Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570918
Mfanufikile Ncube, Joyce B. Mwangama
In order to enhance the quality of life to its citizens, cities invest in digital infrastructures to allow for sustainable environments and smart applications. However, there are significant concerns about the benefits of smart city solutions as compared to the initial investment needed for their implementation. Current proposed solutions involve discarding proprietary platforms and connecting legacy sensors onto open standards-based platform via gateways. Therefore, our approach involves designing a gateway to integrate sensor data onto an ETSI oneM2M compliant smart city solution. Our validation shows that gateways can be designed to integrate different smart city platforms using different access technologies without affecting the independent functionalities of the platforms. The main contribution of this paper is a model that enables sharing of sensor infrastructure between a smart city platform belonging to a private organisation and a citywide platform which may belong to the Municipal.
{"title":"Convergence of Various Smart City Platforms into a Unified Citywide Platform","authors":"Mfanufikile Ncube, Joyce B. Mwangama","doi":"10.1109/africon51333.2021.9570918","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570918","url":null,"abstract":"In order to enhance the quality of life to its citizens, cities invest in digital infrastructures to allow for sustainable environments and smart applications. However, there are significant concerns about the benefits of smart city solutions as compared to the initial investment needed for their implementation. Current proposed solutions involve discarding proprietary platforms and connecting legacy sensors onto open standards-based platform via gateways. Therefore, our approach involves designing a gateway to integrate sensor data onto an ETSI oneM2M compliant smart city solution. Our validation shows that gateways can be designed to integrate different smart city platforms using different access technologies without affecting the independent functionalities of the platforms. The main contribution of this paper is a model that enables sharing of sensor infrastructure between a smart city platform belonging to a private organisation and a citywide platform which may belong to the Municipal.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116015326","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570887
S. Ekwe, L. Akinyemi, S. Oladejo, N. Ventura
This paper investigates a joint uplink and downlink resource allocation problem for a 5G use case. We explore the social-awareness of the network operators to efficiently match users during any form of peer-to-peer communication. Thus, we proposed a peer-selection scheme to improve the overall utility of the network amid limited spectral resources while exploiting the social-ties of network users’. Consequently, we formulate a utility maximization problem as a mixed-integer non-linear programming (MINLP) problem to be solved using genetic algorithm. We perform extensive Monte-Carlo simulations alongside several meta-heuristic algorithms for comparison. The results reveal that our proposed scheme is a good candidate to appreciably and significantly improve the expected utility and network performance.
{"title":"Social-Aware Joint Uplink and Downlink Resource Allocation Scheme Using Genetic Algorithm","authors":"S. Ekwe, L. Akinyemi, S. Oladejo, N. Ventura","doi":"10.1109/africon51333.2021.9570887","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570887","url":null,"abstract":"This paper investigates a joint uplink and downlink resource allocation problem for a 5G use case. We explore the social-awareness of the network operators to efficiently match users during any form of peer-to-peer communication. Thus, we proposed a peer-selection scheme to improve the overall utility of the network amid limited spectral resources while exploiting the social-ties of network users’. Consequently, we formulate a utility maximization problem as a mixed-integer non-linear programming (MINLP) problem to be solved using genetic algorithm. We perform extensive Monte-Carlo simulations alongside several meta-heuristic algorithms for comparison. The results reveal that our proposed scheme is a good candidate to appreciably and significantly improve the expected utility and network performance.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"72 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126258916","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9571020
Sorin Olaru, F. Stoican, S. Kheawhom
The paper aims to review recent developments and points to the challenges and opportunities for the instrumentation, control and management technologies in relationship with the emergence of novel Energy Storage Systems. It exemplifies this trend with a technology that has received an increasing interest in recent studies, the Zinc-Air batteries.
