Unsupervised Deconvolution Neural Network for High Quality Ultrasound ImagingShujaat Khan
High quality US imaging demand large number of measurements that can increase the cost, size and power requirements. Therefore, low-powered, portable and 3D ultrasound imaging system require reconstruction algorithms that can produce high quality images using fewer receive measurements. Number of model specific methods has been proposed which doesn't work under perturbation. For instance, compressive deconvolution ultrasound which provide a reasonable quality with limited measurements however, it has its own down-sides such as high computation cost and accurate estimation of point spread function (PSF). An other major limitation of conventional methods is that they require RF or base-band signal which is difficult to obtain from portable US systems. To deal with the aforementioned issues, in this study we designed a novel deep deconvolution model for image domain-based deconvolution. The proposed deep deconvolution (DeepDeconv) model can be trained in an unsupervised fashion, alleviate the need of paired high and low quality images. The model was evaluated on both the phantom and in-vivo scans for various sampling configurations. The proposed DeepDeconv significantly enhance the details of anatomical structures and using unsupervised learning on average it achieved 2.14dB, 4.96dB and 0.01 units gain in CR, PSNR and SSIM values respectively, which are comparable to the supervised method.
Deep Learning Based Voice Activity Detection and Speech EnhancementNAVER Engineering
The document summarizes speech recognition front-end technologies including voice activity detection (VAD) and speech enhancement. It discusses conventional signal processing based approaches and more recent deep learning based methods. For VAD, it describes adaptive context attention models that can dynamically adjust the context used based on noise type and SNR. For speech enhancement, it proposes a two-step neural network approach consisting of a prior network that makes multiple predictions from noisy features and a post network that combines these using a boosting method to produce enhanced features, allowing end-to-end training without an explicit masking step. The approach aims to better exploit neural network modeling power while reducing computation cost compared to conventional methods or single-step deep learning frameworks.
CT (computed tomography) scanning uses x-rays and computer processing to create cross-sectional images of the body. Sir Godfrey Hounsfield invented the first CT scanner in 1972. A CT scan uses a narrow x-ray beam that rotates around the body and measures the amount of radiation absorbed in different tissues. A computer processes this data to create images of slices through the body. Each slice is made up of many pixels that are assigned numbers representing the density of the tissue. CT scans provide more detailed images than plain x-rays and can detect many abnormalities.
Aggelos Katsaggelos, Professor and AT&T Chair, Northwestern University, Department of Electrical Engineering & Computer Science (IEEE/ SPIE Fellow, IEEE SPS DL), Sparse and Redundant Representations: Theory and Applications
Computerized tomography (CT) was pioneered by Godfrey Hounsfield and Allan Cormack in the 1970s. CT uses X-rays and computer processing to create cross-sectional images of the body. The first CT scanners used a translate-rotate design, while later generations used multiple detectors and spiral scanning for faster, more detailed imaging. Image reconstruction uses back projection to convert attenuation measurements into pixel values and display slices. CT provides excellent anatomical detail and is widely used for diagnosing conditions of the brain, blood vessels, lungs and other organs.
Variational formulation of unsupervised deep learning for ultrasound image ar...Shujaat Khan
Recently, deep learning approaches have been successfully used for ultrasound (US) image artifact removal. However, paired high-quality images for supervised training are difficult to obtain in many practical situations. Inspired by the recent theory of unsupervised learning using optimal transport driven CycleGAN (OT-CycleGAN), here, we investigate the applicability of unsupervised deep learning for US artifact removal problems without matched reference data. Two types of OT-CycleGAN approaches are employed: one with the partial knowledge of the image degradation physics and the other with the lack of such knowledge. Various US artifact removal problems are then addressed using the two types of OT-CycleGAN. Experimental results for various unsupervised US artifact removal tasks confirmed that our unsupervised learning method delivers results comparable to supervised learning in many practical applications.
Unsupervised Deconvolution Neural Network for High Quality Ultrasound ImagingShujaat Khan
High quality US imaging demand large number of measurements that can increase the cost, size and power requirements. Therefore, low-powered, portable and 3D ultrasound imaging system require reconstruction algorithms that can produce high quality images using fewer receive measurements. Number of model specific methods has been proposed which doesn't work under perturbation. For instance, compressive deconvolution ultrasound which provide a reasonable quality with limited measurements however, it has its own down-sides such as high computation cost and accurate estimation of point spread function (PSF). An other major limitation of conventional methods is that they require RF or base-band signal which is difficult to obtain from portable US systems. To deal with the aforementioned issues, in this study we designed a novel deep deconvolution model for image domain-based deconvolution. The proposed deep deconvolution (DeepDeconv) model can be trained in an unsupervised fashion, alleviate the need of paired high and low quality images. The model was evaluated on both the phantom and in-vivo scans for various sampling configurations. The proposed DeepDeconv significantly enhance the details of anatomical structures and using unsupervised learning on average it achieved 2.14dB, 4.96dB and 0.01 units gain in CR, PSNR and SSIM values respectively, which are comparable to the supervised method.
