More Like this

1. A Deep Learning-Based Real-time Seizure Detection System

   https://doi.org/10.1109/SPMB50085.2020.9353623  

   Shawki, N. ; Elseify, T. ; Cap, T. ; Shah, V. ; Obeid, I. ; Picone, J. ( December 2020 , Proceedings of the IEEE Signal Processing in Medicine and Biology Symposium (SPMB))

   Obeid, Iyad Selesnick (Ed.)

   Electroencephalography (EEG) is a popular clinical monitoring tool used for diagnosing brain-related disorders such as epilepsy [1]. As monitoring EEGs in a critical-care setting is an expensive and tedious task, there is a great interest in developing real-time EEG monitoring tools to improve patient care quality and efficiency [2]. However, clinicians require automatic seizure detection tools that provide decisions with at least 75% sensitivity and less than 1 false alarm (FA) per 24 hours [3]. Some commercial tools recently claim to reach such performance levels, including the Olympic Brainz Monitor [4] and Persyst 14 [5]. In this abstract, we describe our efforts to transform a high-performance offline seizure detection system [3] into a low latency real-time or online seizure detection system. An overview of the system is shown in Figure 1. The main difference between an online versus offline system is that an online system should always be causal and has minimum latency which is often defined by domain experts. The offline system, shown in Figure 2, uses two phases of deep learning models with postprocessing [3]. The channel-based long short term memory (LSTM) model (Phase 1 or P1) processes linear frequency cepstral coefficients (LFCC) [6] features from each EEG channel separately. We use the hypotheses generated by the P1 model and create additional features that carry information about the detected events and their confidence. The P2 model uses these additional features and the LFCC features to learn the temporal and spatial aspects of the EEG signals using a hybrid convolutional neural network (CNN) and LSTM model. Finally, Phase 3 aggregates the results from both P1 and P2 before applying a final postprocessing step. The online system implements Phase 1 by taking advantage of the Linux piping mechanism, multithreading techniques, and multi-core processors. To convert Phase 1 into an online system, we divide the system into five major modules: signal preprocessor, feature extractor, event decoder, postprocessor, and visualizer. The system reads 0.1-second frames from each EEG channel and sends them to the feature extractor and the visualizer. The feature extractor generates LFCC features in real time from the streaming EEG signal. Next, the system computes seizure and background probabilities using a channel-based LSTM model and applies a postprocessor to aggregate the detected events across channels. The system then displays the EEG signal and the decisions simultaneously using a visualization module. The online system uses C++, Python, TensorFlow, and PyQtGraph in its implementation. The online system accepts streamed EEG data sampled at 250 Hz as input. The system begins processing the EEG signal by applying a TCP montage [8]. Depending on the type of the montage, the EEG signal can have either 22 or 20 channels. To enable the online operation, we send 0.1-second (25 samples) length frames from each channel of the streamed EEG signal to the feature extractor and the visualizer. Feature extraction is performed sequentially on each channel. The signal preprocessor writes the sample frames into two streams to facilitate these modules. In the first stream, the feature extractor receives the signals using stdin. In parallel, as a second stream, the visualizer shares a user-defined file with the signal preprocessor. This user-defined file holds raw signal information as a buffer for the visualizer. The signal preprocessor writes into the file while the visualizer reads from it. Reading and writing into the same file poses a challenge. The visualizer can start reading while the signal preprocessor is writing into it. To resolve this issue, we utilize a file locking mechanism in the signal preprocessor and visualizer. Each of the processes temporarily locks the file, performs its operation, releases the lock, and tries to obtain the lock after a waiting period. The file locking mechanism ensures that only one process can access the file by prohibiting other processes from reading or writing while one process is modifying the file [9]. The feature extractor uses circular buffers to save 0.3 seconds or 75 samples from each channel for extracting 0.2-second or 50-sample long center-aligned windows. The module generates 8 absolute LFCC features where the zeroth cepstral coefficient is replaced by a temporal domain energy term. For extracting the rest of the features, three pipelines are used. The differential energy feature is calculated in a 0.9-second absolute feature window with a frame size of 0.1 seconds. The difference between the maximum and minimum temporal energy terms is calculated in this range. Then, the first derivative or the delta features are calculated using another 0.9-second window. Finally, the second derivative or delta-delta features are calculated using a 0.3-second window [6]. The differential energy for the delta-delta features is not included. In total, we extract 26 features from the raw sample windows which add 1.1 seconds of delay to the system. We used the Temple University Hospital Seizure Database (TUSZ) v1.2.1 for developing the online system [10]. The statistics for this dataset are shown in Table 1. A channel-based LSTM model was trained using the features derived from the train set using the online feature extractor module. A window-based normalization technique was applied to those features. In the offline model, we scale features by normalizing using the maximum absolute value of a channel [11] before applying a sliding window approach. Since the online system has access to a limited amount of data, we normalize based on the observed window. The model uses the feature vectors with a frame size of 1 second and a window size of 7 seconds. We evaluated the model using the offline P1 postprocessor to determine the efficacy of the delayed features and the window-based normalization technique. As shown by the results of experiments 1 and 4 in Table 2, these changes give us a comparable performance to the offline model. The online event decoder module utilizes this trained model for computing probabilities for the seizure and background classes. These posteriors are then postprocessed to remove spurious