constrained minimum variance spatial filtering[J]. IEEE Transactions on Biomedical Engineering, 1997, 44(9): 867-880.
[25] Mosher J C, Lewis P S, Leahy R M. Multiple dipole modeling and localization from spatio-temporal MEG data[J]. IEEE Transactions on Biomedical Engineering, 1992, 39(6): 541-557.
[26] Cornwell B R, Carver F W, Coppola R, et al. Evoked amygdala responses to negative faces revealed by adaptive MEG beamformers[J]. Brain Research, 2008, 1244: 103-112.
[27] Xiang J, Wang Y, Chen Y, et al. Noninvasive localization of epileptogenic zones with ictal high-frequency neuromagnetic signals: Case report[J]. Journal of Neurosurgery: Pediatrics, 2010, 5(1): 113-122.
[28] Papadelis C, Tamilia E, Stufflebeam S, et al. Interictal high frequency oscillations detected with simultaneous magnetoencephalography and electroencephalography as biomarker of pediatric epilepsy[J]. JoVE (Journal of Visualized Experiments), 2016 (118): e54883.
[29] 张欢欢. 基于皮层脑电的癫痫脑网络研究及癫痫预测探索[D]. 电子科技大学, 2017.
[30] Li Y H, Ye X L, Liu Q Q, et al. Localization of epileptogenic zone based on graph analysis of stereo-EEG[J]. Epilepsy Research, 2016, 128: 149-157.
[31] Wu J, Zhou T, Li T. Detecting epileptic seizures in EEG signals with complementary ensemble empirical mode decomposition and extreme gradient boosting[J]. Entropy, 2020, 22(2): 140.
[32] Diykh M, Li Y, Wen P. EEG sleep stages classification based on time domain features and structural graph similarity[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2016, 24(11): 1159-1168.
[33] Khorshidtalab A, Salami M J E, Hamedi M. Robust classification of motor imagery EEG signals using statistical time–domain features[J]. Physiological Measurement, 2013, 34(11): 1563.
[34] Vidaurre C, Krämer N, Blankertz B, et al. Time domain parameters as a feature for EEG-based brain–computer interfaces[J]. Neural Networks, 2009, 22(9): 1313-1319.
[35] Xie S, Lawniczak A T, Song Y, et al. Feature extraction via dynamic PCA for epilepsy diagnosis and epileptic seizure detection[C]//2010 IEEE International Workshop on Machine Learning for Signal Processing. IEEE, 2010: 337-342.
[36] Sankar R, Natour J. Automatic computer analysis of transients in EEG[J]. Computers in Biology and Medicine, 1992, 22(6): 407-422.
[37] Gotman J. Automatic recognition of epileptic seizures in the EEG[J]. Electroencephalography and Clinical Neurophysiology, 1982, 54(5): 530-540.
[38] Gotman J. Automatic seizure detection: improvements and evaluation[J]. Electroencephalography and Clinical Neurophysiology, 1990, 76(4): 317-324.
[39] Qu H, Gotman J. Improvement in seizure detection performance by automatic adaptation to the EEG of each patient[J]. Electroencephalography and Clinical Neurophysiology, 1993, 86(2): 79-87.
[40] Jerger K K, Netoff T I, Francis J T, et al. Early seizure detection[J]. Journal of Clinical Neurophysiology, 2001, 18(3): 259-268.
[41] Hopfengärtner R, Kerling F, Bauer V, et al. An efficient, robust and fast method for the offline detection of epileptic seizures in long-term scalp EEG recordings[J]. Clinical Neurophysiology, 2007, 118(11): 2332-2343.
[42] Liang S F, Wang H C, Chang W L. Combination of EEG complexity and spectral analysis for epilepsy diagnosis and seizure detection[J]. EURASIP Journal on Advances in Signal Processing, 2010, 2010: 1-15.
[43] Alkan A, Kiymik M K. Comparison of AR and Welch methods in epileptic seizure detection[J]. Journal of Medical Systems, 2006, 30: 413-419.
[44] Mousavi S R, Niknazar M, Vahdat B V. Epileptic seizure detection using AR model on EEG signals[C]//2008 Cairo International Biomedical Engineering Conference. IEEE, 2008: 1-4.
