Search results for: aerial imaging and detection

Authors: Niloufar Yadgari

Abstract:

GANs are a potent form of deep learning models that have found success in various fields. They are part of the larger group of generative techniques, which aim to produce authentic data using a probabilistic model that learns distributions from actual samples. In clinical settings, GANs have demonstrated improved abilities in capturing spatially intricate, nonlinear, and possibly subtle disease impacts in contrast to conventional generative techniques. This review critically evaluates the current research on how GANs are being used in imaging studies of different neurological conditions like Alzheimer's disease, brain tumors, aging of the brain, and multiple sclerosis. We offer a clear explanation of different GAN techniques for each use case in neuroimaging and delve into the key hurdles, unanswered queries, and potential advancements in utilizing GANs in this field. Our goal is to connect advanced deep learning techniques with neurology studies, showcasing how GANs can assist in clinical decision-making and enhance our comprehension of the structural and functional aspects of brain disorders.

Keywords: GAN, pathology, generative adversarial network, neuro imaging

Procedia PDF Downloads 32

Authors: M. A. Alavianmehr, A. Tashk, A. Sodagaran

Abstract:

Modeling background and moving objects are significant techniques for video surveillance and other video processing applications. This paper presents a foreground detection algorithm that is robust against illumination changes and noise based on adaptive mixture Gaussian model (GMM), and provides a novel and practical choice for intelligent video surveillance systems using static cameras. In the previous methods, the image of still objects (background image) is not significant. On the contrary, this method is based on forming a meticulous background image and exploiting it for separating moving objects from their background. The background image is specified either manually, by taking an image without vehicles, or is detected in real-time by forming a mathematical or exponential average of successive images. The proposed scheme can offer low image degradation. The simulation results demonstrate high degree of performance for the proposed method.

Keywords: image processing, background models, video surveillance, foreground detection, Gaussian mixture model

Procedia PDF Downloads 516

Authors: Jia Li, Xing Wei, Yuchen Hong, Yang Lu

Abstract:

Detecting vehicle behavior has always been the focus of intelligent transportation, but with the explosive growth of the number of vehicles and the complexity of the road environment, the vehicle behavior videos captured by traditional surveillance have been unable to satisfy the study of vehicle behavior. The traditional method of manually labeling vehicle behavior is too time-consuming and labor-intensive, but the existing object detection and tracking algorithms have poor practicability and low behavioral location detection rate. This paper proposes a vehicle behavior detection algorithm based on the dual-stream convolution network and the multi-dimensional video dynamic detection network. In the videos, the straight-line behavior of the vehicle will default to the background behavior. The Changing lanes, turning and turning around are set as target behaviors. The purpose of this model is to automatically mark the target behavior of the vehicle from the untrimmed videos. First, the target behavior proposals in the long video are extracted through the dual-stream convolution network. The model uses a dual-stream convolutional network to generate a one-dimensional action score waveform, and then extract segments with scores above a given threshold M into preliminary vehicle behavior proposals. Second, the preliminary proposals are pruned and identified using the multi-dimensional video dynamic detection network. Referring to the hierarchical reinforcement learning, the multi-dimensional network includes a Timer module and a Spacer module, where the Timer module mines time information in the video stream and the Spacer module extracts spatial information in the video frame. The Timer and Spacer module are implemented by Long Short-Term Memory (LSTM) and start from an all-zero hidden state. The Timer module uses the Transformer mechanism to extract timing information from the video stream and extract features by linear mapping and other methods. Finally, the model fuses time information and spatial information and obtains the location and category of the behavior through the softmax layer. This paper uses recall and precision to measure the performance of the model. Extensive experiments show that based on the dataset of this paper, the proposed model has obvious advantages compared with the existing state-of-the-art behavior detection algorithms. When the Time Intersection over Union (TIoU) threshold is 0.5, the Average-Precision (MP) reaches 36.3% (the MP of baselines is 21.5%). In summary, this paper proposes a vehicle behavior detection model based on multi-dimensional dynamic detection network. This paper introduces spatial information and temporal information to extract vehicle behaviors in long videos. Experiments show that the proposed algorithm is advanced and accurate in-vehicle timing behavior detection. In the future, the focus will be on simultaneously detecting the timing behavior of multiple vehicles in complex traffic scenes (such as a busy street) while ensuring accuracy.

Keywords: vehicle behavior detection, convolutional neural network, long short-term memory, deep learning

Procedia PDF Downloads 130

Authors: Muhammad Ali

Abstract:

Cyber-attacks and anomaly detection on the Internet of Things (IoT) infrastructure is emerging concern in the domain of data-driven intrusion. Rapidly increasing IoT risk is now making headlines around the world. denial of service, malicious control, data type probing, malicious operation, DDos, scan, spying, and wrong setup are attacks and anomalies that can affect an IoT system failure. Everyone talks about cyber security, connectivity, smart devices, and real-time data extraction. IoT devices expose a wide variety of new cyber security attack vectors in network traffic. For further than IoT development, and mainly for smart and IoT applications, there is a necessity for intelligent processing and analysis of data. So, our approach is too secure. We train several machine learning models that have been compared to accurately predicting attacks and anomalies on IoT systems, considering IoT applications, with ANOVA-based feature selection with fewer prediction models to evaluate network traffic to help prevent IoT devices. The machine learning (ML) algorithms that have been used here are KNN, SVM, NB, D.T., and R.F., with the most satisfactory test accuracy with fast detection. The evaluation of ML metrics includes precision, recall, F1 score, FPR, NPV, G.M., MCC, and AUC & ROC. The Random Forest algorithm achieved the best results with less prediction time, with an accuracy of 99.98%.