Deep Learning Based Voice Activity Detection and Speech EnhancementNAVER Engineering
The document summarizes speech recognition front-end technologies including voice activity detection (VAD) and speech enhancement. It discusses conventional signal processing based approaches and more recent deep learning based methods. For VAD, it describes adaptive context attention models that can dynamically adjust the context used based on noise type and SNR. For speech enhancement, it proposes a two-step neural network approach consisting of a prior network that makes multiple predictions from noisy features and a post network that combines these using a boosting method to produce enhanced features, allowing end-to-end training without an explicit masking step. The approach aims to better exploit neural network modeling power while reducing computation cost compared to conventional methods or single-step deep learning frameworks.
CT (computed tomography) scanning uses x-rays and computer processing to create cross-sectional images of the body. Sir Godfrey Hounsfield invented the first CT scanner in 1972. A CT scan uses a narrow x-ray beam that rotates around the body and measures the amount of radiation absorbed in different tissues. A computer processes this data to create images of slices through the body. Each slice is made up of many pixels that are assigned numbers representing the density of the tissue. CT scans provide more detailed images than plain x-rays and can detect many abnormalities.
Aggelos Katsaggelos, Professor and AT&T Chair, Northwestern University, Department of Electrical Engineering & Computer Science (IEEE/ SPIE Fellow, IEEE SPS DL), Sparse and Redundant Representations: Theory and Applications
Computerized tomography (CT) was pioneered by Godfrey Hounsfield and Allan Cormack in the 1970s. CT uses X-rays and computer processing to create cross-sectional images of the body. The first CT scanners used a translate-rotate design, while later generations used multiple detectors and spiral scanning for faster, more detailed imaging. Image reconstruction uses back projection to convert attenuation measurements into pixel values and display slices. CT provides excellent anatomical detail and is widely used for diagnosing conditions of the brain, blood vessels, lungs and other organs.
Variational formulation of unsupervised deep learning for ultrasound image ar...Shujaat Khan
Recently, deep learning approaches have been successfully used for ultrasound (US) image artifact removal. However, paired high-quality images for supervised training are difficult to obtain in many practical situations. Inspired by the recent theory of unsupervised learning using optimal transport driven CycleGAN (OT-CycleGAN), here, we investigate the applicability of unsupervised deep learning for US artifact removal problems without matched reference data. Two types of OT-CycleGAN approaches are employed: one with the partial knowledge of the image degradation physics and the other with the lack of such knowledge. Various US artifact removal problems are then addressed using the two types of OT-CycleGAN. Experimental results for various unsupervised US artifact removal tasks confirmed that our unsupervised learning method delivers results comparable to supervised learning in many practical applications.
Towards fine-precision automated immobilization in maskless radiosurgeryOlalekan Ogunmolu
This document summarizes a presentation on developing an automated patient immobilization system for head and neck cancer radiotherapy without the use of masks. It describes current radiotherapy techniques, issues with positioning errors, and an initial study using a soft robot system with an air bladder and visual feedback to automatically control a mannequin's head motion in one degree of freedom. Nonparametric analysis was used to model the system dynamics and design a PID controller to track position trajectories within 14 seconds, demonstrating a potential solution for accurate in-treatment positioning and motion compensation.
This slide best explains the introduction of CT, basis and types of CT image reconstructions with detailed explanation about Interpolation, convolution, Fourier slice theorem, Fourier transformation and brief explanation about the image domain i.e digital image processing.
The document describes a proposed patient positioning system for maskless head and neck radiotherapy using a soft robot. The system uses a Kinect camera for vision-based sensing of patient head position. A soft robot consisting of an inflatable air bladder and pneumatic valves would manipulate the patient's head to correct for any motion during treatment. Preliminary results show the system was able to control 1 degree of freedom of motion (flexion/extension) of a mannequin head using proportional valve control and Kinect vision feedback to a control system. Further work is needed to validate the system for actual use in radiotherapy treatment.
This document provides an overview of a refresher course on deep learning for CT reconstruction. The course covers an introduction to CNNs, the biological and mathematical origins of deep learning, and various applications of deep learning to CT reconstruction problems like low-dose CT, sparse-view CT, and interior tomography. Successful demonstrations of deep learning approaches for these image reconstruction problems are highlighted from recent works. The role of convolutional and pooling layers in CNNs is discussed. Reasons for why deep learning works well for reconstruction problems are explored, including the mathematical understanding of CNNs in terms of deep convolutional framelets and the lifting of signals to higher dimensional spaces using Hankel matrices. The document concludes by discussing some open problems and providing information on datasets
Currently, magnetic resonance imaging (MRI) has been utilized extensively to obtain high contrast medical image due to its safety which can be applied repetitively. To extract important information from an MRI medical images, an efficient image segmentation or edge detection is required. Edges are represented as important contour features in the medical image since they are the boundaries where distinct intensity changes or discontinuities occur. However, in practices, it is found rather difficult to design an edge detector that is capable of finding all the true edges in an image as there is always noise, and the subjectivity of sensitiveness in detecting the edges. Many traditional algorithms have been proposed to detect the edge, such as Canny, Sobel, Prewitt, Roberts, Zerocross, and Laplacian of Gaussian (LoG). Moreover, many researches have shown the potential of using Artificial Neural Network (ANN) for edge detection. Although many algorithms have been conducted on edge detection for medical images, however higher computational cost and subjective image quality could be further improved. Therefore, the objective of this paper is to develop a fast ANN based edge detection algorithm for MRI medical images. First, we developed features based on horizontal, vertical, and diagonal difference. Then, Canny edge detector will be used as the training output. Finally, optimized parameters will be obtained, including number of hidden layers and output threshold. The edge detection image will be analysed its quality subjectively and computational. Results showed that the proposed algorithm provided better image quality while it has faster processing time around three times time compared to other traditional algorithms, such as Sobel and Canny edge detector.
