Darren Dancey
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Darren Dancey has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Raheel Nawaz
Staffordshire University, Stoke-on-Trent, United Kingdom
5 shared publications - Lianghao Han
School of Medicine, Tongji University, Shanghai 200092, China
4 shared publications - Liangxiu Han
Manchester Metropolitan University
4 shared publications - Daoqiang Zhang
Nanjing University of Aeronautics and Astronautics
2 shared publications - Sheng Chang
Department of Space Microwave Remote Sensing System, Aerospace Information Research Institute, Beijing 100190, China
2 shared publications - Xin Zhang
Department of Computing and Mathematics, Manchester Metropolitan University, Manchester M1 5GD, UK
2 shared publications
Latest Publications
- Open Access
- 2 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 108603-108616
sMRI-PatchNet: A Novel Efficient Explainable Patch-Based Deep Learning Network for Alzheimer’s Disease Diagnosis With Structural MRI
Structural magnetic resonance imaging (sMRI) can identify subtle brain changes due to its high contrast for soft tissues and high spatial resolution. It has been widely used in diagnosing neurological brain diseases, such as Alzheimer’s disease (AD). However, the size of 3D high-resolution data poses a significant challenge for data analysis and processing. Since only a few areas of the brain show structural changes highly associated with AD, the patch-based methods dividing the whole data into several regular patches have shown promising for more efficient image analysis. The major challenges of the patch-based methods include identifying the discriminative patches, combining features from the discrete discriminative patches, and designing appropriate classifiers. This work proposes a novel efficient patch-based deep learning network (sMRI-PatchNet) with explainable patch localisation and selection for AD diagnosis. Specifically, it consists of two primary components: 1) A fast and efficient explainable patch selection method for determining the most discriminative patches; and 2) A novel patch-based network for extracting deep features and AD classification with position embeddings to retain position information, capable of capturing the global and local information of inter- and intra-patches. This method has been applied for the AD classification and the prediction of the transitional state moderate cognitive impairment (MCI) conversion with real datasets. The experimental evaluation shows that the proposed method can identify discriminative pathological locations effectively with a significant reduction on patch numbers used, providing better performance in terms of accuracy, computing performance, and generalizability, in contrast to the state-of-the-art methods.
- 2 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 108603-108616
Open AccesssMRI-PatchNet: A Novel Efficient Explainable Patch-Based Deep Learning Network for Alzheimer’s Disease Diagnosis With Structural MRI
Structural magnetic resonance imaging (sMRI) can identify subtle brain changes due to its high contrast for soft tissues and high spatial resolution. It has been widely used in diagnosing neurological brain diseases, such as Alzheimer’s disease (AD). However, the size of 3D high-resolution data poses a significant challenge for data analysis and processing. Since only a few areas of the brain show structural changes highly associated with AD, the patch-based methods dividing the whole data into several regular patches have shown promising for more efficient image analysis. The major challenges of the patch-based methods include identifying the discriminative patches, combining features from the discrete discriminative patches, and designing appropriate classifiers. This work proposes a novel efficient patch-based deep learning network (sMRI-PatchNet) with explainable patch localisation and selection for AD diagnosis. Specifically, it consists of two primary components: 1) A fast and efficient explainable patch selection method for determining the most discriminative patches; and 2) A novel patch-based network for extracting deep features and AD classification with position embeddings to retain position information, capable of capturing the global and local information of inter- and intra-patches. This method has been applied for the AD classification and the prediction of the transitional state moderate cognitive impairment (MCI) conversion with real datasets. The experimental evaluation shows that the proposed method can identify discriminative pathological locations effectively with a significant reduction on patch numbers used, providing better performance in terms of accuracy, computing performance, and generalizability, in contrast to the state-of-the-art methods.