detections. The online postprocessor receives and saves 8 seconds of class posteriors in a buffer for further processing. It applies multiple heuristic filters (e.g., probability threshold) to make an overall decision by combining events across the channels. These filters evaluate the average confidence, the duration of a seizure, and the channels where the seizures were observed. The postprocessor delivers the label and confidence to the visualizer. The visualizer starts to display the signal as soon as it gets access to the signal file, as shown in Figure 1 using the “Signal File” and “Visualizer” blocks. Once the visualizer receives the label and confidence for the latest epoch from the postprocessor, it overlays the decision and color codes that epoch. The visualizer uses red for seizure with the label SEIZ and green for the background class with the label BCKG. Once the streaming finishes, the system saves three files: a signal file in which the sample frames are saved in the order they were streamed, a time segmented event (TSE) file with the overall decisions and confidences, and a hypotheses (HYP) file that saves the label and confidence for each epoch. The user can plot the signal and decisions using the signal and HYP files with only the visualizer by enabling appropriate options. For comparing the performance of different stages of development, we used the test set of TUSZ v1.2.1 database. It contains 1015 EEG records of varying duration. The any-overlap performance [12] of the overall system shown in Figure 2 is 40.29% sensitivity with 5.77 FAs per 24 hours. For comparison, the previous state-of-the-art model developed on this database performed at 30.71% sensitivity with 6.77 FAs per 24 hours [3]. The individual performances of the deep learning phases are as follows: Phase 1’s (P1) performance is 39.46% sensitivity and 11.62 FAs per 24 hours, and Phase 2 detects seizures with 41.16% sensitivity and 11.69 FAs per 24 hours. We trained an LSTM model with the delayed features and the window-based normalization technique for developing the online system. Using the offline decoder and postprocessor, the model performed at 36.23% sensitivity with 9.52 FAs per 24 hours. The trained model was then evaluated with the online modules. The current performance of the overall online system is 45.80% sensitivity with 28.14 FAs per 24 hours. Table 2 summarizes the performances of these systems. The performance of the online system deviates from the offline P1 model because the online postprocessor fails to combine the events as the seizure probability fluctuates during an event. The modules in the online system add a total of 11.1 seconds of delay for processing each second of the data, as shown in Figure 3. In practice, we also count the time for loading the model and starting the visualizer block. When we consider these facts, the system consumes 15 seconds to display the first hypothesis. The system detects seizure onsets with an average latency of 15 seconds. Implementing an automatic seizure detection model in real time is not trivial. We used a variety of techniques such as the file locking mechanism, multithreading, circular buffers, real-time event decoding, and signal-decision plotting to realize the system. A video demonstrating the system is available at: https://www.isip.piconepress.com/projects/nsf_pfi_tt/resources/videos/realtime_eeg_analysis/v2.5.1/video_2.5.1.mp4. The final conference submission will include a more detailed analysis of the online performance of each module. ACKNOWLEDGMENTS Research reported in this publication was most recently supported by the National Science Foundation Partnership for Innovation award number IIP-1827565 and the Pennsylvania Commonwealth Universal Research Enhancement Program (PA CURE). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the official views of any of these organizations. REFERENCES [1] A. Craik, Y. He, and J. L. Contreras-Vidal, “Deep learning for electroencephalogram (EEG) classification tasks: a review,” J. Neural Eng., vol. 16, no. 3, p. 031001, 2019. https://doi.org/10.1088/1741-2552/ab0ab5. [2] A. C. Bridi, T. Q. Louro, and R. C. L. Da Silva, “Clinical Alarms in intensive care: implications of alarm fatigue for the safety of patients,” Rev. Lat. Am. Enfermagem, vol. 22, no. 6, p. 1034, 2014. https://doi.org/10.1590/0104-1169.3488.2513. [3] M. Golmohammadi, V. Shah, I. Obeid, and J. Picone, “Deep Learning Approaches for Automatic Seizure Detection from Scalp Electroencephalograms,” in Signal Processing in Medicine and Biology: Emerging Trends in Research and Applications, 1st ed., I. Obeid, I. Selesnick, and J. Picone, Eds. New York, New York, USA: Springer, 2020, pp. 233–274. https://doi.org/10.1007/978-3-030-36844-9_8. [4] “CFM Olympic Brainz Monitor.” [Online]. Available: https://newborncare.natus.com/products-services/newborn-care-products/newborn-brain-injury/cfm-olympic-brainz-monitor. [Accessed: 17-Jul-2020]. [5] M. L. Scheuer, S. B. Wilson, A. Antony, G. Ghearing, A. Urban, and A. I. Bagic, “Seizure Detection: Interreader Agreement and Detection Algorithm Assessments Using a Large Dataset,” J. Clin. Neurophysiol., 2020. https://doi.org/10.1097/WNP.0000000000000709. [6] A. Harati, M. Golmohammadi, S. Lopez, I. Obeid, and J. Picone, “Improved EEG Event Classification Using Differential Energy,” in Proceedings of the IEEE Signal Processing in Medicine and Biology Symposium, 2015, pp. 1–4. https://doi.org/10.1109/SPMB.2015.7405421. [7] V. Shah, C. Campbell, I. Obeid, and J. Picone, “Improved Spatio-Temporal Modeling in Automated Seizure Detection using Channel-Dependent Posteriors,” Neurocomputing, 2021. [8] W. Tatum, A. Husain, S. Benbadis, and P. Kaplan, Handbook of EEG Interpretation. New York City, New York, USA: Demos Medical Publishing, 2007. [9] D. P. Bovet and C. Marco, Understanding the Linux Kernel, 3rd ed. O’Reilly Media, Inc., 2005. https://www.oreilly.com/library/view/understanding-the-linux/0596005652/. [10] V. Shah et al., “The Temple University Hospital Seizure Detection Corpus,” Front. Neuroinform., vol. 12, pp. 1–6, 2018. https://doi.org/10.3389/fninf.2018.00083. [11] F. Pedregosa et al., “Scikit-learn: Machine Learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011. https://dl.acm.org/doi/10.5555/1953048.2078195. [12] J. Gotman, D. Flanagan, J. Zhang, and B. Rosenblatt, “Automatic seizure detection in the newborn: Methods and initial evaluation,” Electroencephalogr. Clin. Neurophysiol., vol. 103, no. 3, pp. 356–362, 1997. https://doi.org/10.1016/S0013-4694(97)00003-9. 

