Artificial intelligence for automatic classification of needle EMG signals: A scoping review.

OBJECTIVE: This scoping review provides an overview of artificial intelligence (AI), including machine and deep learning techniques, in the interpretation of clinical needle electromyography (nEMG) signals. METHODS: A comprehensive search of Medline, Embase and Web of Science was conducted to find peer-reviewed journal articles. All papers published after 2010 were included. The methodological quality of the included studies was assessed with CLAIM (checklist for artificial intelligence in medical imaging). RESULTS: 51 studies were identified that fulfilled the inclusion criteria. 61% used open-source EMGlab data set to develop models to classify nEMG signal in healthy, amyotrophic lateral sclerosis (ALS) and myopathy (25 subjects). Only two articles developed models to classify signals recorded at rest. Most articles reported high performance accuracies, but many were subject to bias and overtraining. CONCLUSIONS: Current AI-models of nEMG signals are not sufficient for clinical implementation. Suggestions for future research include emphasizing the need for an optimal training and validation approach using large datasets of clinical nEMG data from a diverse patient population. SIGNIFICANCE: The outcomes of this study and the suggestions made aim to contribute to developing AI-models that can effectively handle signal quality variability and are suitable for daily clinical practice in interpreting nEMG signals.

High performance clean versus artifact dry electrode EEG data classification using Convolutional Neural Network transfer learning.

Objective: Convolutional Neural Networks (CNNs) are promising for artifact detection in electroencephalography (EEG) data, but require large amounts of data. Despite increasing use of dry electrodes for EEG data acquisition, dry electrode EEG datasets are sparse. We aim to develop an algorithm for clean versus artifact dry electrode EEG data classification using transfer learning. Methods: Dry electrode EEG data were acquired in 13 subjects while physiological and technical artifacts were induced. Data were per 2-second segment labeled as clean or artifact and split in an 80% train and 20% test set. With the train set, we fine-tuned a pre-trained CNN for clean versus artifact wet electrode EEG data classification using 3-fold cross validation. The three fine-tuned CNNs were combined in one final clean versus artifact classification algorithm, in which the majority vote was used for classification. We calculated accuracy, F1-score, precision, and recall of the pre-trained CNN and fine-tuned algorithm when applied to unseen test data. Results: The algorithm was trained on 0.40 million and tested on 0.17 million overlapping EEG segments. The pre-trained CNN had a test accuracy of 65.6%. The fine-tuned clean versus artifact classification algorithm had an improved test accuracy of 90.7%, F1-score of 90.2%, precision of 89.1% and recall of 91.2%. Conclusions: Despite a relatively small dry electrode EEG dataset, transfer learning enabled development of a high performing CNN-based algorithm for clean versus artifact classification. Significance: Development of CNNs for classification of dry electrode EEG data is challenging as dry electrode EEG datasets are sparse. Here, we show that transfer learning can be used to overcome this problem.

Correlation between EEG spectral power and cerebral perfusion in patients with acute ischemic stroke.

Dry electrode electroencephalography (EEG) has the potential to diagnose ischemic stroke in the acute phase. In the current study we determined the correlation between EEG spectral power and ischemic stroke size and location as determined by computed tomography perfusion (CTP). Dry electrode EEG recordings were performed in patients with acute ischemic stroke in the emergency room. CTP preceded the EEG recordings as part of standard imaging protocol. Infarct core volume, total hypoperfused volume and local cerebral blood flow (CBF) were estimated with CTP. Additionally, global and local EEG spectral power were determined. We used Spearman's correlation coefficients to evaluate the correlation between variables. We included 27 patients (median age 72 [IQR:69-80] years, 15/27 [56%] men). Median CTP-to-EEG time was 32 (range:8-138) minutes. Hypoperfused volumes were estimated for 12/27 (44%) patients. Infarct core volume correlated best with global delta power (ρ = 0.76, p < 0.01), total hypoperfused volume with global alpha power (ρ = -0.58, p = 0.05), and local CBF with local alpha power (ρ = 0.43, p < 0.01). We conclude that dry electrode EEG signals slow down with increasing hypoperfused volume, which could potentially be used to discriminate between small and large ischemic strokes.

Deriving reference values for nerve conduction studies from existing data using mixture model clustering.