- Open Access
sMRI-PatchNet: A novel explainable patch-based deep learning network for Alzheimer's disease diagnosis and discriminative atrophy localisation with Structural MRI
Structural magnetic resonance imaging (sMRI) can identify subtle brain changes due to its high contrast for soft tissues and high spatial resolution. It has been widely used in diagnosing neurological brain diseases, such as Alzheimer disease (AD). However, the size of 3D high-resolution data poses a significant challenge for data analysis and processing. Since only a few areas of the brain show structural changes highly associated with AD, the patch-based methods dividing the whole image data into several small regular patches have shown promising for more efficient sMRI-based image analysis. The major challenges of the patch-based methods on sMRI include identifying the discriminative patches, combining features from the discrete discriminative patches, and designing appropriate classifiers. This work proposes a novel patch-based deep learning network (sMRI-PatchNet) with explainable patch localisation and selection for AD diagnosis using sMRI. Specifically, it consists of two primary components: 1) A fast and efficient explainable patch selection mechanism for determining the most discriminative patches based on computing the SHapley Additive exPlanations (SHAP) contribution to a transfer learning model for AD diagnosis on massive medical data; and 2) A novel patch-based network for extracting deep features and AD classfication from the selected patches with position embeddings to retain position information, capable of capturing the global and local information of inter- and intra-patches. This method has been applied for the AD classification and the prediction of the transitional state moderate cognitive impairment (MCI) conversion with real datasets.
Open AccesssMRI-PatchNet: A novel explainable patch-based deep learning network for Alzheimer's disease diagnosis and discriminative atrophy localisation with Structural MRI
Structural magnetic resonance imaging (sMRI) can identify subtle brain changes due to its high contrast for soft tissues and high spatial resolution. It has been widely used in diagnosing neurological brain diseases, such as Alzheimer disease (AD). However, the size of 3D high-resolution data poses a significant challenge for data analysis and processing. Since only a few areas of the brain show structural changes highly associated with AD, the patch-based methods dividing the whole image data into several small regular patches have shown promising for more efficient sMRI-based image analysis. The major challenges of the patch-based methods on sMRI include identifying the discriminative patches, combining features from the discrete discriminative patches, and designing appropriate classifiers. This work proposes a novel patch-based deep learning network (sMRI-PatchNet) with explainable patch localisation and selection for AD diagnosis using sMRI. Specifically, it consists of two primary components: 1) A fast and efficient explainable patch selection mechanism for determining the most discriminative patches based on computing the SHapley Additive exPlanations (SHAP) contribution to a transfer learning model for AD diagnosis on massive medical data; and 2) A novel patch-based network for extracting deep features and AD classfication from the selected patches with position embeddings to retain position information, capable of capturing the global and local information of inter- and intra-patches. This method has been applied for the AD classification and the prediction of the transitional state moderate cognitive impairment (MCI) conversion with real datasets.
- 25 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 60  (01962892) , 1-19
A Latent Encoder Coupled Generative Adversarial Network (LE-GAN) for Efficient Hyperspectral Image Super-Resolution
Realistic hyperspectral image (HSI) super-resolution (SR) techniques aim to generate a high-resolution (HR) HSI with higher spectral and spatial fidelity from its low-resolution (LR) counterpart. The generative adversarial network (GAN) has proven to be an effective deep learning framework for image SR. However, the optimization process of existing GAN-based models frequently suffers from the problem of mode collapse, leading to the limited capacity of spectral–spatial invariant reconstruction. This may cause the spectral–spatial distortion to the generated HSI, especially with a large upscaling factor. To alleviate the problem of mode collapse, this work has proposed a novel GAN model coupled with a latent encoder (LE-GAN), which can map the generated spectral–spatial features from the image space to the latent space and produce a coupling component to regularize the generated samples. Essentially, we treat an HSI as a high-dimensional manifold embedded in a latent space. Thus, the optimization of GAN models is converted to the problem of learning the distributions of HR HSI samples in the latent space, making the distributions of the generated SR HSIs closer to those of their original HR counterparts. We have conducted experimental evaluations on the model performance of SR and its capability in alleviating mode collapse. The proposed approach has been tested and validated based on two real HSI datasets with different sensors (i.e., AVIRIS and UHD-185) for various upscaling factors (i.e., $\times 2$ , $\times 4$ , and $\times 8$ ) and added noise levels (i.e., $\infty $ , 40, and 80 dB) and compared with the state-of-the-art SR models (i.e., hyperspectral coupled network (HyCoNet), low tensor-train rank (LTTR), band attention GAN (BAGAN), SR-GAN, and WGAN). Experimental results show that the proposed model outperforms the competitors on the SR quality, robustness, and alleviation of mode collapse. The proposed approach is able to capture spectral and spatial details and generate more faithful samples than its competitors. It has also been found that the proposed model is more robust to noise and less sensitive to the upscaling factor and has been proven to be effective in improving the convergence of the generator and the spectral–spatial fidelity of the SR HSIs.