The document summarizes a research project on multi-resolution data fusion using agent-based sensors. The project aims to develop collaborative signal processing techniques that are energy-aware, fault-tolerant, and progressively improve accuracy. Key accomplishments include developing mobile agent-based collaborative signal processing, energy-aware task scheduling algorithms, analytical battery modeling, and sensor deployment algorithms. The project has resulted in several publications and integrated some techniques successfully, while other integration efforts faced challenges.
Machine learning for Tomographic Imaging.pdfMunir Ahmad
This document discusses using machine learning techniques for tomographic imaging reconstruction and denoising. It begins with an overview of tomographic imaging and PET/CT as an example. It then discusses tomographic data acquisition through PET imaging and sinogram generation. Various analytical and iterative reconstruction methods are described along with their limitations related to noise and ill-posed problems. Neural network approaches for image reconstruction from sinograms, CT image denoising, and mapping iterative reconstruction algorithms to neural networks are proposed to overcome these limitations. Specific network architectures discussed include a simple FBP mapping network, residual learning networks, and networks that unroll iterative algorithms. Applications to PET, SPECT, and developing new techniques like positronium imaging are envisioned.
Machine learning for Tomographic Imaging.pptxMunir Ahmad
This document discusses using machine learning techniques for tomographic imaging reconstruction and denoising. It begins with an overview of tomographic imaging and PET/CT as an example. It then discusses tomographic data acquisition through PET imaging and sinogram generation. Various analytical and iterative reconstruction methods are described along with their limitations related to noise and ill-posed problems. Several studies applying neural networks to different aspects of tomographic reconstruction are summarized, including mapping filtered backprojection to a neural network, using CNNs for low-dose CT denoising, incorporating denoising models within iterative reconstruction networks, and mapping iterative PET reconstruction algorithms to neural networks. Potential applications of deep learning to new modalities like positronium imaging are also mentioned.
1. The document presents a method for super resolution of text images using ant colony optimization. It involves registering multiple low resolution images, fusing them, performing soft classification to assign pixel values to multiple classes, and using ant colony optimization for super resolution mapping to increase the resolution.
2. Key steps include SURF-based image registration, intensity-based and discrete wavelet transform fusion, decision tree-based soft classification, and ant colony optimization to assign pixel values based on pheromone updating to increase resolution.
3. Test cases on images with angular displacement, blurred text, etc. show that the method increases resolution successfully but can add some noise, though processing is faster than alternatives. Ant colony optimization
An artificial neural network was used to accurately identify the interaction positions of gamma photons in a gamma camera detector module. Training datasets were acquired along lines parallel to the x and y axes to simplify the training process and optimize the neural network structure. The proposed method improved discrimination accuracy at the edges of the detector compared to conventional algorithms and reduced the energy resolution from 22.8% to 15.7%, demonstrating its effectiveness for gamma camera systems.
1. The document describes using a deep neural network to detect changes between two SAR images by preclassifying the images, training the neural network on selected samples, and analyzing the results.
2. A similarity matrix and variance matrix are calculated during preclassification to identify and jointly label similar pixels, while different pixels are labeled separately. Good samples are selected to train the neural network.
3. The neural network is tested on images with different types and levels of noise and performs well at change detection, with performance increasing as noise decreases. Future work could focus on accelerating the training process.
Universal plane wave compounding for high quality us imaging using deep learningShujaat Khan
Plane-wave compounding is to sum up several successive plane waves incident at different angles to form an image. By applying time-reversal of the received signals, transmit focusing can be synthesized. Unfortunately, to improve the temporal resolution, the number of plane waves should be reduced, which often degrades the image quality. To address this problem, an image domain learning method using neural networks has been proposed, but the network needs to be retrained when the number of plane waves changes. Herein, we propose, for the first time, a universal plane-wave compounding scheme using deep learning to directly process plane waves and RF data acquired at different view angles and sub-sampling rate to generate high quality US images.
Deep learning for image super resolutionPrudhvi Raj
Using Deep Convolutional Networks, the machine can learn end-to-end mapping between the low/high-resolution images. Unlike traditional methods, this method jointly optimizes all the layers of the image. A light-weight CNN structure is used, which is simple to implement and provides formidable trade-off from the existential methods.
Deep learning for image super resolutionPrudhvi Raj
Using Deep Convolutional Networks, the machine can learn end-to-end mapping between the low/high-resolution images. Unlike traditional methods, this method jointly optimizes all the layers of the image. A light-weight CNN structure is used, which is simple to implement and provides formidable trade-off from the existential methods.