[45] Chisci L, Mavino A, Perferi G, et al. Real-time epileptic seizure prediction using AR models and support vector machines[J]. IEEE Transactions on Biomedical Engineering, 2010, 57(5): 1124-1132.
[46] 艾玲梅, 黄力宇, 黄远桂, 等. 利用双谱分析的癫痫脑电特征研究[J]. 西安交通大学学报, 2004, 38(10): 1097-1100.
[47] Wang H, Shi W, Choy C S. Hardware design of real time epileptic seizure detection based on STFT and SVM[J]. IEEE Access, 2018, 6: 67277-67290.
[48] Sharan R V, Berkovsky S. Epileptic seizure detection using multi-channel EEG wavelet power spectra and 1-D convolutional neural networks[C]//2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). IEEE, 2020: 545-548.
[49] Geng M, Zhou W, Liu G, et al. Epileptic seizure detection based on stockwell transform and bidirectional long short-term memory[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2020, 28(3): 573-580.
[50] Liu G, Zhou W, Geng M. Automatic seizure detection based on S-transform and deep convolutional neural network[J]. International Journal of Neural Systems, 2020, 30(04): 1950024.
[51] Chakraborty M, Mitra D. Epilepsy seizure detection using kurtosis based VMD’s parameters selection and bandwidth features[J]. Biomedical Signal Processing and Control, 2021, 64: 102255.
[52] Tzallas A T, Tsipouras M G, Fotiadis D I. Epileptic seizure detection in EEGs using time–frequency analysis[J]. IEEE Transactions on Information Technology in Biomedicine, 2009, 13(5): 703-710.
[53] Kıymık M K, Güler İ, Dizibüyük A, et al. Comparison of STFT and wavelet transform methods in determining epileptic seizure activity in EEG signals for real-time application[J]. Computers in Biology and Medicine, 2005, 35(7): 603-616.
[54] Daubechies I. The wavelet transform, time-frequency localization and signal analysis[J]. IEEE Transactions on Information Theory, 1990, 36(5): 961-1005.
[55] Khan Y U, Gotman J. Wavelet based automatic seizure detection in intracerebral electroencephalogram[J]. Clinical Neurophysiology, 2003, 114(5): 898-908.
[56] Subasi A. Epileptic seizure detection using dynamic wavelet network[J]. Expert Systems with Applications, 2005, 29(2): 343-355.
[57] Kannathal N, Choo M L, Acharya U R, et al. Entropies for detection of epilepsy in EEG[J]. Computer Methods and Programs in Biomedicine, 2005, 80(3): 187-194.
[58] Übeyli E D. Lyapunov exponents/probabilistic neural networks for analysis of EEG signals[J]. Expert Systems with Applications, 2010, 37(2): 985-992.
[59] Acharya U R, Oh S L, Hagiwara Y, et al. Deep convolutional neural network for the automated detection and diagnosis of seizure using EEG signals[J]. Computers in Biology and Medicine, 2018, 100: 270-278.
[60] O’Shea A, Lightbody G, Boylan G, et al. Neonatal seizure detection from raw multi-channel EEG using a fully convolutional architecture[J]. Neural Networks, 2020, 123: 12-25.
[61] Ullah I, Hussain M, Aboalsamh H. An automated system for epilepsy detection using EEG brain signals based on deep learning approach[J]. Expert Systems with Applications, 2018, 107: 61-71.
[62] Wei X, Zhou L, Chen Z, et al. Automatic seizure detection using three-dimensional CNN based on multi-channel EEG[J]. BMC Medical Informatics and Decision Making, 2018, 18(5): 71-80.
[63] Nogay H S, Adeli H. Detection of epileptic seizure using pretrained deep convolutional neural network and transfer learning[J]. European Neurology, 2020, 83(6): 602-614.
[64] Türk Ö, Özerdem M S. Epilepsy detection by using scalogram based convolutional neural network from EEG signals[J]. Brain Sciences, 2019, 9(5): 115.