Keywords: machine learning, analysis of variance, Internet of Thing, network security, intrusion detection

Procedia PDF Downloads 125

Authors: Alisha M. Ebrahim, Bijoy K. Menon, Eddy Lang, Shelagh B. Coutts, Katie Lin

Abstract:

Introduction: Frontline emergency physicians require clear and evidence-based approaches to guide neuroimaging investigations for patients presenting with suspected acute stroke or transient ischemic attack (TIA). Various forms of computed tomography (CT) are currently available for initial investigation, including non-contrast CT (NCCT), CT angiography head and neck (CTA), and CT perfusion (CTP). However, there is uncertainty around optimal imaging choice for cost-effectiveness, particularly for minor or resolved neurological symptoms. In addition to the cost of CTA and CTP testing, there is also a concern for increased incidental findings, which may contribute to the burden of overdiagnosis. Methods: In this cross-sectional observational study, analysis was conducted on 586 anonymized triage and diagnostic imaging (DI) reports for neuroimaging orders completed on patients presenting to adult emergency departments (EDs) with a suspected stroke or TIA from January-December 2019. The primary outcome of interest is the diagnostic yield of NCCT+CTA compared to NCCT alone for patients presenting to urban academic EDs with Canadian Emergency Department Information System (CEDIS) complaints of “symptoms of stroke” (specifically acute stroke and TIA indications). DI reports were coded into 4 pre-specified categories (endorsed by a panel of stroke experts): no abnormalities, clinically significant findings (requiring immediate or follow-up clinical action), incidental findings (not meeting prespecified criteria for clinical significance), and both significant and incidental findings. Standard descriptive statistics were performed. A two-sided p-value <0.05 was considered significant. Results: 75% of patients received NCCT+CTA imaging, 21% received NCCT alone, and 4% received NCCT+CTA+CTP. The diagnostic yield of NCCT+CTA imaging for prespecified clinically significant findings was 24%, compared to only 9% in those who received NCCT alone. The proportion of incidental findings was 30% in the NCCT only group and 32% in the NCCT+CTA group. CTP did not significantly increase the yield of significant or incidental findings. Conclusion: In this cohort of patients presenting with suspected stroke or TIA, an NCCT+CTA neuroimaging strategy had a higher diagnostic yield for clinically significant findings than NCCT alone without significantly increasing the number of incidental findings identified.

Keywords: stroke, diagnostic yield, neuroimaging, emergency department, CT

Procedia PDF Downloads 100

Authors: Nizar Karim Uribe-Orihuela, Fernando Brambila-Paz, Ivette Caldelas, Rodrigo Montufar-Chaveznava

Abstract:

In this paper, we present the advances corresponding to the implementation of a strategy to visualize geographical data using a Software Development Kit (SDK) for video games. We use multispectral images from Landsat 7 platform and Laser Imaging Detection and Ranging (LIDAR) data from The National Institute of Geography and Statistics of Mexican (INEGI). We select a place of interest to visualize from Landsat platform and make some processing to the image (rotations, atmospheric correction and enhancement). The resulting image will be our gray scale color-map to fusion with the LIDAR data, which was selected using the same coordinates than in Landsat. The LIDAR data is translated to 8-bit raw data. Both images are fused in a software developed using Unity (an SDK employed for video games). The resulting image is then displayed and can be explored moving around. The idea is the software could be used for students of geology and geophysics at the Engineering School of the National University of Mexico. They will download the software and images corresponding to a geological place of interest to a smartphone and could virtually visit and explore the site with a virtual reality visor such as Google cardboard.

Keywords: virtual reality, interactive technologies, geographical data visualization, video games technologies, educational material

Procedia PDF Downloads 246

Authors: Raahil Sheikh, Abhishek Maurya, Priya Gujjar, Himanshu Dwivedi, Prathamesh Minde

Abstract:

UAVs have been an initial member of our environment since it's the first used by Austrian warfare in Venice. At that stage, they were just pilotless balloons equipped with bombs to be dropped on enemy territory. Over time, technological advancements allowed UAVs to be controlled remotely or autonomously. This study shall mainly focus on the intensification of pre-existing manual drones equipping them with a variety of sensors and making them autonomous, and capable, and purposing them for a variety of roles, including thermal sensing, data collection, tracking creatures, forest fires, volcano detection, hydrothermal studies, urban heat, Island measurement, and other environmental research. The system can also be used for reconnaissance, research, 3D mapping, and search and rescue missions. This study mainly focuses on automating tedious tasks and reducing human errors as much as possible, reducing deployment time, and increasing the overall efficiency, efficacy, and reliability of the UAVs. Creation of a comprehensive Ground Control System UI (GCS) enabling less trained professionals to be able to use the UAV with maximum potency. With the inclusion of such an autonomous system, artificially intelligent paths and environmental gusts and concerns can be avoided.

Keywords: UAV, drone, autonomous system, thermal imaging

Procedia PDF Downloads 75

Authors: Elmahdi Khalil, Saad Enniari, Mostapha Zbakh

Abstract:

Cloud computing promises to increase innovation and the velocity with witch applications are deployed, all while helping any enterprise meet most IT service needs at a lower total cost of ownership and higher return investment. As the march of cloud continues, it brings both new opportunities and new security challenges. To take advantages of those opportunities while minimizing risks, we think that Intrusion Detection Systems (IDS) integrated in the cloud is one of the best existing solutions nowadays in the field. The concept of intrusion detection was known since past and was first proposed by a well-known researcher named Anderson in 1980's. Since that time IDS's are evolving. Although, several efforts has been made in the area of Intrusion Detection systems for cloud computing environment, many attacks still prevail. Therefore, the work presented in this paper proposes a multi criteria analysis and a comparative study between several IDS architectures designated to work in a cloud computing environments. To achieve this objective, in the first place we will search in the state of the art of several consistent IDS architectures designed to work in a cloud environment. Whereas, in a second step we will establish the criteria that will be useful for the evaluation of architectures. Later, using the approach of multi criteria decision analysis Mac Beth (Measuring Attractiveness by a Categorical Based Evaluation Technique we will evaluate the criteria and assign to each one the appropriate weight according to their importance in the field of IDS architectures in cloud computing. The last step is to evaluate architectures against the criteria and collecting results of the model constructed in the previous steps.