An Efficient Thresholding Neural Network Technique for High Noise Densities E...CSCJournals
Medical images when infected with high noise densities lose usefulness for diagnosis and early detection purposes. Thresholding neural networks (TNN) with a new class of smooth nonlinear function have been widely used to improve the efficiency of the denoising procedure. This paper introduces better solution for medical images in noisy environments which serves in early detection of breast cancer tumor. The proposed algorithm is based on two consecutive phases. Image denoising, where an adaptive learning TNN with remarkable time improvement and good image quality is introduced. A semi-automatic segmentation to extract suspicious regions or regions of interest (ROIs) is presented as an evaluation for the proposed technique. A set of data is then applied to show algorithm superior image quality and complexity reduction especially in high noisy environments.
This paper presents a technique for denoising digital radiographic images using a wavelet-based hidden Markov model. The method first applies the Anscombe transformation to adjust for Poisson noise, then uses the dual-tree complex wavelet transform for decomposition. A hidden Markov tree model is used to capture correlations between wavelet coefficients across scales. Two correction functions are applied to shrink coefficients before inverse transformation. Evaluation on phantom and clinical images showed the method outperforms Gaussian filtering in terms of noise reduction, detail quality and bone sharpness, though some edges had artifacts.
Ijri ece-01-02 image enhancement aided denoising using dual tree complex wave...Ijripublishers Ijri
This paper presents a novel way to reduce noise introduced or exacerbated by image enhancement methods, in particular algorithms based on the random spray sampling technique, but not only. According to the nature of sprays, output images of spray-based methods tend to exhibit noise with unknown statistical distribution. To avoid inappropriate assumptions on the statistical characteristics of noise, a different one is made. In fact, the non-enhanced image is considered to be either free of noise or affected by non-perceivable levels of noise. Taking advantage of the higher sensitivity of the human visual system to changes in brightness, the analysis can be limited to the luma channel of both the non-enhanced and enhanced image. Also, given the importance of directional content in human vision, the analysis is performed through the dual-tree complex wavelet transform , lanczos interpolator and edge preserving smoothing filters. Unlike the discrete wavelet transform, the DTWCT allows for distinction of data directionality in the transform space. For each level of the transform, the standard deviation of the non-enhanced image coefficients is computed across the six orientations of the DTWCT, then it is normalized.
Keywords: dual-tree complex wavelet transform (DTWCT), lanczos interpolator, edge preserving smoothing filters.
The document summarizes research on developing planning and control frameworks for communication-aware coordination of unmanned vehicle networks. It describes using an information-theoretic approach to optimize robot motion to maximize information gain over noisy communication links. Experimental results show decentralized algorithms allow vehicles to form optimal communication chains and relay networks by considering communication constraints. Field experiments demonstrate these approaches can improve tracking performance for heterogeneous teams of unmanned aircraft and vehicles operating in realistic communication environments.
Automatic System for Detection and Classification of Brain TumorsFatma Sayed Ibrahim
Automatic system for brain tumors detection based on DICOM MRI images
Surveying methodologies of from preprocessing to classifications
Implementing comparative study.
Proposed technique with highest accuracy and lest elapsed time.
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...University of Maribor
Slides from talk presenting:
Aleš Zamuda: Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapter and Networking.
Presentation at IcETRAN 2024 session:
"Inter-Society Networking Panel GRSS/MTT-S/CIS
Panel Session: Promoting Connection and Cooperation"
IEEE Slovenia GRSS
IEEE Serbia and Montenegro MTT-S
IEEE Slovenia CIS
11TH INTERNATIONAL CONFERENCE ON ELECTRICAL, ELECTRONIC AND COMPUTING ENGINEERING
3-6 June 2024, Niš, Serbia
Towards fine-precision automated immobilization in maskless radiosurgeryOlalekan Ogunmolu
This document summarizes a presentation on developing an automated patient immobilization system for head and neck cancer radiotherapy without the use of masks. It describes current radiotherapy techniques, issues with positioning errors, and an initial study using a soft robot system with an air bladder and visual feedback to automatically control a mannequin's head motion in one degree of freedom. Nonparametric analysis was used to model the system dynamics and design a PID controller to track position trajectories within 14 seconds, demonstrating a potential solution for accurate in-treatment positioning and motion compensation.
This slide best explains the introduction of CT, basis and types of CT image reconstructions with detailed explanation about Interpolation, convolution, Fourier slice theorem, Fourier transformation and brief explanation about the image domain i.e digital image processing.
The document describes a proposed patient positioning system for maskless head and neck radiotherapy using a soft robot. The system uses a Kinect camera for vision-based sensing of patient head position. A soft robot consisting of an inflatable air bladder and pneumatic valves would manipulate the patient's head to correct for any motion during treatment. Preliminary results show the system was able to control 1 degree of freedom of motion (flexion/extension) of a mannequin head using proportional valve control and Kinect vision feedback to a control system. Further work is needed to validate the system for actual use in radiotherapy treatment.
