اختيار الموقع            تسجيل دخول
 

تصفح المحتوي RDA
التصفح حسب الموضوعات
التصفح حسب اللغة
التصفح حسب الناشر
التصفح حسب تاريخ النشر
التصفح حسب مكان النشر
التصفح حسب المؤلفين
تصفح الهيئات
التصفح المؤتمرات
التصفح حسب نوع المادة
التصفح حسب العلاقة بالعمل
تم العثور علي : 7651
 تم العثور علي : 7651
  
 
إعادة البحث

Thesis 2024.

Articles 2024.
Vol. 97, No. 1 (June 2024) /

Articles 2024.
Vol. 99, No. 1 (July 2024) /

Thesis 2024
The use of image classification in medical fields is one of the most important uses - including skin cancer image classification. Skin cancer is a major health problem across the world - and early identification is critical for successful treatment. Skin cancer - which is defined by abnormal skin cell development - is a common and dangerous disease worldwide. Despite advances in digital diagnosis tools - many present skin cancer detection technologies frequently fail to attain adequate levels of accuracy. Disease detection - computer-aided diagnosis - and patient risk identification rely heavily on computer vision. This is particularly true for skin cancer - which may be lethal if not detected early on. Several computer-aided diagnosis and detection systems have already been developed to do this.
In this dissertation
- two approaches for classifying skin cancer images were examined and compared with the proposed methods. Machine Learning (ML) and Deep Learning (DL) are these two approaches. ML approaches include Artificial Neural Networks - Support Vector Machines - Naïve Bayes - and Decision Tree. Both Convolutional Neural Networks and Pretrained Deep Neural Networks (PDNN) were employed in the DL approach.
Two methods for detecting and binary classifying dermoscopic skin cancer images into benign and malignant were proposed. The first proposed method employs K-Nearest Neighbor (KNN) as a classifier with several PDNN serving as feature extractors
- (KNN-PDNN). These networks include AlexNet - VGG-16 - VGG-19 - EfficientNet-B0 - ResNet-18 - ResNet-50 - ResNet-101 - DenseNet-201 - Inception-v3 - and MobileNet-v2. The second proposed method employs some PDNN with the Improved Grey Wolf Optimizer (I-GWO) - (PDNN-I-GWO). The PDNN used in this technique are AlexNet - ResNet-18 - SqueezeNet - ShuffleNet - and DarkNet-19.
The experiments of KNN-PDNN method used 4000 images from the ISIC archive dataset to train and test images. In certain PDNN
- the KNN-PDNN method’s accuracy exceeded 99%. The PDNN-I-GWO method investigated two datasets: MED-NODE and DermIS. The outcomes showed that the proposed methods outperformed the other tested approaches. The highest accuracy achieved by this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively. The highest accuracy achieved with this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively.
The dissertation consists of five chapters as follows:
Chapter 1: Introduction
An introduction to the dissertation is given
- explaining the importance of the research point and the goals it seeks to achieve - and an explanation of the problems found in some of the old techniques that we seek to improve in this thesis and the extent of their impact on classifying skin cancer images. This chapter also summarizes what the other chapters contain and the order in which they are reviewed in the thesis.
Chapter 2: Literature Review
This chapter covers background on skin cancer image classification and presents some previous works and methods used and their features and characteristics.
Chapter 3: Proposed System
The third chapter presents the proposed algorithms that were represented and applied in the dissertation. It reviews them in detail and discusses the additions and modifications that were made to achieve high accuracy. This chapter also presents the preprocessing of images before using them in the proposed methods. In addition
- it includes different datasets for training and testing images.
Chapter 4: Experimental Results
It reviews all the experiments
- their accompanying results - and details of the images that were used in the experiments. This dissertation also includes many comparisons between the proposed and modified algorithms that were used during the image classification process. This included using several methods and methods to evaluate and compare the performance of these algorithms.
Chapter 5: Conclusions and Recommendations for Future Work
It presents a summary of the results reached as well as some recommended points for future work that can be used to develop the work presented in this dissertation or related works
- The use of image classification in medical fields is one of the most important uses - including skin cancer image classification. Skin cancer is a major health problem across the world - and early identification is critical for successful treatment. Skin cancer - which is defined by abnormal skin cell development - is a common and dangerous disease worldwide. Despite advances in digital diagnosis tools - many present skin cancer detection technologies frequently fail to attain adequate levels of accuracy. Disease detection - computer-aided diagnosis - and patient risk identification rely heavily on computer vision. This is particularly true for skin cancer - which may be lethal if not detected early on. Several computer-aided diagnosis and detection systems have already been developed to do this.
In this dissertation
- two approaches for classifying skin cancer images were examined and compared with the proposed methods. Machine Learning (ML) and Deep Learning (DL) are these two approaches. ML approaches include Artificial Neural Networks - Support Vector Machines - Naïve Bayes - and Decision Tree. Both Convolutional Neural Networks and Pretrained Deep Neural Networks (PDNN) were employed in the DL approach.
Two methods for detecting and binary classifying dermoscopic skin cancer images into benign and malignant were proposed. The first proposed method employs K-Nearest Neighbor (KNN) as a classifier with several PDNN serving as feature extractors
- (KNN-PDNN). These networks include AlexNet - VGG-16 - VGG-19 - EfficientNet-B0 - ResNet-18 - ResNet-50 - ResNet-101 - DenseNet-201 - Inception-v3 - and MobileNet-v2. The second proposed method employs some PDNN with the Improved Grey Wolf Optimizer (I-GWO) - (PDNN-I-GWO). The PDNN used in this technique are AlexNet - ResNet-18 - SqueezeNet - ShuffleNet - and DarkNet-19.
The experiments of KNN-PDNN method used 4000 images from the ISIC archive dataset to train and test images. In certain PDNN
- the KNN-PDNN method’s accuracy exceeded 99%. The PDNN-I-GWO method investigated two datasets: MED-NODE and DermIS. The outcomes showed that the proposed methods outperformed the other tested approaches. The highest accuracy achieved by this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively. The highest accuracy achieved with this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively.
The dissertation consists of five chapters as follows:
Chapter 1: Introduction
An introduction to the dissertation is given
- explaining the importance of the research point and the goals it seeks to achieve - and an explanation of the problems found in some of the old techniques that we seek to improve in this thesis and the extent of their impact on classifying skin cancer images. This chapter also summarizes what the other chapters contain and the order in which they are reviewed in the thesis.
Chapter 2: Literature Review
This chapter covers background on skin cancer image classification and presents some previous works and methods used and their features and characteristics.
Chapter 3: Proposed System
The third chapter presents the proposed algorithms that were represented and applied in the dissertation. It reviews them in detail and discusses the additions and modifications that were made to achieve high accuracy. This chapter also presents the preprocessing of images before using them in the proposed methods. In addition
- it includes different datasets for training and testing images.
Chapter 4: Experimental Results
It reviews all the experiments
- their accompanying results - and details of the images that were used in the experiments. This dissertation also includes many comparisons between the proposed and modified algorithms that were used during the image classification process. This included using several methods and methods to evaluate and compare the performance of these algorithms.
Chapter 5: Conclusions and Recommendations for Future Work
It presents a summary of the results reached as well as some recommended points for future work that can be used to develop the work presented in this dissertation or related works