Keywords: cloud computing, cloud security, intrusion detection/prevention system, multi-criteria decision analysis

Procedia PDF Downloads 472

Authors: K. Tunde Olagunju, C. Scott Allen, Freek Van Der Meer

Abstract:

Identification of hydrocarbon oil in remote sensing images is often the first step in monitoring oil during spill events. Most remote sensing methods adopt techniques for hydrocarbon identification to achieve detection in order to model an appropriate cleanup program. Identification on optical sensors does not only allow for detection but also for characterization and quantification. Until recently, in optical remote sensing, quantification and characterization are only potentially possible using high-resolution laboratory and airborne imaging spectrometers (hyperspectral data). Unlike multispectral, hyperspectral data are not freely available, as this data category is mainly obtained via airborne survey at present. In this research, two (2) operational high-resolution multispectral satellites (WorldView-3 and Sentinel-2) are theoretically assessed for their suitability for hydrocarbon characterization, using the hydrocarbon spectral slope model (HYSS). This method utilized the two most persistent hydrocarbon diagnostic/absorption features at 1.73 µm and 2.30 µm for hydrocarbon mapping on multispectral data. In this research, spectra measurement of seven (7) different hydrocarbon oils (crude and refined oil) taken on ten (10) different substrates with the use of laboratory ASD Fieldspec were convolved to Sentinel-2 and WorldView-3 resolution, using their full width half maximum (FWHM) parameter. The resulting hydrocarbon slope values obtained from the studied samples enable clear qualitative discrimination of most hydrocarbons, despite the presence of different background substrates, particularly on WorldView-3. Due to close conformity of central wavelengths and narrow bandwidths to key hydrocarbon bands used in HYSS, the statistical significance for qualitative analysis on WorldView-3 sensors for all studied hydrocarbon oil returned with 95% confidence level (P-value ˂ 0.01), except for Diesel. Using multifactor analysis of variance (MANOVA), the discriminating power of HYSS is statistically significant for most hydrocarbon-substrate combinations on Sentinel-2 and WorldView-3 FWHM, revealing the potential of these two operational multispectral sensors as rapid response tools for hydrocarbon mapping. One notable exception is highly transmissive hydrocarbons on Sentinel-2 data due to the non-conformity of spectral bands with key hydrocarbon absorptions and the relatively coarse bandwidth (> 100 nm).

Keywords: hydrocarbon, oil spill, remote sensing, hyperspectral, multispectral, hydrocarbon-substrate combination, Sentinel-2, WorldView-3

Procedia PDF Downloads 216

Authors: Raahil Sheikh, Prathamesh Minde, Priya Gujjar, Himanshu Dwivedi, Abhishek Maurya

Abstract:

UAVs have been an initial member of our environment since it's the first used by Austrian warfare in Venice. At that stage, they were just pilotless balloons equipped with bombs to be dropped on enemy territory. Over time, technological advancements allowed UAVs to be controlled remotely or autonomously. This study shall mainly focus on the intensification of pre-existing manual drones equipping them with a variety of sensors and making them autonomous, and capable, and purposing them for a variety of roles, including thermal sensing, data collection, tracking creatures, forest fires, volcano detection, hydrothermal studies, urban heat, Island measurement, and other environmental research. The system can also be used for reconnaissance, research, 3D mapping, and search and rescue missions. This study mainly focuses on automating tedious tasks and reducing human errors as much as possible, reducing deployment time, and increasing the overall efficiency, efficacy, and reliability of the UAVs. Creation of a comprehensive Ground Control System UI (GCS) enabling less trained professionals to be able to use the UAV with maximum potency. With the inclusion of such an autonomous system, artificially intelligent paths and environmental gusts and concerns can be avoided

Keywords: UAV, autonomous systems, drones, geo thermal imaging

Procedia PDF Downloads 85

Authors: Marjila Burhanzoi, Kenta Onohara, Tomoaki Ikegami

Abstract:

Faults in photovoltaic (PV) modules should be detected to the greatest extent as early as possible. For that conventional fault detection methods such as electrical characterization, visual inspection, infrared (IR) imaging, ultraviolet fluorescence and electroluminescence (EL) imaging are used, but they either fail to detect the location or category of fault, or they require expensive equipment and are not convenient for onsite application. Hence, these methods are not convenient to use for monitoring small-scale PV systems. Therefore, low cost and efficient inspection techniques with the ability of onsite application are indispensable for PV modules. In this study in order to establish efficient inspection technique, correlation between faults and magnetic flux density on the surface is of crystalline PV modules are investigated. Magnetic flux on the surface of normal and faulted PV modules is measured under the short circuit and illuminated conditions using two different sensor devices. One device is made of small integrated sensors namely 9-axis motion tracking sensor with a 3-axis electronic compass embedded, an IR temperature sensor, an optical laser position sensor and a microcontroller. This device measures the X, Y and Z components of the magnetic flux density (Bx, By and Bz) few mm above the surface of a PV module and outputs the data as line graphs in LabVIEW program. The second device is made of a laser optical sensor and two magnetic line sensor modules consisting 16 pieces of magnetic sensors. This device scans the magnetic field on the surface of PV module and outputs the data as a 3D surface plot of the magnetic flux intensity in a LabVIEW program. A PC equipped with LabVIEW software is used for data acquisition and analysis for both devices. To show the effectiveness of this method, measured results are compared to those of a normal reference module and their EL images. Through the experiments it was confirmed that the magnetic field in the faulted areas have different profiles which can be clearly identified in the measured plots. Measurement results showed a perfect correlation with the EL images and using position sensors it identified the exact location of faults. This method was applied on different modules and various faults were detected using it. The proposed method owns the ability of on-site measurement and real-time diagnosis. Since simple sensors are used to make the device, it is low cost and convenient to be sued by small-scale or residential PV system owners.