OBJECTIVE: to obtain locally valid reference values (RVs) from existing nerve conduction study (NCS) data. METHODS: we used age, sex, height and limb temperature-based mixture model clustering (MMC) to identify normal and abnormal measurements on NCS data from two university hospitals. We compared MMC-derived RVs to published data; examined the effect of using different variables; validated MMC-derived RVs using independent data from 26 healthy control subjects and investigated their clinical applicability for the diagnosis of polyneuropathy. RESULTS: MMC-derived RVs were similar to published RVs. Clustering can be achieved using only sex and age as variables. MMC is likely to yield reliable results with fewer abnormal than normal measurements and when the total number of measurements is at least 300. Measurements from healthy controls fell within the 95% MMC-derived prediction interval in 97.4% of cases. CONCLUSIONS: MMC can be used to obtain RVs from existing data, providing a locally valid, accurate reflection of the (ab)normality of an NCS result. SIGNIFICANCE: MMC can be used to generate locally valid RVs for any test for which sufficient data are available.1.

Hemodynamic Differences in Intracranial Aneurysms before and after Rupture.

BACKGROUND AND PURPOSE: Rupture risk of intracranial aneurysms may depend on hemodynamic characteristics. This has been assessed by comparing hemodynamic data of ruptured and unruptured aneurysms. However, aneurysm geometry may change before, during, or just after rupture; this difference causes potential changes in hemodynamics. We assessed changes in hemodynamics in a series of intracranial aneurysms, by using 3D imaging before and after rupture. MATERIALS AND METHODS: For 9 aneurysms in 9 patients, we used MRA, CTA, and 3D rotational angiography before and after rupture to generate geometric models of the aneurysm and perianeurysmal vasculature. Intra-aneurysmal hemodynamics were simulated by using computational fluid dynamics. Two neuroradiologists qualitatively assessed flow complexity, flow stability, inflow concentration, and flow impingement in consensus, by using flow-velocity streamlines and wall shear stress distributions. RESULTS: Hemodynamics changed in 6 of the 9 aneurysms. The median time between imaging before and after rupture was 678 days (range, 14-1461 days) in these 6 cases, compared with 151 days (range, 34-183 days) in the 3 cases with unaltered hemodynamics. Changes were observed for flow complexity (n = 3), flow stability (n = 3), inflow concentration (n = 2), and region of flow impingement (n = 3). These changes were in all instances associated with aneurysm displacement due to rupture-related hematomas, growth, or newly formed lobulations. CONCLUSIONS: Hemodynamic characteristics of intracranial aneurysms can be altered by geometric changes before, during, or just after rupture. Associations of hemodynamic characteristics with aneurysm rupture obtained from case-control studies comparing ruptured with unruptured aneurysms should therefore be interpreted with caution.

Generalized versus patient-specific inflow boundary conditions in computational fluid dynamics simulations of cerebral aneurysmal hemodynamics.

BACKGROUND AND PURPOSE: Attempts have been made to associate intracranial aneurysmal hemodynamics with aneurysm growth and rupture status. Hemodynamics in aneurysms is traditionally determined with computational fluid dynamics by using generalized inflow boundary conditions in a parent artery. Recently, patient-specific inflow boundary conditions are being implemented more frequently. Our purpose was to compare intracranial aneurysm hemodynamics based on generalized versus patient-specific inflow boundary conditions. MATERIALS AND METHODS: For 36 patients, geometric models of aneurysms were determined by using 3D rotational angiography. 2D phase-contrast MR imaging velocity measurements of the parent artery were performed. Computational fluid dynamics simulations were performed twice: once by using patient-specific phase-contrast MR imaging velocity profiles and once by using generalized Womersley profiles as inflow boundary conditions. Resulting mean and maximum wall shear stress and oscillatory shear index values were analyzed, and hemodynamic characteristics were qualitatively compared. RESULTS: Quantitative analysis showed statistically significant differences for mean and maximum wall shear stress values between both inflow boundary conditions (P < .001). Qualitative assessment of hemodynamic characteristics showed differences in 21 cases: high wall shear stress location (n = 8), deflection location (n = 3), lobulation wall shear stress (n = 12), and/or vortex and inflow jet stability (n = 9). The latter showed more instability for the generalized inflow boundary conditions in 7 of 9 patients. CONCLUSIONS: Using generalized and patient-specific inflow boundary conditions for computational fluid dynamics results in different wall shear stress magnitudes and hemodynamic characteristics. Generalized inflow boundary conditions result in more vortices and inflow jet instabilities. This study emphasizes the necessity of patient-specific inflow boundary conditions for calculation of hemodynamics in cerebral aneurysms by using computational fluid dynamics techniques.