[65] 韦晓燕, 周霖, 陈秋源, 等. 基于深度卷积神经网络的癫痫脑电自动检测[J]. 中国数字医学, 2019, 14(1): 9-13.
[66] Pantazis D, Adler A. MEG source localization via Deep Learning[J]. Sensors, 2021, 21(13): 4278.
[67] Pincus S M. Approximate entropy as a measure of system complexity[J]. Proceedings of the National Academy of Sciences, 1991, 88(6): 2297-2301.
[68] Richman J S, Moorman J R. Physiological time-series analysis using approximate entropy and sample entropy[J]. American Journal of Physiology-heart and Circulatory Physiology, 2000.
[69] Chen W, Wang Z, Xie H, et al. Characterization of surface EMG signal based on fuzzy entropy[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2007, 15(2): 266-272.
[70] Porta A, Baselli G, Lombardi F, et al. Conditional entropy approach for the evaluation of the coupling strength[J]. Biological Cybernetics, 1999, 81: 119-129.
[71] Grassberger P, Procaccia I. Estimation of the Kolmogorov entropy from a chaotic signal[J]. Physical Review A, 1983, 28(4): 2591.
[72] Hurst H E. Long-term storage capacity of reservoirs[J]. Transactions of the American Society of Civil Engineers, 1951, 116(1): 770-799.
[73] Wolf A, Swift J B, Swinney H L, et al. Determining Lyapunov exponents from a time series[J]. Physica D: nonlinear phenomena, 1985, 16(3): 285-317.
[74] Schirrmeister R T, Springenberg J T, Fiederer L D J, et al. Deep learning with convolutional neural networks for EEG decoding and visualization[J]. Human Brain Mapping, 2017, 38(11): 5391-5420.
[75] Lawhern V J, Solon A J, Waytowich N R, et al. EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces[J]. Journal of Neural Engineering, 2018, 15(5): 056013.
[76] Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain[J]. Neuroimage, 2002, 15(1): 273-289.
[77] Cohen I, Huang Y, Chen J, et al. Pearson correlation coefficient[J]. Noise Reduction in Speech Processing, 2009: 1-4.
[78] Nolte G, Bai O, Wheaton L, et al. Identifying true brain interaction from EEG data using the imaginary part of coherency[J]. Clinical Neurophysiology, 2004, 115(10): 2292-2307.
[79] Pascual-Marqui R D. Coherence and phase synchronization: generalization to pairs of multivariate time series, and removal of zero-lag contributions[J]. arXiv preprint arXiv:0706.1776, 2007.
[80] Hipp J F, Hawellek D J, Corbetta M, et al. Large-scale cortical correlation structure of spontaneous oscillatory activity[J]. Nature Neuroscience, 2012, 15(6): 884-890.
[81] Tass P, Rosenblum M G, Weule J, et al. Detection of n: m phase locking from noisy data: application to magnetoencephalography[J]. Physical Review Letters, 1998, 81(15): 3291.
[82] Stam C J, Nolte G, Daffertshofer A. Phase lag index: assessment of functional connectivity from multi channel EEG and MEG with diminished bias from common sources[J]. Human Brain Mapping, 2007, 28(11): 1178-1193.
[83] Colclough G L, Woolrich M W, Tewarie P K, et al. How reliable are MEG resting-state connectivity metrics?[J]. Neuroimage, 2016, 138: 284-293.
[84] Kolda T G, Bader B W. Tensor decompositions and applications[J]. SIAM Review, 2009, 51(3): 455-500.
[85] Woo S, Park J, Lee J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision (ECCV). 2018: 3-19.
[86] Van der Maaten L, Hinton G. Visualizing data using t-SNE[J]. Journal of Machine Learning Research, 2008, 9(11).
[87] Shriram R, Sundhararajan M, Daimiwal N. EEG based cognitive workload assessment for maximum efficiency[J]. Int. Organ. Sci. Res. IOSR, 2013, 7: 34-38.
[88] 周昌贵. 临床脑电图手册[J]. 成都: 四川辞书出版社, 1990: 86-89.
[89] Pantazis D, Adler A. MEG source localization via Deep Learning[J]. Sensors, 2021, 21(13): 4278.