{"title":"Challenges and opportunities for the control of Energy Storage Systems. A focus on the Zinc-Air batteries.","authors":"Sorin Olaru, F. Stoican, S. Kheawhom","doi":"10.1109/africon51333.2021.9571020","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9571020","url":null,"abstract":"The paper aims to review recent developments and points to the challenges and opportunities for the instrumentation, control and management technologies in relationship with the emergence of novel Energy Storage Systems. It exemplifies this trend with a technology that has received an increasing interest in recent studies, the Zinc-Air batteries.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126594298","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Many call centre shift scheduling approaches focus on one call centre day when determining the number of agents to be assigned to each shift. However, this approach makes the assumption that shifts will be filled with the same agents everyday, and ignores the practicalities of an actual call centre like day-offs, which would require shift assignments over longer time horizons. Moreover, many of these shift scheduling approaches use the arrival rate and service rate as inputs. This presents an issue because it might be difficult to estimate these rates with confidence from the data, especially the arrival rate which fluctuates during the day. We present a local search heuristic approach of assigning shifts and day-offs to existing call centre agents using hill climbing, tabu search, and simulated annealing. This is achieved without increasing the staffing costs. Our methods use individual calls data directly, therefore removing the need to estimate the arrival rate, and minimising the need to estimate the service rate. The methods are applied to real-life data from a call centre and the results show improvements in the achieved service level and a significant drop in the number of abandoned calls.
{"title":"Call Centre Shift Schedule Optimisation using Local Search Heuristics","authors":"Liketso Nthimo, Tshepiso Mokoena, Abiodun Modupe, Vukosi Marivate","doi":"10.1109/africon51333.2021.9570947","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570947","url":null,"abstract":"Many call centre shift scheduling approaches focus on one call centre day when determining the number of agents to be assigned to each shift. However, this approach makes the assumption that shifts will be filled with the same agents everyday, and ignores the practicalities of an actual call centre like day-offs, which would require shift assignments over longer time horizons. Moreover, many of these shift scheduling approaches use the arrival rate and service rate as inputs. This presents an issue because it might be difficult to estimate these rates with confidence from the data, especially the arrival rate which fluctuates during the day. We present a local search heuristic approach of assigning shifts and day-offs to existing call centre agents using hill climbing, tabu search, and simulated annealing. This is achieved without increasing the staffing costs. Our methods use individual calls data directly, therefore removing the need to estimate the arrival rate, and minimising the need to estimate the service rate. The methods are applied to real-life data from a call centre and the results show improvements in the achieved service level and a significant drop in the number of abandoned calls.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128266109","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570956
N. Mduma, D. Machuve
Student dropout is among the challenges that face most schools in developing countries particularly in Africa. In addressing the student dropout problem, a thorough understanding of the fundamental causative factors is essential. Several researchers have identified and proposed causes, methods and strategies that will help to reduce or stop the student dropout problem, however, most of the proposed solutions did not show promising results and the dropout trend continue to increase over time. Machine learning on the other hand has gained much attention when addressing society’s problems in different sectors including education. This is attributed by the fact that, machine learning models when accurately trained, provide convenient and reliable results as compared to the traditional approaches. This study focused on developing a machine learning model that will help to predict and identify students who are at risk of dropping out of school. Three datasets from Tanzania, Kenya and Uganda were used to develop the model and disclose the best classifier from the three commonly used i.e. Multilayer Perceptron, Logistic Regression and Random Forest. Classifiers were evaluated using Geometric Mean and F-measure to examine their performance. Results revealed that, Logistic Regression achieved the highest performance as compared to the other two. The study, therefore, recommends the developed model to be used by relevant authorities in identifying and predicting students who are at risk of dropping out of schools, and make informative decisions on addressing the student dropout problem.