Keywords: fault diagnosis, fault location, integrated sensors, PV modules

Procedia PDF Downloads 224

Authors: Heshalini Rajagopal, Li Sze Chow, Raveendran Paramesran

Abstract:

Magnetic Resonance Imaging (MRI) is one of the most important medical imaging modality. Subjective assessment of the image quality is regarded as the gold standard to evaluate MR images. In this study, a database of 210 MR images which contains ten reference images and 200 distorted images is presented. The reference images were distorted with four types of distortions: Rician Noise, Gaussian White Noise, Gaussian Blur and DCT compression. The 210 images were assessed by ten subjects. The subjective scores were presented in Difference Mean Opinion Score (DMOS). The DMOS values were compared with four FR-IQA metrics. We have used Pearson Linear Coefficient (PLCC) and Spearman Rank Order Correlation Coefficient (SROCC) to validate the DMOS values. The high correlation values of PLCC and SROCC shows that the DMOS values are close to the objective FR-IQA metrics.

Keywords: medical resonance (MR) images, difference mean opinion score (DMOS), full reference image quality assessment (FR-IQA)

Procedia PDF Downloads 458

Authors: Chetan Gupta, Ramesh Gupta

Abstract:

Unmanned aerial vehicles, of all sizes, are prime targets of the wing morphing concept as their lightweight structures demand high aerodynamic stability while traversing unsteady atmospheric conditions. In this research study, a hybrid morphing technology is developed to aid the trailing edge of the aircraft wing to alter its camber as a monolithic element rather than functioning as conventional appendages like flaps. Kinematic tailoring, actuation techniques involving shape memory alloys (SMA), piezoelectrics – individually fall short of providing a simplistic solution to the conundrum of morphing aircraft wings. On the other hand, the feature of negligible hysteresis while actuating using compliant mechanisms has shown higher levels of applicability and deliverability in morphing wings of even large aircrafts. This research paper delves into designing a wing section model with a periodic, multi-stable compliant structure requiring lower orders of topological optimization. The design is sub-divided into three smaller domains with external hyperelastic connections to achieve deflections ranging from -15° to +15° at the trailing edge of the wing. To facilitate this functioning, a hybrid actuation system by combining the larger bandwidth feature of piezoelectric macro-fibre composites and relatively higher work densities of shape memory alloy wires are used. Finite element analysis is applied to optimize piezoelectric actuation of the internal compliant structure. A coupled fluid-surface interaction analysis is conducted on the wing section during morphing to study the development of the velocity boundary layer at low Reynold’s numbers of airflow.

Keywords: compliant mechanism, hybrid morphing, piezoelectrics, shape memory alloys

Procedia PDF Downloads 312

Authors: B. Lokuketagoda, M. Weerakoon, U. Madushan, A. N. Senaratne, K. Y. Abeywardena

Abstract:

Ransomware has become a common threat in past few years and the recent threat reports show an increase of growth in Ransomware infections. Researchers have identified different variants of Ransomware families since 2015. Lack of knowledge of the user about the threat is a major concern. Ransomware detection methodologies are still growing through the industry. Email is the easiest method to send Ransomware to its victims. Uninformed users tend to click on links and attachments without much consideration assuming the emails are genuine. As a solution to this in this paper R-Killer Ransomware detection tool is introduced. Tool can be integrated with existing email services. The core detection Engine (CDE) discussed in the paper focuses on separating suspicious samples from emails and handling them until a decision is made regarding the suspicious mail. It has the capability of preventing execution of identified ransomware processes. On the other hand, Sandboxing and URL analyzing system has the capability of communication with public threat intelligence services to gather known threat intelligence. The R-Killer has its own mechanism developed in its Proactive Monitoring System (PMS) which can monitor the processes created by downloaded email attachments and identify potential Ransomware activities. R-killer is capable of gathering threat intelligence without exposing the user’s data to public threat intelligence services, hence protecting the confidentiality of user data.

Keywords: ransomware, deep learning, recurrent neural networks, email, core detection engine

Procedia PDF Downloads 214

Authors: Mohamad Kassab, Amal El Fallah Seghrouchni, Frederic Barbaresco, Raed Abu Zitar

Abstract:

Detecting objects such as drones is a challenging task as their relative size and maneuvering capabilities deceive machine learning models and cause them to misclassify drones as birds or other objects. In this work, we investigate applying several deep learning techniques to benchmark real data sets of flying drones. A deep learning paradigm is proposed for the purpose of mitigating the complexity of those systems. The proposed paradigm consists of a hybrid between the AdderNet deep learning paradigm and the Single Shot Detector (SSD) paradigm. The goal was to minimize multiplication operations numbers in the filtering layers within the proposed system and, hence, reduce complexity. Some standard machine learning technique, such as SVM, is also tested and compared to other deep learning systems. The data sets used for training and testing were either complete or filtered in order to remove the images with mall objects. The types of data were RGB or IR data. Comparisons were made between all these types, and conclusions were presented.