- 25 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 60  (01962892) , 1-19
A Latent Encoder Coupled Generative Adversarial Network (LE-GAN) for Efficient Hyperspectral Image Super-Resolution
Realistic hyperspectral image (HSI) super-resolution (SR) techniques aim to generate a high-resolution (HR) HSI with higher spectral and spatial fidelity from its low-resolution (LR) counterpart. The generative adversarial network (GAN) has proven to be an effective deep learning framework for image SR. However, the optimization process of existing GAN-based models frequently suffers from the problem of mode collapse, leading to the limited capacity of spectral–spatial invariant reconstruction. This may cause the spectral–spatial distortion to the generated HSI, especially with a large upscaling factor. To alleviate the problem of mode collapse, this work has proposed a novel GAN model coupled with a latent encoder (LE-GAN), which can map the generated spectral–spatial features from the image space to the latent space and produce a coupling component to regularize the generated samples. Essentially, we treat an HSI as a high-dimensional manifold embedded in a latent space. Thus, the optimization of GAN models is converted to the problem of learning the distributions of HR HSI samples in the latent space, making the distributions of the generated SR HSIs closer to those of their original HR counterparts. We have conducted experimental evaluations on the model performance of SR and its capability in alleviating mode collapse. The proposed approach has been tested and validated based on two real HSI datasets with different sensors (i.e., AVIRIS and UHD-185) for various upscaling factors (i.e., $\times 2$ , $\times 4$ , and $\times 8$ ) and added noise levels (i.e., $\infty $ , 40, and 80 dB) and compared with the state-of-the-art SR models (i.e., hyperspectral coupled network (HyCoNet), low tensor-train rank (LTTR), band attention GAN (BAGAN), SR-GAN, and WGAN). Experimental results show that the proposed model outperforms the competitors on the SR quality, robustness, and alleviation of mode collapse. The proposed approach is able to capture spectral and spatial details and generate more faithful samples than its competitors. It has also been found that the proposed model is more robust to noise and less sensitive to the upscaling factor and has been proven to be effective in improving the convergence of the generator and the spectral–spatial fidelity of the SR HSIs.
- Open Access
- 24 January 2022
- Published by MDPI AG in Electronics
- Vol. 11  (3) , 350
SCNN-Attack: A Side-Channel Attack to Identify YouTube Videos in a VPN and Non-VPN Network Traffic
Encryption Protocols e.g., HTTPS is utilized to secure the traffic between servers and clients for YouTube and other video streaming services, and to further secure the communication, VPNs are used. However, these protocols are not sufficient to hide the identity of the videos from someone who can sniff the network traffic. The present work explores the methodologies and features to identify the videos in a VPN and non-VPN network traffic. To identify such videos, a side-channel attack using a Sequential Convolution Neural Network is proposed. The results demonstrate that a sequence of bytes per second from even one-minute sniffing of network traffic is sufficient to predict the video with high accuracy. The accuracy is increased to 90% accuracy in the non-VPN, 66% accuracy in the VPN, and 77% in the mixed VPN and non-VPN traffic, for models with two-minute sniffing.