{"title":"Machine Learning Model for Predicting Student Dropout: A Case of Tanzania, Kenya and Uganda","authors":"N. Mduma, D. Machuve","doi":"10.1109/africon51333.2021.9570956","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570956","url":null,"abstract":"Student dropout is among the challenges that face most schools in developing countries particularly in Africa. In addressing the student dropout problem, a thorough understanding of the fundamental causative factors is essential. Several researchers have identified and proposed causes, methods and strategies that will help to reduce or stop the student dropout problem, however, most of the proposed solutions did not show promising results and the dropout trend continue to increase over time. Machine learning on the other hand has gained much attention when addressing society’s problems in different sectors including education. This is attributed by the fact that, machine learning models when accurately trained, provide convenient and reliable results as compared to the traditional approaches. This study focused on developing a machine learning model that will help to predict and identify students who are at risk of dropping out of school. Three datasets from Tanzania, Kenya and Uganda were used to develop the model and disclose the best classifier from the three commonly used i.e. Multilayer Perceptron, Logistic Regression and Random Forest. Classifiers were evaluated using Geometric Mean and F-measure to examine their performance. Results revealed that, Logistic Regression achieved the highest performance as compared to the other two. The study, therefore, recommends the developed model to be used by relevant authorities in identifying and predicting students who are at risk of dropping out of schools, and make informative decisions on addressing the student dropout problem.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128441251","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570990
Andrea Koble, Ágnes Győrfi, Szabolcs Csaholczi, Béla Surányi, Lehel Dénes-Fazakas, L. Kovács, L. Szilágyi
The main drawback of magnetic resonance imaging (MRI) represents the lack of a standard intensity scale. All observed numerical values are relative and can only be interpreted together with their context. Before feeding MRI data volumes to supervised learning segmentation procedures, their histograms need to be registered to each other, or in other words, they need a so-called normalization. The most popular histogram normalization technique used to assist brain MRI segmentation is the algorithm proposed by Nyuĺ et al in 2000, which aligns the histograms of a batch of MRI volumes without paying attention to possible focal lesions that might distort the histogram. Alternately, some recent works applied histogram normalization based on a simple linear transform, and reported achieving slightly better accuracy with them. This paper proposes to investigate, which is the most suitable method and parameter settings for histogram normalization to be performed before the segmentation of brain MRI images, separately in the cases of absence and presence of focal lesions.
{"title":"Identifying the most suitable histogram normalization technique for machine learning based segmentation of multispectral brain MRI data","authors":"Andrea Koble, Ágnes Győrfi, Szabolcs Csaholczi, Béla Surányi, Lehel Dénes-Fazakas, L. Kovács, L. Szilágyi","doi":"10.1109/africon51333.2021.9570990","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570990","url":null,"abstract":"The main drawback of magnetic resonance imaging (MRI) represents the lack of a standard intensity scale. All observed numerical values are relative and can only be interpreted together with their context. Before feeding MRI data volumes to supervised learning segmentation procedures, their histograms need to be registered to each other, or in other words, they need a so-called normalization. The most popular histogram normalization technique used to assist brain MRI segmentation is the algorithm proposed by Nyuĺ et al in 2000, which aligns the histograms of a batch of MRI volumes without paying attention to possible focal lesions that might distort the histogram. Alternately, some recent works applied histogram normalization based on a simple linear transform, and reported achieving slightly better accuracy with them. This paper proposes to investigate, which is the most suitable method and parameter settings for histogram normalization to be performed before the segmentation of brain MRI images, separately in the cases of absence and presence of focal lesions.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127297163","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570929
Xi Huang, Shibin Zhang, Wen Cheng
In this paper, an efficient quantum private comparison (QPC) protocol based on GHZ-type states is proposed. Two participants can compare the equality of two classical bits in each comparison, which could greatly reduce comparison times and increase efficiency. A semi-honest third-party (TP) is involved in assisting the participants to compare their secrets. TP may misbehave on her own, but she is not allowed to conspire with any participants. Besides, the proposed protocol needs Hadamard operation as well as single-particle measurements and Bell measurements, which are easy to implement with current technologies. Finally, the analysis shows the proposed protocol is correct and it can resist various attacks including outside attacks and dishonest participant attacks.