This document provides an overview of a refresher course on deep learning for CT reconstruction. The course covers an introduction to CNNs, the biological and mathematical origins of deep learning, and various applications of deep learning to CT reconstruction problems like low-dose CT, sparse-view CT, and interior tomography. Successful demonstrations of deep learning approaches for these image reconstruction problems are highlighted from recent works. The role of convolutional and pooling layers in CNNs is discussed. Reasons for why deep learning works well for reconstruction problems are explored, including the mathematical understanding of CNNs in terms of deep convolutional framelets and the lifting of signals to higher dimensional spaces using Hankel matrices. The document concludes by discussing some open problems and providing information on datasets
Currently, magnetic resonance imaging (MRI) has been utilized extensively to obtain high contrast medical image due to its safety which can be applied repetitively. To extract important information from an MRI medical images, an efficient image segmentation or edge detection is required. Edges are represented as important contour features in the medical image since they are the boundaries where distinct intensity changes or discontinuities occur. However, in practices, it is found rather difficult to design an edge detector that is capable of finding all the true edges in an image as there is always noise, and the subjectivity of sensitiveness in detecting the edges. Many traditional algorithms have been proposed to detect the edge, such as Canny, Sobel, Prewitt, Roberts, Zerocross, and Laplacian of Gaussian (LoG). Moreover, many researches have shown the potential of using Artificial Neural Network (ANN) for edge detection. Although many algorithms have been conducted on edge detection for medical images, however higher computational cost and subjective image quality could be further improved. Therefore, the objective of this paper is to develop a fast ANN based edge detection algorithm for MRI medical images. First, we developed features based on horizontal, vertical, and diagonal difference. Then, Canny edge detector will be used as the training output. Finally, optimized parameters will be obtained, including number of hidden layers and output threshold. The edge detection image will be analysed its quality subjectively and computational. Results showed that the proposed algorithm provided better image quality while it has faster processing time around three times time compared to other traditional algorithms, such as Sobel and Canny edge detector.
The document summarizes a research project on multi-resolution data fusion using agent-based sensors. The project aims to develop collaborative signal processing techniques that are energy-aware, fault-tolerant, and progressively improve accuracy. Key accomplishments include developing mobile agent-based collaborative signal processing, energy-aware task scheduling algorithms, analytical battery modeling, and sensor deployment algorithms. The project has resulted in several publications and integrated some techniques successfully, while other integration efforts faced challenges.
Machine learning for Tomographic Imaging.pdfMunir Ahmad
This document discusses using machine learning techniques for tomographic imaging reconstruction and denoising. It begins with an overview of tomographic imaging and PET/CT as an example. It then discusses tomographic data acquisition through PET imaging and sinogram generation. Various analytical and iterative reconstruction methods are described along with their limitations related to noise and ill-posed problems. Neural network approaches for image reconstruction from sinograms, CT image denoising, and mapping iterative reconstruction algorithms to neural networks are proposed to overcome these limitations. Specific network architectures discussed include a simple FBP mapping network, residual learning networks, and networks that unroll iterative algorithms. Applications to PET, SPECT, and developing new techniques like positronium imaging are envisioned.
Machine learning for Tomographic Imaging.pptxMunir Ahmad
This document discusses using machine learning techniques for tomographic imaging reconstruction and denoising. It begins with an overview of tomographic imaging and PET/CT as an example. It then discusses tomographic data acquisition through PET imaging and sinogram generation. Various analytical and iterative reconstruction methods are described along with their limitations related to noise and ill-posed problems. Several studies applying neural networks to different aspects of tomographic reconstruction are summarized, including mapping filtered backprojection to a neural network, using CNNs for low-dose CT denoising, incorporating denoising models within iterative reconstruction networks, and mapping iterative PET reconstruction algorithms to neural networks. Potential applications of deep learning to new modalities like positronium imaging are also mentioned.
1. The document presents a method for super resolution of text images using ant colony optimization. It involves registering multiple low resolution images, fusing them, performing soft classification to assign pixel values to multiple classes, and using ant colony optimization for super resolution mapping to increase the resolution.
2. Key steps include SURF-based image registration, intensity-based and discrete wavelet transform fusion, decision tree-based soft classification, and ant colony optimization to assign pixel values based on pheromone updating to increase resolution.
3. Test cases on images with angular displacement, blurred text, etc. show that the method increases resolution successfully but can add some noise, though processing is faster than alternatives. Ant colony optimization
An artificial neural network was used to accurately identify the interaction positions of gamma photons in a gamma camera detector module. Training datasets were acquired along lines parallel to the x and y axes to simplify the training process and optimize the neural network structure. The proposed method improved discrimination accuracy at the edges of the detector compared to conventional algorithms and reduced the energy resolution from 22.8% to 15.7%, demonstrating its effectiveness for gamma camera systems.
1. The document describes using a deep neural network to detect changes between two SAR images by preclassifying the images, training the neural network on selected samples, and analyzing the results.
2. A similarity matrix and variance matrix are calculated during preclassification to identify and jointly label similar pixels, while different pixels are labeled separately. Good samples are selected to train the neural network.
3. The neural network is tested on images with different types and levels of noise and performs well at change detection, with performance increasing as noise decreases. Future work could focus on accelerating the training process.
Universal plane wave compounding for high quality us imaging using deep learningShujaat Khan
Plane-wave compounding is to sum up several successive plane waves incident at different angles to form an image. By applying time-reversal of the received signals, transmit focusing can be synthesized. Unfortunately, to improve the temporal resolution, the number of plane waves should be reduced, which often degrades the image quality. To address this problem, an image domain learning method using neural networks has been proposed, but the network needs to be retrained when the number of plane waves changes. Herein, we propose, for the first time, a universal plane-wave compounding scheme using deep learning to directly process plane waves and RF data acquired at different view angles and sub-sampling rate to generate high quality US images.