- 24 January 2022
- Published by MDPI AG in Electronics
- Vol. 11  (3) , 350
Open AccessSCNN-Attack: A Side-Channel Attack to Identify YouTube Videos in a VPN and Non-VPN Network Traffic
Encryption Protocols e.g., HTTPS is utilized to secure the traffic between servers and clients for YouTube and other video streaming services, and to further secure the communication, VPNs are used. However, these protocols are not sufficient to hide the identity of the videos from someone who can sniff the network traffic. The present work explores the methodologies and features to identify the videos in a VPN and non-VPN network traffic. To identify such videos, a side-channel attack using a Sequential Convolution Neural Network is proposed. The results demonstrate that a sequence of bytes per second from even one-minute sniffing of network traffic is sufficient to predict the video with high accuracy. The accuracy is increased to 90% accuracy in the non-VPN, 66% accuracy in the VPN, and 77% in the mixed VPN and non-VPN traffic, for models with two-minute sniffing.
- Open Access
- 24 February 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 60  (01962892) , 1-20
A Biologically Interpretable Two-Stage Deep Neural Network (BIT-DNN) for Vegetation Recognition From Hyperspectral Imagery
Spectral-spatial-based deep learning models have recently proven to be effective in hyper-spectral image (HSI) classification for various earth monitoring applications such as land cover classification and agricultural monitoring. However, due to the nature of ``black-box'' model representation, how to explain and interpret the learning process and the model decision, especially for vegetation classification, remains an open challenge. This study proposes a novel interpretable deep learning model--a biologically interpretable two-stage deep neural network (BIT-DNN), by incorporating the prior-knowledge (i.e., biophysical and biochemical attributes and their hierarchical structures of target entities)-based spectral-spatial feature transformation into the proposed framework, capable of achieving both high accuracy and interpretability on HSI-based classification tasks. The proposed model introduces a two-stage feature learning process: in the first stage, an enhanced interpretable feature block extracts the low-level spectral features associated with the biophysical and biochemical attributes of target entities; and in the second stage, an interpretable capsule block extracts and encapsulates the high-level joint spectral-spatial features representing the hierarchical structure of biophysical and biochemical attributes of these target entities, which provides the model an improved performance on classification and intrinsic interpretability with reduced computational complexity. We have tested and evaluated the model using four real HSI data sets for four separate tasks (i.e., plant species classification, land cover classification, urban scene recognition, and crop disease recognition tasks). The proposed model has been compared with five state-of-the-art deep learning models. The results demonstrate that the proposed model has competitive advantages in terms of both classification accuracy and model interpretability, especially for vegetation classification.
- 24 February 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 60  (01962892) , 1-20
Open AccessA Biologically Interpretable Two-Stage Deep Neural Network (BIT-DNN) for Vegetation Recognition From Hyperspectral Imagery
Spectral-spatial-based deep learning models have recently proven to be effective in hyper-spectral image (HSI) classification for various earth monitoring applications such as land cover classification and agricultural monitoring. However, due to the nature of ``black-box'' model representation, how to explain and interpret the learning process and the model decision, especially for vegetation classification, remains an open challenge. This study proposes a novel interpretable deep learning model--a biologically interpretable two-stage deep neural network (BIT-DNN), by incorporating the prior-knowledge (i.e., biophysical and biochemical attributes and their hierarchical structures of target entities)-based spectral-spatial feature transformation into the proposed framework, capable of achieving both high accuracy and interpretability on HSI-based classification tasks. The proposed model introduces a two-stage feature learning process: in the first stage, an enhanced interpretable feature block extracts the low-level spectral features associated with the biophysical and biochemical attributes of target entities; and in the second stage, an interpretable capsule block extracts and encapsulates the high-level joint spectral-spatial features representing the hierarchical structure of biophysical and biochemical attributes of these target entities, which provides the model an improved performance on classification and intrinsic interpretability with reduced computational complexity. We have tested and evaluated the model using four real HSI data sets for four separate tasks (i.e., plant species classification, land cover classification, urban scene recognition, and crop disease recognition tasks). The proposed model has been compared with five state-of-the-art deep learning models. The results demonstrate that the proposed model has competitive advantages in terms of both classification accuracy and model interpretability, especially for vegetation classification.