Keywords: drones detection, deep learning, birds versus drones, precision of detection, AdderNet

Procedia PDF Downloads 182

Authors: Ava Zarif Sanayei, Sedigheh Sina

Abstract:

Despite the advancement of radiation-free interventions in the gastrointestinal and hepatobiliary fields, endoscopy and endoscopic retrograde cholangiopancreatography (ERCP) remain indispensable procedures that necessitate radiation exposure. ERCP, in particular, relies heavily on radiation-guided imaging to ensure precise delivery of therapy. Meanwhile, interventional radiology (IR) procedures also utilize imaging modalities like X-rays and CT scans to guide therapy, often under local anesthesia via small needle insertion. However, the complexity of these procedures raises concerns about radiation exposure to healthcare professionals, including nurses, who play a crucial role in these interventions. This study aims to assess the radiation exposure to the hands and fingers of nurses 1 and 2, who are directly involved in ERCP procedures utilizing (TLD-100) dosimeters at the Gastrointestinal Endoscopy department of a clinic in Shiraz, Iran. The dosimeters were initially calibrated using various phantoms and then a group was prepared and used over a two-month period. For personal equivalent dose measurement, two TLD chips were mounted on a finger ring to monitor exposure to the hands and fingers. Upon completion of the monitoring period, the TLDs were analyzed using a TLD reader, showing that Nurse 1 received an equivalent dose of 298.26 µSv and Nurse 2 received an equivalent dose of 195.39 µSv. The investigation revealed that the total radiation exposure to the nurses did not exceed the annual limit for occupational exposure. Nevertheless, it is essential to prioritize radiation protection measures to prevent potential harm. The study showed that positioning staff members and placing two nurses in a specific location contributed to somehow equal doses. To reduce exposure further, we suggest providing education and training on radiation safety principles, particularly for technologists.

Keywords: dose measurement, ERCP, interventional radiology, medical imaging

Procedia PDF Downloads 34

Authors: Kelemewerk Destalem, Joongjae Cho, Jaeseong Lee, Ju H. Park, Joonhyuk Yoo

Abstract:

Background subtraction and temporal difference are often used for moving object detection in video. Both approaches are computationally simple and easy to be deployed in real-time image processing. However, while the background subtraction is highly sensitive to dynamic background and illumination changes, the temporal difference approach is poor at extracting relevant pixels of the moving object and at detecting the stopped or slowly moving objects in the scene. In this paper, we propose a moving object detection scheme based on adaptive background subtraction and temporal difference exploiting dynamic background updates. The proposed technique consists of a histogram equalization, a linear combination of background and temporal difference, followed by the novel frame-based and pixel-based background updating techniques. Finally, morphological operations are applied to the output images. Experimental results show that the proposed algorithm can solve the drawbacks of both background subtraction and temporal difference methods and can provide better performance than that of each method.

Keywords: background subtraction, background updating, real time, light weight algorithm, temporal difference

Procedia PDF Downloads 342

Authors: Sean Paulsen, Michael Casey

Abstract:

In this work we present a self-supervised pretraining framework for transformers on functional Magnetic Resonance Imaging (fMRI) data. First, we pretrain our architecture on two self-supervised tasks simultaneously to teach the model a general understanding of the temporal and spatial dynamics of human auditory cortex during music listening. Our pretraining results are the first to suggest a synergistic effect of multitask training on fMRI data. Second, we finetune the pretrained models and train additional fresh models on a supervised fMRI classification task. We observe significantly improved accuracy on held-out runs with the finetuned models, which demonstrates the ability of our pretraining tasks to facilitate transfer learning. This work contributes to the growing body of literature on transformer architectures for pretraining and transfer learning with fMRI data, and serves as a proof of concept for our pretraining tasks and multitask pretraining on fMRI data.

Keywords: transfer learning, fMRI, self-supervised, brain decoding, transformer, multitask training

Procedia PDF Downloads 90

Authors: Samson Munn, Virginia H. Kim, Huija Chen, Sean Maldonado, Michelle Kim, Paul Koscheski, Babak N. Kalantari, Gregory Eckel, Albert Lee

Abstract:

In terms of screening mammography quality, neither the portion of reports that advise call-back imaging that should be bilateral versus unilateral nor how much the unilateral call-backs may appropriately diverge from 50–50 (left versus right) is known. Many factors may affect detection laterality: display arrangement, reflections preferentially striking one display location, hanging protocols, seating positions with respect to others and displays, visual field cuts, health, etc. The call-back bilateral fraction may reflect radiologist experience (not in our data) or confidence level. Thus, laterality and bilaterality of call-backs advised in screening mammography reports could be worthy quality metrics. Here, laterality data did not reveal a concern until drilling down to individuals. Bilateral screening mammogram report recommendations by five breast imaging, attending radiologists at Harbor-UCLA Medical Center (Torrance, California) 9/1/15--8/31/16 and 9/1/16--8/31/17 were retrospectively reviewed. Recommended call-backs for bilateral versus unilateral, and for left versus right, findings were counted. Chi-square (χ²) statistic was applied. Year 1: of 2,665 bilateral screening mammograms, reports of 556 (20.9%) recommended call-back, of which 99 (17.8% of the 556) were for bilateral findings. Of the 457 unilateral recommendations, 222 (48.6%) regarded the left breast. Year 2: of 2,106 bilateral screening mammograms, reports of 439 (20.8%) recommended call-back, of which 65 (14.8% of the 439) were for bilateral findings. Of the 374 unilateral recommendations, 182 (48.7%) regarded the left breast. Individual ranges of call-backs that were bilateral were 13.2–23.3%, 10.2–22.5%, and 13.6–17.9%, by year(s) 1, 2, and 1+2, respectively; these ranges were unrelated to experience level; the two-year mean was 15.8% (SD=1.9%). The lowest χ² p value of the group's sidedness disparities years 1, 2, and 1+2 was > 0.4. Regarding four individual radiologists, the lowest p value was 0.42. However, the fifth radiologist disfavored the left, with p values of 0.21, 0.19, and 0.07, respectively; that radiologist had the greatest number of years of experience. There was a concerning, 93% likelihood that bias against left breast findings evidenced by one of our radiologists was not random. Notably, very soon after the period under review, he retired, presented with leukemia, and died. We call for research to be done, particularly by large departments with many radiologists, of two possible, new, quality metrics in screening mammography: laterality and bilaterality. (Images, patient outcomes, report validity, and radiologist psychological confidence levels were not assessed. No intervention nor subsequent data collection was conducted. This uncomplicated collection of data and simple appraisal were not designed, nor had there been any intention to develop or contribute, to generalizable knowledge (per U.S. DHHS 45 CFR, part 46)).