Deep learning for image super resolutionPrudhvi Raj
Using Deep Convolutional Networks, the machine can learn end-to-end mapping between the low/high-resolution images. Unlike traditional methods, this method jointly optimizes all the layers of the image. A light-weight CNN structure is used, which is simple to implement and provides formidable trade-off from the existential methods.
Deep learning for image super resolutionPrudhvi Raj
Using Deep Convolutional Networks, the machine can learn end-to-end mapping between the low/high-resolution images. Unlike traditional methods, this method jointly optimizes all the layers of the image. A light-weight CNN structure is used, which is simple to implement and provides formidable trade-off from the existential methods.
An Efficient Thresholding Neural Network Technique for High Noise Densities E...CSCJournals
Medical images when infected with high noise densities lose usefulness for diagnosis and early detection purposes. Thresholding neural networks (TNN) with a new class of smooth nonlinear function have been widely used to improve the efficiency of the denoising procedure. This paper introduces better solution for medical images in noisy environments which serves in early detection of breast cancer tumor. The proposed algorithm is based on two consecutive phases. Image denoising, where an adaptive learning TNN with remarkable time improvement and good image quality is introduced. A semi-automatic segmentation to extract suspicious regions or regions of interest (ROIs) is presented as an evaluation for the proposed technique. A set of data is then applied to show algorithm superior image quality and complexity reduction especially in high noisy environments.
This paper presents a technique for denoising digital radiographic images using a wavelet-based hidden Markov model. The method first applies the Anscombe transformation to adjust for Poisson noise, then uses the dual-tree complex wavelet transform for decomposition. A hidden Markov tree model is used to capture correlations between wavelet coefficients across scales. Two correction functions are applied to shrink coefficients before inverse transformation. Evaluation on phantom and clinical images showed the method outperforms Gaussian filtering in terms of noise reduction, detail quality and bone sharpness, though some edges had artifacts.
Ijri ece-01-02 image enhancement aided denoising using dual tree complex wave...Ijripublishers Ijri
This paper presents a novel way to reduce noise introduced or exacerbated by image enhancement methods, in particular algorithms based on the random spray sampling technique, but not only. According to the nature of sprays, output images of spray-based methods tend to exhibit noise with unknown statistical distribution. To avoid inappropriate assumptions on the statistical characteristics of noise, a different one is made. In fact, the non-enhanced image is considered to be either free of noise or affected by non-perceivable levels of noise. Taking advantage of the higher sensitivity of the human visual system to changes in brightness, the analysis can be limited to the luma channel of both the non-enhanced and enhanced image. Also, given the importance of directional content in human vision, the analysis is performed through the dual-tree complex wavelet transform , lanczos interpolator and edge preserving smoothing filters. Unlike the discrete wavelet transform, the DTWCT allows for distinction of data directionality in the transform space. For each level of the transform, the standard deviation of the non-enhanced image coefficients is computed across the six orientations of the DTWCT, then it is normalized.
Keywords: dual-tree complex wavelet transform (DTWCT), lanczos interpolator, edge preserving smoothing filters.
The document summarizes research on developing planning and control frameworks for communication-aware coordination of unmanned vehicle networks. It describes using an information-theoretic approach to optimize robot motion to maximize information gain over noisy communication links. Experimental results show decentralized algorithms allow vehicles to form optimal communication chains and relay networks by considering communication constraints. Field experiments demonstrate these approaches can improve tracking performance for heterogeneous teams of unmanned aircraft and vehicles operating in realistic communication environments.
Automatic System for Detection and Classification of Brain TumorsFatma Sayed Ibrahim
Automatic system for brain tumors detection based on DICOM MRI images
Surveying methodologies of from preprocessing to classifications
Implementing comparative study.
Proposed technique with highest accuracy and lest elapsed time.
Similar to MULTI-DOMAIN UNPAIRED ULTRASOUND IMAGE ARTIFACT REMOVAL USING A SINGLE CONVOLUTIONAL NEURAL NETWORK.pptx (20)
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...University of Maribor
Slides from talk presenting:
Aleš Zamuda: Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapter and Networking.
Presentation at IcETRAN 2024 session:
"Inter-Society Networking Panel GRSS/MTT-S/CIS
Panel Session: Promoting Connection and Cooperation"
IEEE Slovenia GRSS
IEEE Serbia and Montenegro MTT-S
IEEE Slovenia CIS
11TH INTERNATIONAL CONFERENCE ON ELECTRICAL, ELECTRONIC AND COMPUTING ENGINEERING
3-6 June 2024, Niš, Serbia
Embedded machine learning-based road conditions and driving behavior monitoringIJECEIAES
Car accident rates have increased in recent years, resulting in losses in human lives, properties, and other financial costs. An embedded machine learning-based system is developed to address this critical issue. The system can monitor road conditions, detect driving patterns, and identify aggressive driving behaviors. The system is based on neural networks trained on a comprehensive dataset of driving events, driving styles, and road conditions. The system effectively detects potential risks and helps mitigate the frequency and impact of accidents. The primary goal is to ensure the safety of drivers and vehicles. Collecting data involved gathering information on three key road events: normal street and normal drive, speed bumps, circular yellow speed bumps, and three aggressive driving actions: sudden start, sudden stop, and sudden entry. The gathered data is processed and analyzed using a machine learning system designed for limited power and memory devices. The developed system resulted in 91.9% accuracy, 93.6% precision, and 92% recall. The achieved inference time on an Arduino Nano 33 BLE Sense with a 32-bit CPU running at 64 MHz is 34 ms and requires 2.6 kB peak RAM and 139.9 kB program flash memory, making it suitable for resource-constrained embedded systems.