- Open Access
- 10 November 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 208240-208252
Proximal Actuation of an Elastically Loaded Scissors Mechanism for the Leg Design of a Quadruped Robot
Spring Loaded Pantographs (SLPs) are frequently used in designing lightweight limbs for multi-legged robots. Quadruped robots that incorporate cable-pulled SLP legs have proven to be agile, robust and capable of conserving energy during their gait cycle. In such designs, the extension of the distal segments via the knee joint is dependent upon the length of the cable. In this article we propose the use of an Elastically Loaded Scissors Mechanism (ELS Mechanism or ELSM), which is a variant of the SLP. Driven by ’pulling’ onto the proximal joint of the scissors as opposed to the distal joint, this proposed leg utilizes the increased mechanical advantage of the scissors mechanism to ’amplify’ input angles to larger output displacement by the knee joint. Analysis and Simulations reveal that the proposed mechanism achieves increased motion speed as compared to the SLP mechanism. This, however, comes at the cost of higher load on the actuator which serves as an engineering trade-off. This is validated by experimentation using motion capture and load motor techniques of the SLP and ELS configurations in a physical quadruped robot.
- 10 November 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 208240-208252
Open AccessProximal Actuation of an Elastically Loaded Scissors Mechanism for the Leg Design of a Quadruped Robot
Spring Loaded Pantographs (SLPs) are frequently used in designing lightweight limbs for multi-legged robots. Quadruped robots that incorporate cable-pulled SLP legs have proven to be agile, robust and capable of conserving energy during their gait cycle. In such designs, the extension of the distal segments via the knee joint is dependent upon the length of the cable. In this article we propose the use of an Elastically Loaded Scissors Mechanism (ELS Mechanism or ELSM), which is a variant of the SLP. Driven by ’pulling’ onto the proximal joint of the scissors as opposed to the distal joint, this proposed leg utilizes the increased mechanical advantage of the scissors mechanism to ’amplify’ input angles to larger output displacement by the knee joint. Analysis and Simulations reveal that the proposed mechanism achieves increased motion speed as compared to the SLP mechanism. This, however, comes at the cost of higher load on the actuator which serves as an engineering trade-off. This is validated by experimentation using motion capture and load motor techniques of the SLP and ELS configurations in a physical quadruped robot.
- Open Access
A Modified Bayesian Framework for Multi-Sensor Target Tracking with Out-of-Sequence-Measurements
Target detection and tracking is important in military as well as in civilian applications. In order to detect and track high-speed incoming threats, modern surveillance systems are equipped with multiple sensors to overcome the limitations of single-sensor based tracking systems. This research proposes the use of information from RADAR and Infrared sensors (IR) for tracking and estimating target state dynamics. A new technique is developed for information fusion of the two sensors in a way that enhances performance of the data association algorithm. The measurement acquisition and processing time of these sensors is not the same; consequently the fusion center measurements arrive out of sequence. To ensure the practicality of system, proposed algorithm compensates the Out of Sequence Measurements (OOSMs) in cluttered environment. This is achieved by a novel algorithm which incorporates a retrodiction based approach to compensate the effects of OOSMs in a modified Bayesian technique. The proposed modification includes a new gating strategy to fuse and select measurements from two sensors which originate from the same target. The state estimation performance is evaluated in terms of Root Mean Squared Error (RMSE) for both position and velocity, whereas, track retention statistics are evaluated to gauge the performance of the proposed tracking algorithm. The results clearly show that the proposed technique improves track retention and and false track discrimination (FTD).
Open AccessA Modified Bayesian Framework for Multi-Sensor Target Tracking with Out-of-Sequence-Measurements
Target detection and tracking is important in military as well as in civilian applications. In order to detect and track high-speed incoming threats, modern surveillance systems are equipped with multiple sensors to overcome the limitations of single-sensor based tracking systems. This research proposes the use of information from RADAR and Infrared sensors (IR) for tracking and estimating target state dynamics. A new technique is developed for information fusion of the two sensors in a way that enhances performance of the data association algorithm. The measurement acquisition and processing time of these sensors is not the same; consequently the fusion center measurements arrive out of sequence. To ensure the practicality of system, proposed algorithm compensates the Out of Sequence Measurements (OOSMs) in cluttered environment. This is achieved by a novel algorithm which incorporates a retrodiction based approach to compensate the effects of OOSMs in a modified Bayesian technique. The proposed modification includes a new gating strategy to fuse and select measurements from two sensors which originate from the same target. The state estimation performance is evaluated in terms of Root Mean Squared Error (RMSE) for both position and velocity, whereas, track retention statistics are evaluated to gauge the performance of the proposed tracking algorithm. The results clearly show that the proposed technique improves track retention and and false track discrimination (FTD).