Keywords: mammography, screening mammography, quality, quality metrics, laterality

Procedia PDF Downloads 162

Authors: Ifedapo Francis Awolowo

Abstract:

The unrelenting series of embarrassing audit failures should stimulate a paradigm shift in accounting. And in this age of information revolution, there is need for a constant improvement on the products or services one offers to the market in order to be relevant. This study explores the perceptions of external auditors, forensic accountants and accounting academics on whether a paradigm shift to forensic accounting can reduce financial statement frauds. Through Neo-empiricism/inductive analytical approach, findings reveal that a paradigm shift to forensic accounting might be the right step in the right direction in order to increase the chances of fraud prevention and detection in the financial statement. This research has implication on accounting education on the need to incorporate forensic accounting into present day accounting curriculum. Accounting professional bodies, accounting standard setters and accounting firms all have roles to play in incorporating forensic accounting education into accounting curriculum. Particularly, there is need to alter the ISA 240 to make the prevention and detection of frauds the responsibilities of bot those charged with the management and governance of companies and statutory auditors.

Keywords: financial statement fraud, forensic accounting, fraud prevention and detection, auditing, audit expectation gap, corporate governance

Procedia PDF Downloads 366

Authors: Anoop T. R., Otman Basir, Robert F. Hess, Eileen E. Birch, Brooke A. Koritala, Reed M. Jost, Becky Luu, David Stager, Ben Thompson

Abstract:

Strabismus refers to a misalignment of the eyes. Early detection and treatment of strabismus in childhood can prevent the development of permanent vision loss due to abnormal development of visual brain areas. We developed a two-stage method for strabismus detection and classification based on photographs of the face. The first stage detects the presence or absence of strabismus, and the second stage classifies the type of strabismus. The first stage comprises face detection using Haar cascade, facial landmark estimation, face alignment, aligned face landmark detection, segmentation of the eye region, and detection of strabismus using VGG 16 convolution neural networks. Face alignment transforms the face to a canonical pose to ensure consistency in subsequent analysis. Using facial landmarks, the eye region is segmented from the aligned face and fed into a VGG 16 CNN model, which has been trained to classify strabismus. The CNN determines whether strabismus is present and classifies the type of strabismus (exotropia, esotropia, and vertical deviation). If stage 1 detects strabismus, the eye region image is fed into stage 2, which starts with the estimation of pupil center coordinates using mask R-CNN deep neural networks. Then, the distance between the pupil coordinates and eye landmarks is calculated along with the angle that the pupil coordinates make with the horizontal and vertical axis. The distance and angle information is used to characterize the degree and direction of the strabismic eye misalignment. This model was tested on 100 clinically labeled images of children with (n = 50) and without (n = 50) strabismus. The True Positive Rate (TPR) and False Positive Rate (FPR) of the first stage were 94% and 6% respectively. The classification stage has produced a TPR of 94.73%, 94.44%, and 100% for esotropia, exotropia, and vertical deviations, respectively. This method also had an FPR of 5.26%, 5.55%, and 0% for esotropia, exotropia, and vertical deviation, respectively. The addition of one more feature related to the location of corneal light reflections may reduce the FPR, which was primarily due to children with pseudo-strabismus (the appearance of strabismus due to a wide nasal bridge or skin folds on the nasal side of the eyes).

Keywords: strabismus, deep neural networks, face detection, facial landmarks, face alignment, segmentation, VGG 16, mask R-CNN, pupil coordinates, angle deviation, horizontal and vertical deviation

Procedia PDF Downloads 93

Authors: Siti Aishah Hasbullah

Abstract:

Studies on identification of pork content in food have grown rapidly to meet the Halal food standard in Malaysia. The used mitochondria DNA (mtDNA) approaches for the identification of pig species is thought to be the most precise marker due to the mtDNA genes are present in thousands of copies per cell, the large variability of mtDNA. The standard method commonly used for DNA detection is based on polymerase chain reaction (PCR) method combined with gel electrophoresis but has major drawback. Its major drawbacks are laborious, need longer time and toxic to handle. Therefore, the need for simplicity and fast assay of DNA is vital and has triggered us to develop DNA biosensors for porcine DNA detection. Therefore, the aim of this project is to develop electrochemical DNA biosensor based on ruthenium (II) complex, [Ru(bpy)2(p-PIP)]2+ as DNA hybridization label. The interaction of DNA and [Ru(bpy)2(p-HPIP)]2+ will be studied by electrochemical transduction using Gold Screen-Printed Electrode (GSPE) modified with gold nanoparticles (AuNPs) and succinimide acrylic microspheres. The electrochemical detection by redox active ruthenium (II) complex was measured by cyclic voltammetry (CV) and differential pulse voltammetry (DPV). The results indicate that the interaction of [Ru(bpy)2(PIP)]2+ with hybridization complementary DNA has higher response compared to single-stranded and mismatch complementary DNA. Under optimized condition, this porcine DNA biosensor incorporated modified GSPE shows good linear range towards porcine DNA.