A SYSTEMATIC RISK ASSESSMENT APPROACH FOR SECURING THE SMART IRRIGATION SYSTEMSIJNSA Journal
The smart irrigation system represents an innovative approach to optimize water usage in agricultural and landscaping practices. The integration of cutting-edge technologies, including sensors, actuators, and data analysis, empowers this system to provide accurate monitoring and control of irrigation processes by leveraging real-time environmental conditions. The main objective of a smart irrigation system is to optimize water efficiency, minimize expenses, and foster the adoption of sustainable water management methods. This paper conducts a systematic risk assessment by exploring the key components/assets and their functionalities in the smart irrigation system. The crucial role of sensors in gathering data on soil moisture, weather patterns, and plant well-being is emphasized in this system. These sensors enable intelligent decision-making in irrigation scheduling and water distribution, leading to enhanced water efficiency and sustainable water management practices. Actuators enable automated control of irrigation devices, ensuring precise and targeted water delivery to plants. Additionally, the paper addresses the potential threat and vulnerabilities associated with smart irrigation systems. It discusses limitations of the system, such as power constraints and computational capabilities, and calculates the potential security risks. The paper suggests possible risk treatment methods for effective secure system operation. In conclusion, the paper emphasizes the significant benefits of implementing smart irrigation systems, including improved water conservation, increased crop yield, and reduced environmental impact. Additionally, based on the security analysis conducted, the paper recommends the implementation of countermeasures and security approaches to address vulnerabilities and ensure the integrity and reliability of the system. By incorporating these measures, smart irrigation technology can revolutionize water management practices in agriculture, promoting sustainability, resource efficiency, and safeguarding against potential security threats.
Literature Review Basics and Understanding Reference Management.pptxDr Ramhari Poudyal
Three-day training on academic research focuses on analytical tools at United Technical College, supported by the University Grant Commission, Nepal. 24-26 May 2024
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTjpsjournal1
The rivalry between prominent international actors for dominance over Central Asia's hydrocarbon
reserves and the ancient silk trade route, along with China's diplomatic endeavours in the area, has been
referred to as the "New Great Game." This research centres on the power struggle, considering
geopolitical, geostrategic, and geoeconomic variables. Topics including trade, political hegemony, oil
politics, and conventional and nontraditional security are all explored and explained by the researcher.
Using Mackinder's Heartland, Spykman Rimland, and Hegemonic Stability theories, examines China's role
in Central Asia. This study adheres to the empirical epistemological method and has taken care of
objectivity. This study analyze primary and secondary research documents critically to elaborate role of
china’s geo economic outreach in central Asian countries and its future prospect. China is thriving in trade,
pipeline politics, and winning states, according to this study, thanks to important instruments like the
Shanghai Cooperation Organisation and the Belt and Road Economic Initiative. According to this study,
China is seeing significant success in commerce, pipeline politics, and gaining influence on other
governments. This success may be attributed to the effective utilisation of key tools such as the Shanghai
Cooperation Organisation and the Belt and Road Economic Initiative.
Introduction- e - waste – definition - sources of e-waste– hazardous substances in e-waste - effects of e-waste on environment and human health- need for e-waste management– e-waste handling rules - waste minimization techniques for managing e-waste – recycling of e-waste - disposal treatment methods of e- waste – mechanism of extraction of precious metal from leaching solution-global Scenario of E-waste – E-waste in India- case studies.
Understanding Inductive Bias in Machine LearningSUTEJAS
This presentation explores the concept of inductive bias in machine learning. It explains how algorithms come with built-in assumptions and preferences that guide the learning process. You'll learn about the different types of inductive bias and how they can impact the performance and generalizability of machine learning models.
The presentation also covers the positive and negative aspects of inductive bias, along with strategies for mitigating potential drawbacks. We'll explore examples of how bias manifests in algorithms like neural networks and decision trees.
By understanding inductive bias, you can gain valuable insights into how machine learning models work and make informed decisions when building and deploying them.
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODELgerogepatton
As digital technology becomes more deeply embedded in power systems, protecting the communication
networks of Smart Grids (SG) has emerged as a critical concern. Distributed Network Protocol 3 (DNP3)
represents a multi-tiered application layer protocol extensively utilized in Supervisory Control and Data
Acquisition (SCADA)-based smart grids to facilitate real-time data gathering and control functionalities.