- Open Access
- 2 April 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 62990-63003
CPU and RAM Energy-Based SLA-Aware Workload Consolidation Techniques for Clouds
Cloud computing offers hardware and software resources delivered as services. It provides solutions for dynamic as well as “pay as you go” provision of resources. Energy consumption of these resources is high which leads to higher operational costs and carbon emissions in data centers. A number of research studies have been conducted on energy efficiency of data centers, but most of them concentrate on single factor energy consumption, i.e., energy consumed by CPU only, and energy consumption by Random Access Memory (RAM) is neglected. However, recently the focus has been turned towards impact of energy consumption by RAM on data centers. Studies have shown that RAM consumes about 25% of joint energy consumed by a server’s CPU and RAM. In this paper, two energy-aware virtual machine (VM) consolidation schemes are proposed that take into account a server’s capacity in terms of CPU and RAM to reduce the overall energy consumption. The proposed schemes are compared with existing schemes using CloudSim simulator. The results show that the proposed schemes reduce the energy cost with improved Service Level Agreement (SLA).
- 2 April 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 62990-63003
Open AccessCPU and RAM Energy-Based SLA-Aware Workload Consolidation Techniques for Clouds
Cloud computing offers hardware and software resources delivered as services. It provides solutions for dynamic as well as “pay as you go” provision of resources. Energy consumption of these resources is high which leads to higher operational costs and carbon emissions in data centers. A number of research studies have been conducted on energy efficiency of data centers, but most of them concentrate on single factor energy consumption, i.e., energy consumed by CPU only, and energy consumption by Random Access Memory (RAM) is neglected. However, recently the focus has been turned towards impact of energy consumption by RAM on data centers. Studies have shown that RAM consumes about 25% of joint energy consumed by a server’s CPU and RAM. In this paper, two energy-aware virtual machine (VM) consolidation schemes are proposed that take into account a server’s capacity in terms of CPU and RAM to reduce the overall energy consumption. The proposed schemes are compared with existing schemes using CloudSim simulator. The results show that the proposed schemes reduce the energy cost with improved Service Level Agreement (SLA).
- Top Cited PapersOpen Access
- 18 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 4171-4181
Skin Lesion Segmentation in Dermoscopic Images With Ensemble Deep Learning Methods
Early detection of skin cancer, particularly melanoma, is crucial to enable advanced treatment. Due to the rapid growth in the number of skin cancers, there is a growing need of computerised analysis for skin lesions. The state-of-the-art public available datasets for skin lesions are often accompanied with a very limited amount of segmentation ground truth labeling. Also, the available segmentation datasets consist of noisy expert annotations reflecting the fact that precise annotations to represent the boundary of skin lesions are laborious and expensive. The lesion boundary segmentation is vital to locate the lesion accurately in dermoscopic images and lesion diagnosis of different skin lesion types. In this work, we propose the fully automated deep learning ensemble methods to achieve high sensitivity and high specificity in lesion boundary segmentation. We trained the ensemble methods based on Mask RCNN and DeeplabV3+ methods on ISIC-2017 segmentation training set and evaluate the performance of the ensemble networks on ISIC-2017 testing set and PH2 dataset. Our results showed that the proposed ensemble methods segmented the skin lesions with Sensitivity of 89.93% and Specificity of 97.94% for the ISIC-2017 testing set. The proposed ensemble method Ensemble-A outperformed FrCN, FCNs, U-Net, and SegNet in Sensitivity by 4.4%, 8.8%, 22.7%, and 9.8% respectively. Furthermore, the proposed ensemble method Ensemble-S achieved a specificity score of 97.98% for clinically benign cases, 97.30% for the melanoma cases, and 98.58% for the seborrhoeic keratosis cases on ISIC-2017 testing set, exhibiting better performance than FrCN, FCNs, U-Net, and SegNet.