Keywords: gold, screen printed electrode, ruthenium, porcine DNA

Procedia PDF Downloads 309

Authors: Renata Gerhardt, Detlev Belder

Abstract:

Raman spectroscopy has attracted much attention as a structurally descriptive and label-free detection method. It is particularly suited for chemical analysis given as it is non-destructive and molecules can be identified via the fingerprint region of the spectra. In this work possibilities are investigated how to integrate Raman spectroscopy as a detection method for chip-based chromatography, making use of a droplet interface. A demanding task in lab-on-a-chip applications is the specific and sensitive detection of low concentrated analytes in small volumes. Fluorescence detection is frequently utilized but restricted to fluorescent molecules. Furthermore, no structural information is provided. Another often applied technique is mass spectrometry which enables the identification of molecules based on their mass to charge ratio. Additionally, the obtained fragmentation pattern gives insight into the chemical structure. However, it is only applicable as an end-of-the-line detection because analytes are destroyed during measurements. In contrast to mass spectrometry, Raman spectroscopy can be applied on-chip and substances can be processed further downstream after detection. A major drawback of Raman spectroscopy is the inherent weakness of the Raman signal, which is due to the small cross-sections associated with the scattering process. Enhancement techniques, such as surface enhanced Raman spectroscopy (SERS), are employed to overcome the poor sensitivity even allowing detection on a single molecule level. In SERS measurements, Raman signal intensity is improved by several orders of magnitude if the analyte is in close proximity to nanostructured metal surfaces or nanoparticles. The main gain of lab-on-a-chip technology is the building block-like ability to seamlessly integrate different functionalities, such as synthesis, separation, derivatization and detection on a single device. We intend to utilize this powerful toolbox to realize Raman detection in chip-based chromatography. By interfacing on-chip separations with a droplet generator, the separated analytes are encapsulated into numerous discrete containers. These droplets can then be injected with a silver nanoparticle solution and investigated via Raman spectroscopy. Droplet microfluidics is a sub-discipline of microfluidics which instead of a continuous flow operates with the segmented flow. Segmented flow is created by merging two immiscible phases (usually an aqueous phase and oil) thus forming small discrete volumes of one phase in the carrier phase. The study surveys different chip designs to realize coupling of chip-based chromatography with droplet microfluidics. With regards to maintaining a sufficient flow rate for chromatographic separation and ensuring stable eluent flow over the column different flow rates of eluent and oil phase are tested. Furthermore, the detection of analytes in droplets with surface enhanced Raman spectroscopy is examined. The compartmentalization of separated compounds preserves the analytical resolution since the continuous phase restricts dispersion between the droplets. The droplets are ideal vessels for the insertion of silver colloids thus making use of the surface enhancement effect and improving the sensitivity of the detection. The long-term goal of this work is the first realization of coupling chip based chromatography with droplets microfluidics to employ surface enhanced Raman spectroscopy as means of detection.

Keywords: chip-based separation, chip LC, droplets, Raman spectroscopy, SERS

Procedia PDF Downloads 245

Authors: Sylwia Baluta, Joanna Cabaj, Karol Malecha

Abstract:

The evolution of biosensors was driven by the need for faster and more versatile analytical methods for application in important areas including clinical, diagnostics, food analysis or environmental monitoring, with minimum sample pretreatment. Rapid and sensitive neurotransmitters detection is extremely important in modern medicine. These compounds mainly occur in the brain and central nervous system of mammals. Any changes in the neurotransmitters concentration may lead to many diseases, such as Parkinson’s or schizophrenia. Classical techniques of chemical analysis, despite many advantages, do not permit to obtain immediate results or automatization of measurements.

Keywords: adrenaline, biosensor, dopamine, laccase, tyrosinase

Procedia PDF Downloads 142

Authors: Haonan Hu, Shuge Lei, Dasheng Sun, Huabin Zhang, Kehong Yuan, Jian Dai, Jijun Tang

Abstract:

This paper focuses on the classification task of breast ultrasound images and conducts research on the reliability measurement of classification results. A dual-channel evaluation framework was developed based on the proposed inference reliability and predictive reliability scores. For the inference reliability evaluation, human-aligned and doctor-agreed inference rationals based on the improved feature attribution algorithm SP-RISA are gracefully applied. Uncertainty quantification is used to evaluate the predictive reliability via the test time enhancement. The effectiveness of this reliability evaluation framework has been verified on the breast ultrasound clinical dataset YBUS, and its robustness is verified on the public dataset BUSI. The expected calibration errors on both datasets are significantly lower than traditional evaluation methods, which proves the effectiveness of the proposed reliability measurement.

Keywords: medical imaging, ultrasound imaging, XAI, uncertainty measurement, trustworthy AI

Procedia PDF Downloads 101

Authors: Alejandro Molina-Villegas, Silvia Fernández-Sabido, Eduardo Mendoza-Vargas, Fátima Miranda-Pestaña

Abstract:

The automatic detection of indigenous languages ​​in digital texts is essential to promote their inclusion in digital media. Underrepresented languages, such as Maya, are often excluded from language detection tools like Google’s language-detection library, LANGDETECT. This study addresses these limitations by developing a hybrid language detection solution that accurately distinguishes Maya (YUA) from Spanish (ES). Two strategies are employed: the first focuses on creating a profile for the Maya language within the LANGDETECT library, while the second involves training a Naive Bayes classification model with two categories, YUA and ES. The process includes comprehensive data preprocessing steps, such as cleaning, normalization, tokenization, and n-gram counting, applied to text samples collected from various sources, including articles from La Jornada Maya, a major newspaper in Mexico and the only media outlet that includes a Maya section. After the training phase, a portion of the data is used to create the YUA profile within LANGDETECT, which achieves an accuracy rate above 95% in identifying the Maya language during testing. Additionally, the Naive Bayes classifier, trained and tested on the same database, achieves an accuracy close to 98% in distinguishing between Maya and Spanish, with further validation through F1 score, recall, and logarithmic scoring, without signs of overfitting. This strategy, which combines the LANGDETECT profile with a Naive Bayes model, highlights an adaptable framework that can be extended to other underrepresented languages in future research. This fills a gap in Natural Language Processing and supports the preservation and revitalization of these languages.

