Thio Brandon J, Sinha Nishant, Davis Kathryn A, Sinha Saurabh R, Grill Warren M
Department of Biomedical Engineering, Duke University, Durham, NC 27708, USA.
Department of Bioengineering, University of Pennsylvania, Philadelphia, PA 19104, USA.
Brain. 2025 Mar 6;148(3):764-775. doi: 10.1093/brain/awae297.
Epilepsy surgery can eliminate seizures in patients with drug-resistant focal epilepsy. Surgical intervention requires proper identification of the epileptic network and often involves implanting stereo-EEG electrodes in patients where non-invasive methods are insufficient. However, only ∼60% of patients achieve seizure-freedom following surgery. Quantitative methods have been developed to help improve surgical outcomes. However, previous quantitative methods that localized interictal spike and seizure activity using stereo-EEG recordings did not account for the propagation path encoded by the temporal dynamics of stereo-EEG recordings. Reconstructing the seizure propagation path can aid in determining whether a signal originated from the seizure onset or propagation zone, which directly informs treatment decisions. We developed a novel source reconstruction algorithm, Temporally Dependent Iterative Expansion (TEDIE), that accurately reconstructs propagating and expanding neural sources over time. TEDIE iteratively optimizes the number, location and size of neural sources to minimize the differences between the reconstructed and recorded stereo-EEG signals using temporal information to refine the reconstructions. The TEDIE output comprises a movie of seizure activity projected onto patient-specific brain anatomy. We analysed data from 46 epilepsy patients implanted with stereo-EEG electrodes at Duke Hospital (12 patients) and the Hospital of the University of Pennsylvania (34 patients). We reconstructed seizure recordings and found that TEDIE's seizure onset zone reconstructions were closer to the resected brain region for Engel 1 compared to Engel 2-4 patients, retrospectively validating the clinical utility of TEDIE. We also demonstrated that TEDIE has prospective clinical value, whereby metrics that can be determined presurgically accurately predict whether a patient would achieve seizure-freedom following surgery. Furthermore, we used TEDIE to delineate new potential surgical targets in 12/23 patients who are currently Engel 2-4. We validated TEDIE by accurately reconstructing various dynamic synthetic neural sources with known locations and sizes. TEDIE generated more accurate, focal and interpretable dynamic reconstructions of seizures compared to other algorithms (sLORETA and IRES). Our findings demonstrate that TEDIE is a promising clinical tool that can greatly improve epileptogenic zone localization and epilepsy surgery outcomes.
癫痫手术可以消除药物难治性局灶性癫痫患者的癫痫发作。手术干预需要正确识别癫痫网络，并且在非侵入性方法不足时，通常需要在患者体内植入立体脑电图电极。然而，只有约60%的患者在手术后实现无癫痫发作。已经开发了定量方法来帮助改善手术结果。然而，以前使用立体脑电图记录来定位发作间期棘波和癫痫发作活动的定量方法没有考虑到立体脑电图记录的时间动态所编码的传播路径。重建癫痫发作传播路径有助于确定信号是源自癫痫发作起始区还是传播区，这直接为治疗决策提供依据。我们开发了一种新颖的源重建算法，即时间相关迭代扩展（TEDIE），它可以随着时间准确重建传播和扩展的神经源。TEDIE迭代优化神经源的数量、位置和大小，以利用时间信息细化重建，从而最小化重建的和记录的立体脑电图信号之间的差异。TEDIE的输出包括投影到患者特定脑解剖结构上的癫痫发作活动的动态影像。我们分析了来自杜克医院（12例患者）和宾夕法尼亚大学医院（34例患者）植入立体脑电图电极的46例癫痫患者的数据。我们重建了癫痫发作记录，发现与恩格尔2 - 4级患者相比，TEDIE对恩格尔1级患者的癫痫发作起始区重建更接近切除的脑区，从而回顾性地验证了TEDIE的临床效用。我们还证明了TEDIE具有前瞻性临床价值，即术前可确定的指标能够准确预测患者术后是否能实现无癫痫发作。此外，我们使用TEDIE在目前为恩格尔2 - 4级的23例患者中的12例中划定了新的潜在手术靶点。我们通过准确重建具有已知位置和大小的各种动态合成神经源来验证TEDIE。与其他算法（sLORETA和IRES）相比，TEDIE生成了更准确、更局限且更具可解释性的癫痫动态重建。我们的研究结果表明，TEDIE是一种很有前景的临床工具，它可以极大地改善致痫区定位和癫痫手术结果。