   more » « less
2. PointPWC-Net: Cost Volume on Point Clouds for (Self-)Supervised Scene Flow Estimation

   https://doi.org/10.1007/978-3-030-58558-7_6  

   Wu, Wenxuan ; Wang, Zhi Yuan ; Li, Zhuwen ; Liu, Wei ; Fuxin, Li ( January 2020 , ECCV 2020)

   null (Ed.)

   We propose a novel end-to-end deep scene flow model, called PointPWC-Net, that directly processes 3D point cloud scenes with large motions in a coarse-to-fine fashion. Flow computed at the coarse level is upsampled and warped to a finer level, enabling the algorithm to accommodate for large motion without a prohibitive search space. We introduce novel cost volume, upsampling, and warping layers to efficiently handle 3D point cloud data. Unlike traditional cost volumes that require exhaustively computing all the cost values on a high-dimensional grid, our point-based formulation discretizes the cost volume onto input 3D points, and a PointConv operation efficiently computes convolutions on the cost volume. Experiment results on FlyingThings3D and KITTI outperform the state-of-the-art by a large margin. We further explore novel self-supervised losses to train our model and achieve comparable results to state-of-the-art trained with supervised loss. Without any fine-tuning, our method also shows great generalization ability on the KITTI Scene Flow 2015 dataset, outperforming all previous methods. The code is released at https://github.com/DylanWusee/PointPWC. 