{"title":"Quantum Private Comparison Based on GHZ-type States","authors":"Xi Huang, Shibin Zhang, Wen Cheng","doi":"10.1109/africon51333.2021.9570929","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570929","url":null,"abstract":"In this paper, an efficient quantum private comparison (QPC) protocol based on GHZ-type states is proposed. Two participants can compare the equality of two classical bits in each comparison, which could greatly reduce comparison times and increase efficiency. A semi-honest third-party (TP) is involved in assisting the participants to compare their secrets. TP may misbehave on her own, but she is not allowed to conspire with any participants. Besides, the proposed protocol needs Hadamard operation as well as single-particle measurements and Bell measurements, which are easy to implement with current technologies. Finally, the analysis shows the proposed protocol is correct and it can resist various attacks including outside attacks and dishonest participant attacks.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131927158","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570968
Wassihun Beyene W. Mariam, Y. Negash
One of the main security problems that become the hardest and most serious threat is called Distributed Denial of Service (DDoS) attacks specifically Synchronize (SYN) flood attack. This research focused on the performance evaluation of classification machine learning (ML) algorithms for SYN flood attack detection. The classification models are trained and tested with packet captured dataset gathered from ethio telecom network by generating and capturing packets using Hping3 and Wireshark tools respectively. This dataset has been further preprocessed and evaluated using four classification ML algorithms and three training approaches. The implementation has been performed using WAKA (Waikato Environment for Knowledge Analysis) data mining tool. The experimental results show that the J48 algorithm performs with 98.57% accuracy and AdaBoost, Naïve Bayes and ANN algorithms with 98.52%, 95.31% and 94.85% accuracy respectively. Accordingly based on the performance a model with the J48 algorithm has been recommended for SYN attack detection.
分布式拒绝服务(DDoS)攻击是最严重的安全威胁之一，特别是同步(SYN)洪水攻击。本文主要研究了分类机器学习算法在SYN flood攻击检测中的性能评估。通过使用Hping3和Wireshark工具分别生成和捕获数据包，利用从埃塞俄比亚电信网络收集的数据包捕获数据集对分类模型进行训练和测试。该数据集已经使用四种分类ML算法和三种训练方法进行了进一步的预处理和评估。使用WAKA (Waikato Environment for Knowledge Analysis)数据挖掘工具进行实现。实验结果表明，J48算法的准确率为98.57%，AdaBoost、Naïve贝叶斯和ANN算法的准确率分别为98.52%、95.31%和94.85%。在此基础上，提出了一种基于J48算法的SYN攻击检测模型。
{"title":"Performance Evaluation of Machine Learning Algorithms for Detection of SYN Flood Attack","authors":"Wassihun Beyene W. Mariam, Y. Negash","doi":"10.1109/africon51333.2021.9570968","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570968","url":null,"abstract":"One of the main security problems that become the hardest and most serious threat is called Distributed Denial of Service (DDoS) attacks specifically Synchronize (SYN) flood attack. This research focused on the performance evaluation of classification machine learning (ML) algorithms for SYN flood attack detection. The classification models are trained and tested with packet captured dataset gathered from ethio telecom network by generating and capturing packets using Hping3 and Wireshark tools respectively. This dataset has been further preprocessed and evaluated using four classification ML algorithms and three training approaches. The implementation has been performed using WAKA (Waikato Environment for Knowledge Analysis) data mining tool. The experimental results show that the J48 algorithm performs with 98.57% accuracy and AdaBoost, Naïve Bayes and ANN algorithms with 98.52%, 95.31% and 94.85% accuracy respectively. Accordingly based on the performance a model with the J48 algorithm has been recommended for SYN attack detection.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"186 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131502135","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570899
Lwando Ngcama, N. Mostert
The use of social media in higher education has both benefits and pitfalls. The Nelson Mandela University in South Africa has implemented a social media policy and a set of guidelines to inform and enforce the acceptable use of social media by its staff and students. In order to know whether staff and students at the Nelson Mandela University are aware of and compliant with its policy and guidelines, their level of awareness and compliance was measured. Within this context, the primary objective of this study is to describe the state of awareness and compliance of staff and students at the Nelson Mandela University towards its social media policy and social media guidelines. The level of awareness and compliance of staff and students at the Nelson Mandela University in respect of the university's social media policy and guidelines was measured and described through the use of a survey questionnaire and statistical analysis of the data collected. The results of the analysis indicated an overall medium level of awareness for both staff and students, with a mean average score of 3.213 out of a possible maximum score of 5; while both groups demonstrated an overall high level of compliance towards the social media policy and guidelines, with a mean average score of 4.256.