Robust Intrusion Detection Systems (IDS) are necessary for early threat detection and mitigation because
of the interconnection of these networks, which makes them vulnerable to a variety of cyberattacks. To
solve this issue, this paper develops a hybrid Deep Learning (DL) model specifically designed for intrusion
detection in smart grids. The proposed approach is a combination of the Convolutional Neural Network
(CNN) and the Long-Short-Term Memory algorithms (LSTM). We employed a recent intrusion detection
dataset (DNP3), which focuses on unauthorized commands and Denial of Service (DoS) cyberattacks, to
train and test our model. The results of our experiments show that our CNN-LSTM method is much better
at finding smart grid intrusions than other deep learning algorithms used for classification. In addition,
our proposed approach improves accuracy, precision, recall, and F1 score, achieving a high detection
accuracy rate of 99.50%.
MULTI-DOMAIN UNPAIRED ULTRASOUND IMAGE ARTIFACT REMOVAL USING A SINGLE CONVOLUTIONAL NEURAL NETWORK.pptx
1. Jaeyoung Huh1,
Shujaat Khan1, and Jong Chul Ye2
BISPL - BioImaging, Signal Processing, and Learning lab.
1 Dept. Bio & Brain Engineering
2 Kim Jaechul Graduate School of AI
KAIST, Korea
Multi-Domain Unpaired Ultrasound Image
Artifact Removal Using
a Single Convolutional Neural Network
2. Introduction
Ultrasound image artifact
- Blurring artifact from limitation by the
bandwidth of the transducer.
"Unsupervised deconvolution neural network for high quality ultrasound imaging." International Ultrasonics Symposium (IUS). IEEE, 2020.
No interference Constructive
interference
Destructive
interference
Signal
1
Signal
2
Signal
summation
Deconvolution Despeckle
- Speckle pattern generated by
instructive and destructive signal
interference.
3. Motivation
Conventional Method
US artifact hinder accurate diagnosis.
Many researchers have proposed
various model-based iterative
algorithms
computationally expensive
"Increasing axial resolution of ultrasonic imaging with a joint sparse representation model." TUFFC, 2016.
"A non-local low-rank framework for ultrasound speckle reduction." CVPR, 2017.
4. Motivation
Deep learning based single domain image to
image translation such as CycleGAN.
Still have some technical hurdles to wide
acceptance.
Different types of artifact needs distinct choice of
artifact suppression algorithms.
Single domain translation model
5. Motivation
Recently, multi-domain image translation
models are proposed such as StarGAN.
StarGAN require bi-directional translation
which is unnecessary in US image.
Multi-domain translation model
6. Background
Representative single-domain unpaired image-to-image translation method.
It minimizes the distance between 𝜇, 𝜇𝜃 and 𝜈, 𝜈𝜙.
Geometry of CycleGAN CycleGAN structure
CycleGAN
"Unpaired image-to-image translation using cycle-consistent adversarial networks.“ CVPR, 2017.
7. Background
StarGAN
"Stargan: Unified generative adversarial networks for multi-domain image-to-image translation.“ CVPR, 2018.
StarGAN structure
Representative multi-domain unpaired image-to-image translation method.
It can translate into each domain by using single generator.
Each domain is separated by mask vector composed of one-hot vector.
8. Strategy
Multi-domain Ultrasound image artifact removal using a single CNN
Delay-And-Sum (DAS) image is generated from
raw data obtained from the scanner.
=> not necessary to re-generate from other
domain image.
By combining the concept of CycleGAN and
StarGAN, we proposed uni-directional multi-
domain translation method for US image artifact
removal.
10. Our contribution
Generator, Discriminator Structure
The generator is U-Net with residual block.
The discriminator is multi-head structure.
3x3 Convolution with stride=1 + InstNorm + Relu
3x3 Convolution with stride=2 + InstNorm + Relu
3x3 Convolution with stride=1 + InstNorm
3x3 Transposed convolution with stride=2 + InstNorm + Relu
3x3 Convolution with stride=1 + Tanh
4x4 Convolution with stride=2 + LeakyRelu
3x3 Convolution with stride=1
4x4 Convolution with stride=1
Domain
classification
Real/Fake
classification
11. Training Details
Implementation Details
Total Epoch 1000
Learning Rate
Linear decreasing from 1e-4 after the
half of total epoch
Optimizer Adam Optimizer
Batch size 4
Parameter (𝜆𝑐𝑦𝑐, 𝜆𝐺𝑃, 𝜆𝑐𝑙𝑠)= (20,30,1)
Optimization
Formulation
WGAN with Gradient-Penalty
Dataset Details
Training set
Input - 304 images (8 subjects)
Input -125 images (Phantom)
Target - 429 images
(generated using input image)
Test set
80 images (2 subjects)
16 images (tissue mimicking phantom)
Data
Augmentation
Flipping, Rotating, Random Scaling
Normalization Normalized all images to -1~1
Target
generation
- Deconvolution
"Increasing axial resolution of ultrasonic imaging
with a joint sparse representation model." TUFFC,
2016.
- Despeckle
"A non-local low-rank framework for ultrasound
speckle reduction." CVPR, 2017.
14. Conclusion
Ultrasound (US) image suffer from many artifacts from various sources.
We proposed multi-domain US image artifact removal method using single convolution
neural network.
The method is based on representative image-to-image (I2I) translation algorithm,
CycleGAN and StarGAN.
A single network can provide blurring removed or speckle suppressed image.