- 18 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 4171-4181
Top Cited PapersOpen AccessSkin Lesion Segmentation in Dermoscopic Images With Ensemble Deep Learning Methods
Early detection of skin cancer, particularly melanoma, is crucial to enable advanced treatment. Due to the rapid growth in the number of skin cancers, there is a growing need of computerised analysis for skin lesions. The state-of-the-art public available datasets for skin lesions are often accompanied with a very limited amount of segmentation ground truth labeling. Also, the available segmentation datasets consist of noisy expert annotations reflecting the fact that precise annotations to represent the boundary of skin lesions are laborious and expensive. The lesion boundary segmentation is vital to locate the lesion accurately in dermoscopic images and lesion diagnosis of different skin lesion types. In this work, we propose the fully automated deep learning ensemble methods to achieve high sensitivity and high specificity in lesion boundary segmentation. We trained the ensemble methods based on Mask RCNN and DeeplabV3+ methods on ISIC-2017 segmentation training set and evaluate the performance of the ensemble networks on ISIC-2017 testing set and PH2 dataset. Our results showed that the proposed ensemble methods segmented the skin lesions with Sensitivity of 89.93% and Specificity of 97.94% for the ISIC-2017 testing set. The proposed ensemble method Ensemble-A outperformed FrCN, FCNs, U-Net, and SegNet in Sensitivity by 4.4%, 8.8%, 22.7%, and 9.8% respectively. Furthermore, the proposed ensemble method Ensemble-S achieved a specificity score of 97.98% for clinically benign cases, 97.30% for the melanoma cases, and 98.58% for the seborrhoeic keratosis cases on ISIC-2017 testing set, exhibiting better performance than FrCN, FCNs, U-Net, and SegNet.
- Open Access
- 1 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 400, 185-192
Towards Interdisciplinary Research: A Bibliometric View of Information Communication Technology for Development in Different Disciplines
We present a novel bibliometric view to create a taxonomy of the interdisciplinary field of Information Communication Technology for Development (ICTD or ICT4D), using scientific documents published in the fields related to information communication technologies that were indexed in the Scopus database from 2001 to 2015. Our research approach utilizes a set of relevant terms and venues (journals and conferences) to procure publications on ICTD topics, and further to analyze them statistically to identify emerging or emerged sub-disciplines. The sub-disciplines prominently include Economic Growth/E-Government/Political Economy, Cloud Enterprises/Entrepreneurship/ICT Innovation, and Smart Energy/Smart Grid. We identify the relevance and uniqueness of all ICTD topics, generated by the algorithm, by employing well-known indices of the coherence coefficient and Kullback-Leibler divergence, indicating the qualifications of a given topic as an established sub-discipline of ICTD. This paper attempts to discuss the interdisciplinary nature of the research field ICTD and recommends prospective directions for new subdisciplines based on the text analysis.
- 1 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 400, 185-192
Open AccessTowards Interdisciplinary Research: A Bibliometric View of Information Communication Technology for Development in Different Disciplines
We present a novel bibliometric view to create a taxonomy of the interdisciplinary field of Information Communication Technology for Development (ICTD or ICT4D), using scientific documents published in the fields related to information communication technologies that were indexed in the Scopus database from 2001 to 2015. Our research approach utilizes a set of relevant terms and venues (journals and conferences) to procure publications on ICTD topics, and further to analyze them statistically to identify emerging or emerged sub-disciplines. The sub-disciplines prominently include Economic Growth/E-Government/Political Economy, Cloud Enterprises/Entrepreneurship/ICT Innovation, and Smart Energy/Smart Grid. We identify the relevance and uniqueness of all ICTD topics, generated by the algorithm, by employing well-known indices of the coherence coefficient and Kullback-Leibler divergence, indicating the qualifications of a given topic as an established sub-discipline of ICTD. This paper attempts to discuss the interdisciplinary nature of the research field ICTD and recommends prospective directions for new subdisciplines based on the text analysis.