   more » « less
3. Multiscale Model Reduction of the Unsaturated Flow Problem in Heterogeneous Porous Media with Rough Surface Topography

   https://doi.org/10.3390/math8060904  

   Spiridonov, Denis ; Vasilyeva, Maria ; Chung, Eric T. ; Efendiev, Yalchin ; Jana, Raghavendra ( June 2020 , Mathematics)

   In this paper, we consider unsaturated filtration in heterogeneous porous media with rough surface topography. The surface topography plays an important role in determining the flow process and includes multiscale features. The mathematical model is based on the Richards’ equation with three different types of boundary conditions on the surface: Dirichlet, Neumann, and Robin boundary conditions. For coarse-grid discretization, the Generalized Multiscale Finite Element Method (GMsFEM) is used. Multiscale basis functions that incorporate small scale heterogeneities into the basis functions are constructed. To treat rough boundaries, we construct additional basis functions to take into account the influence of boundary conditions on rough surfaces. We present numerical results for two-dimensional and three-dimensional model problems. To verify the obtained results, we calculate relative errors between the multiscale and reference (fine-grid) solutions for different numbers of multiscale basis functions. We obtain a good agreement between fine-grid and coarse-grid solutions. 

   more » « less
4. Incremental and Approximate Computations for Accelerating Deep CNN Inference

   https://doi.org/10.1145/3397461  

   Nakandala, Supun ; Nagrecha, Kabir ; Kumar, Arun ; Papakonstantinou, Yannis ( December 2020 , ACM Transactions on Database Systems)

   null (Ed.)

   Deep learning now offers state-of-the-art accuracy for many prediction tasks. A form of deep learning called deep convolutional neural networks (CNNs) are especially popular on image, video, and time series data. Due to its high computational cost, CNN inference is often a bottleneck in analytics tasks on such data. Thus, a lot of work in the computer architecture, systems, and compilers communities study how to make CNN inference faster. In this work, we show that by elevating the abstraction level and re-imagining CNN inference as queries , we can bring to bear database-style query optimization techniques to improve CNN inference efficiency. We focus on tasks that perform CNN inference repeatedly on inputs that are only slightly different . We identify two popular CNN tasks with this behavior: occlusion-based explanations (OBE) and object recognition in videos (ORV). OBE is a popular method for “explaining” CNN predictions. It outputs a heatmap over the input to show which regions (e.g., image pixels) mattered most for a given prediction. It leads to many re-inference requests on locally modified inputs. ORV uses CNNs to identify and track objects across video frames. It also leads to many re-inference requests. We cast such tasks in a unified manner as a novel instance of the incremental view maintenance problem and create a comprehensive algebraic framework for incremental CNN inference that reduces computational costs. We produce materialized views of features produced inside a CNN and connect them with a novel multi-query optimization scheme for CNN re-inference. Finally, we also devise novel OBE-specific and ORV-specific approximate inference optimizations exploiting their semantics. We prototype our ideas in Python to create a tool called Krypton that supports both CPUs and GPUs. Experiments with real data and CNNs show that Krypton reduces runtimes by up to 5× (respectively, 35×) to produce exact (respectively, high-quality approximate) results without raising resource requirements. 

   more » « less
5. Expedition 397T Preliminary Report: Return to Walvis Ridge Hotspot

    Sager, W. ; Blum, P. ( December 2022 , Preliminary report)