Keywords: indigenous languages, language detection, Maya language, Naive Bayes classifier, natural language processing, low-resource languages

Procedia PDF Downloads 16

Authors: Ali Raza, Shafiq Ur Rehman Khan, Raja Sher Afgun Usmani, Asif Raza, Basit Umair

Abstract:

Detection of Fake news and missing information is gaining popularity, especially after the advancement in social media and online news platforms. Social media platforms are the main and speediest source of fake news propagation, whereas online news websites contribute to fake news dissipation. In this study, we propose a framework to detect fake news using the temporal features of text and consider user feedback to identify whether the news is fake or not. In recent studies, the temporal features in text documents gain valuable consideration from Natural Language Processing and user feedback and only try to classify the textual data as fake or true. This research article indicates the impact of recurring and non-recurring events on fake and true news. We use two models BERT and Bi-LSTM to investigate, and it is concluded from BERT we get better results and 70% of true news are recurring and rest of 30% are non-recurring.

Keywords: natural language processing, fake news detection, machine learning, Bi-LSTM

Procedia PDF Downloads 22

Authors: Maomao Cao

Abstract:

Objective: To test the performance of ctDNA methylation for the detection of liver cancer. Methods: A total of 1233 individuals have been recruited in 2017. 15 male and 15 female samples (including 10 cases of liver cancer) were randomly selected in the present study. CfDNA was extracted by MagPure Circulating DNA Maxi Kit. The concentration of cfDNA was obtained by Qubit™ dsDNA HS Assay Kit. A pre-constructed predictive model was used to analyze methylation data and to give a predictive score for each cfDNA sample. Individuals with a predictive score greater than or equal to 80 were classified as having liver cancer. CT tests were considered the gold standard. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for the diagnosis of liver cancer were calculated. Results: 9 patients were diagnosed with liver cancer according to the prediction model (with high sensitivity and threshold of 80 points), with scores of 99.2, 91.9, 96.6, 92.4, 91.3, 92.5, 96.8, 91.1, and 92.2, respectively. The sensitivity, specificity, positive predictive value, and negative predictive value of ctDNA methylation for the diagnosis of liver cancer were 0.70, 0.90, 0.78, and 0.86, respectively. Conclusions: ctDNA methylation could be an acceptable diagnostic modality for the detection of liver cancer.

Keywords: liver cancer, ctDNA methylation, detection, diagnostic performance

Procedia PDF Downloads 151

Authors: Murray L. Ireland, Kevin J. Worrall, Rebecca Mackenzie, Thaleia Flessa, Euan McGookin, Douglas Thomson

Abstract:

Robotic rovers which are designed to work in extra-terrestrial environments present a unique challenge in terms of the reliability and availability of systems throughout the mission. Should some fault occur, with the nearest human potentially millions of kilometres away, detection and identification of the fault must be performed solely by the robot and its subsystems. Faults in the system sensors are relatively straightforward to detect, through the residuals produced by comparison of the system output with that of a simple model. However, faults in the input, that is, the actuators of the system, are harder to detect. A step change in the input signal, caused potentially by the loss of an actuator, can propagate through the system, resulting in complex residuals in multiple outputs. These residuals can be difficult to isolate or distinguish from residuals caused by environmental disturbances. While a more complex fault detection method or additional sensors could be used to solve these issues, an alternative is presented here. Using inverse simulation (InvSim), the inputs and outputs of the mathematical model of the rover system are reversed. Thus, for a desired trajectory, the corresponding actuator inputs are obtained. A step fault near the input then manifests itself as a step change in the residual between the system inputs and the input trajectory obtained through inverse simulation. This approach avoids the need for additional hardware on a mass- and power-critical system such as the rover. The InvSim fault detection method is applied to a simple four-wheeled rover in simulation. Additive system faults and an external disturbance force and are applied to the vehicle in turn, such that the dynamic response and sensor output of the rover are impacted. Basic model-based fault detection is then employed to provide output residuals which may be analysed to provide information on the fault/disturbance. InvSim-based fault detection is then employed, similarly providing input residuals which provide further information on the fault/disturbance. The input residuals are shown to provide clearer information on the location and magnitude of an input fault than the output residuals. Additionally, they can allow faults to be more clearly discriminated from environmental disturbances.

Keywords: fault detection, ground robot, inverse simulation, rover

Procedia PDF Downloads 308

Authors: Ju Peng Poh, Jun Yu Charles Lee, Jonathan Chew Hoe Khoo

Abstract:

The emergence of digital twin technology, a digital replica of physical world, has improved the real-time access to data from sensors about the performance of buildings. This digital transformation has opened up many opportunities to improve the management of the building by using the data collected to help monitor consumption patterns and energy leakages. One example is the integration of predictive models for anomaly detection. In this paper, we use the GAM (Generalised Additive Model) for the anomaly detection of Air Handling Units (AHU) power consumption pattern. There is ample research work on the use of GAM for the prediction of power consumption at the office building and nation-wide level. However, there is limited illustration of its anomaly detection capabilities, prescriptive analytics case study, and its integration with the latest development of digital twin technology. In this paper, we applied the general GAM modelling framework on the historical data of the AHU power consumption and cooling load of the building between Jan 2018 to Aug 2019 from an education campus in Singapore to train prediction models that, in turn, yield predicted values and ranges. The historical data are seamlessly extracted from the digital twin for modelling purposes. We enhanced the utility of the GAM model by using it to power a real-time anomaly detection system based on the forward predicted ranges. The magnitude of deviation from the upper and lower bounds of the uncertainty intervals is used to inform and identify anomalous data points, all based on historical data, without explicit intervention from domain experts. Notwithstanding, the domain expert fits in through an optional feedback loop through which iterative data cleansing is performed. After an anomalously high or low level of power consumption detected, a set of rule-based conditions are evaluated in real-time to help determine the next course of action for the facilities manager. The performance of GAM is then compared with other approaches to evaluate its effectiveness. Lastly, we discuss the successfully deployment of this approach for the detection of anomalous power consumption pattern and illustrated with real-world use cases.

Keywords: anomaly detection, digital twin, generalised additive model, GAM, power consumption, supervised learning

Procedia PDF Downloads 154