{"title":"An Evaluation of Social Media Policy Awareness and Compliance at the Nelson Mandela University","authors":"Lwando Ngcama, N. Mostert","doi":"10.1109/africon51333.2021.9570899","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570899","url":null,"abstract":"The use of social media in higher education has both benefits and pitfalls. The Nelson Mandela University in South Africa has implemented a social media policy and a set of guidelines to inform and enforce the acceptable use of social media by its staff and students. In order to know whether staff and students at the Nelson Mandela University are aware of and compliant with its policy and guidelines, their level of awareness and compliance was measured. Within this context, the primary objective of this study is to describe the state of awareness and compliance of staff and students at the Nelson Mandela University towards its social media policy and social media guidelines. The level of awareness and compliance of staff and students at the Nelson Mandela University in respect of the university's social media policy and guidelines was measured and described through the use of a survey questionnaire and statistical analysis of the data collected. The results of the analysis indicated an overall medium level of awareness for both staff and students, with a mean average score of 3.213 out of a possible maximum score of 5; while both groups demonstrated an overall high level of compliance towards the social media policy and guidelines, with a mean average score of 4.256.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134161919","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.1109/africon51333.2021.9570952
David Iclanzan, R. Lung, Zsolt Levente Kucsván, Béla Surányi, Levente Kovács, László Szilágyi
Atlas assisted image segmentation has been quite popular in medical imaging during the last two decades. The atlas is able to provide prior information on the imaged organ’s shape, appearance, and local texture or intensity distribution. In case of segmenting images via pixelwise classification, the final segmentation result is obtained through a fusion of the classification outcome with the local atlas information. In other words, the atlas guides the classifier towards the shape of local structures normally situated at the given location. This paper proposes to demonstrate the advantages a multi-atlas can bring in a segmentation process of the main tissues in infant brain based on multi-spectral MRI records. Three supervised machine learning methods are deployed to segment brain tissues, with and without the use of the atlas. Differences are evaluated using statistical accuracy indicators. Atlases improved the overall segmentation accuracy by 2.5-3.5%, depending on the deployed classifier method.
{"title":"The role of atlases and multi-atlases in brain tissue segmentation based on multispectral magnetic resonance image data","authors":"David Iclanzan, R. Lung, Zsolt Levente Kucsván, Béla Surányi, Levente Kovács, László Szilágyi","doi":"10.1109/africon51333.2021.9570952","DOIUrl":"https://doi.org/10.1109/africon51333.2021.9570952","url":null,"abstract":"Atlas assisted image segmentation has been quite popular in medical imaging during the last two decades. The atlas is able to provide prior information on the imaged organ’s shape, appearance, and local texture or intensity distribution. In case of segmenting images via pixelwise classification, the final segmentation result is obtained through a fusion of the classification outcome with the local atlas information. In other words, the atlas guides the classifier towards the shape of local structures normally situated at the given location. This paper proposes to demonstrate the advantages a multi-atlas can bring in a segmentation process of the main tissues in infant brain based on multi-spectral MRI records. Three supervised machine learning methods are deployed to segment brain tissues, with and without the use of the atlas. Differences are evaluated using statistical accuracy indicators. Atlases improved the overall segmentation accuracy by 2.5-3.5%, depending on the deployed classifier method.","PeriodicalId":170342,"journal":{"name":"2021 IEEE AFRICON","volume":"38 11","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132434015","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：