   International Ocean Discovery Program Expedition 397T sought to address the shortage of drilling time caused by COVID-19 mitigation during Expedition 391 (Walvis Ridge Hotspot) by drilling at two sites omitted from the earlier cruise. A week of coring time was added to a transit of JOIDES Resolution from Cape Town to Lisbon, which would cross Walvis Ridge on its way north. These two sites were located on two of the three seamount trails that emerge from the split in Walvis Ridge morphology into several seamount chains at 2°E. Site U1584 (proposed Site GT-6A) sampled the Gough track on the east, and Site U1585 (proposed Site TT-4A) sampled the Tristan track on the west. Together with Site U1578, drilled on the Center track during Expedition 391, they form a transect across the northern Walvis Ridge Guyot Province. The goal was to core seamount basalts and associated volcanic material for geochemical and isotopic, geochronologic, paleomagnetic, and volcanologic study. Scientifically, one emphasis was to better understand the split in geochemical and isotopic signatures that occurs at the morphologic split. Geochronology would add to the established age progression but also give another dimension to understanding Walvis Ridge seamount formation by giving multiple ages at the same sites. The paleomagnetic study seeks to establish paleolatitudes for Walvis Ridge sites for comparison with those published from hotspot seamount chains in the Pacific, in particular to test whether a component of true polar wander affects hotspot paleolatitude. Hole U1584A cored a 66.4 m thick sedimentary and volcaniclastic section with two lithostratigraphic units. Unit I is a 23 m thick sequence of bioturbated clay and nannofossil chalk with increasing volcaniclastic content downhole. Unit II is a >43 m thick sequence of lapillistone with basalt fragments. Because the seismic section crossing the site shows no evidence as to the depth of the volcaniclastic cover, coring was terminated early. Because there were no other shallow nearby sites with different character on existing seismic lines, the unused operations time from Site U1584 was shifted to the next site. The seismic reflector interpreted as the top of igneous rock at Site U1585 once again resulted from volcaniclastic deposits. Hole U1585A coring began at 144.1 mbsf and penetrated a 273.5 m thick sedimentary and volcaniclastic section atop a 81.2 m thick series of massive basalt flows. The hole was terminated at 498.8 mbsf because allotted operational time expired. The sedimentary section contains four main units. Unit I (144.1–157.02 mbsf) is a bioturbated nannofossil chalk with foraminifera, similar to the shallowest sediments recovered at Site U1584. Unit II (157.02–249.20 mbsf), which is divided into two subunits, is a 92.2 m thick succession of massive and bedded pumice and scoria lapillistone with increased reworking, clast alteration, and tuffaceous chalk intercalations downhole. Unit III (249.20–397.76 mbsf) is 148.6 m thick and consists of a complex succession of pink to greenish gray tuffaceous chalk containing multiple thin, graded ash turbidites and tuffaceous ash layers; intercalated tuffaceous chalk slumps; and several thick coarse lapilli and block-dominated volcaniclastic layers. Befitting the complexity, it is divided into eight subunits (IIIA–IIIH). Three of these subunits (IIIA, IIID, and IIIG) are mainly basalt breccias. Unit IV (397.76–417.60 mbsf) is a volcanic breccia, 19.8 m thick, containing mostly juvenile volcaniclasts. The igneous section, Unit V (417.60–498.80 mbsf) is composed of a small number of massive basaltic lava flows. It is divided into three lithologic units, with Unit 2 represented by a single 3 cm piece of quenched basalt with olivine phenocrysts in a microcrystalline groundmass. This piece may represent a poorly recovered set of pillow lavas. Unit 1 is sparsely to highly olivine-clinopyroxene ± plagioclase phyric massive basalt and is divided into Subunits 1a and 1b based on textural and mineralogical differences, which suggests that they are two different flows. Unit 3 also consists of two massive lava flows with no clear boundary features. Subunit 3a is a 10.3 m thick highly clinopyroxene-plagioclase phyric massive basalt flow with a fine-grained groundmass. Subunit 3b is a featureless massive basalt flow that is moderately to highly clinopyroxene-olivine-plagioclase phyric and >43.7 m thick. Alteration of the lava flows is patchy and moderate to low in grade, with two stages, one at a higher temperature and one at a low temperature, both focused around fractures. The Site U1585 chronologic succession from basalt flows to pelagic sediment indicates volcanic construction and subsidence. Lava eruptions were followed by inundation and shallow-water volcaniclastic sediment deposition, which deepened over time to deepwater conditions. Although the massive flows were probably erupted in a short time and have little variability, volcaniclasts in the sediments may provide geochemical and geochronologic data from a range of time and sources. Chemical analyses indicate that Site U1585 basalt samples are mostly alkalic basalt, with a few trachybasalt flow and clast samples and one basaltic trachyandesite clast. Ti/V ratios lie mostly within the oceanic island basalt (OIB) field but overlap the mid-ocean-ridge basalt (MORB) field. Only a handful of clasts from Site U1584 were analyzed, but geochemical data are similar. Paleomagnetic data from Site U1585 indicate that the sediments and basalt units are strongly magnetic and mostly give coherent inclination data, which indicates that the basaltic section and ~133 m of overlying volcaniclastic sediment is reversely polarized and that this reversal is preserved in a core. Above this, the rest of the sediment section records two normal and two reversed zones. Although there are not enough basalt flows to give a reliable paleolatitude, it may be possible to attain such a result from the sediments. 

   more » « less