Thesis 2024.

Thesis

Thesis 2024.
The use of image classification in medical fields is one of the most important uses - including skin cancer image classification. Skin cancer is a major health problem across the world - and early identification is critical for successful treatment. Skin cancer - which is defined by abnormal skin cell development - is a common and dangerous disease worldwide. Despite advances in digital diagnosis tools - many present skin cancer detection technologies frequently fail to attain adequate levels of accuracy. Disease detection - computer-aided diagnosis - and patient risk identification rely heavily on computer vision. This is particularly true for skin cancer - which may be lethal if not detected early on. Several computer-aided diagnosis and detection systems have already been developed to do this.
In this dissertation
- two approaches for classifying skin cancer images were examined and compared with the proposed methods. Machine Learning (ML) and Deep Learning (DL) are these two approaches. ML approaches include Artificial Neural Networks - Support Vector Machines - Naïve Bayes - and Decision Tree. Both Convolutional Neural Networks and Pretrained Deep Neural Networks (PDNN) were employed in the DL approach.
Two methods for detecting and binary classifying dermoscopic skin cancer images into benign and malignant were proposed. The first proposed method employs K-Nearest Neighbor (KNN) as a classifier with several PDNN serving as feature extractors
- (KNN-PDNN). These networks include AlexNet - VGG-16 - VGG-19 - EfficientNet-B0 - ResNet-18 - ResNet-50 - ResNet-101 - DenseNet-201 - Inception-v3 - and MobileNet-v2. The second proposed method employs some PDNN with the Improved Grey Wolf Optimizer (I-GWO) - (PDNN-I-GWO). The PDNN used in this technique are AlexNet - ResNet-18 - SqueezeNet - ShuffleNet - and DarkNet-19.
The experiments of KNN-PDNN method used 4000 images from the ISIC archive dataset to train and test images. In certain PDNN
- the KNN-PDNN method’s accuracy exceeded 99%. The PDNN-I-GWO method investigated two datasets: MED-NODE and DermIS. The outcomes showed that the proposed methods outperformed the other tested approaches. The highest accuracy achieved by this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively. The highest accuracy achieved with this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively.
The dissertation consists of five chapters as follows:
Chapter 1: Introduction
An introduction to the dissertation is given
- explaining the importance of the research point and the goals it seeks to achieve - and an explanation of the problems found in some of the old techniques that we seek to improve in this thesis and the extent of their impact on classifying skin cancer images. This chapter also summarizes what the other chapters contain and the order in which they are reviewed in the thesis.
Chapter 2: Literature Review
This chapter covers background on skin cancer image classification and presents some previous works and methods used and their features and characteristics.
Chapter 3: Proposed System
The third chapter presents the proposed algorithms that were represented and applied in the dissertation. It reviews them in detail and discusses the additions and modifications that were made to achieve high accuracy. This chapter also presents the preprocessing of images before using them in the proposed methods. In addition
- it includes different datasets for training and testing images.
Chapter 4: Experimental Results
It reviews all the experiments
- their accompanying results - and details of the images that were used in the experiments. This dissertation also includes many comparisons between the proposed and modified algorithms that were used during the image classification process. This included using several methods and methods to evaluate and compare the performance of these algorithms.
Chapter 5: Conclusions and Recommendations for Future Work
It presents a summary of the results reached as well as some recommended points for future work that can be used to develop the work presented in this dissertation or related works
- The use of image classification in medical fields is one of the most important uses - including skin cancer image classification. Skin cancer is a major health problem across the world - and early identification is critical for successful treatment. Skin cancer - which is defined by abnormal skin cell development - is a common and dangerous disease worldwide. Despite advances in digital diagnosis tools - many present skin cancer detection technologies frequently fail to attain adequate levels of accuracy. Disease detection - computer-aided diagnosis - and patient risk identification rely heavily on computer vision. This is particularly true for skin cancer - which may be lethal if not detected early on. Several computer-aided diagnosis and detection systems have already been developed to do this.
In this dissertation
- two approaches for classifying skin cancer images were examined and compared with the proposed methods. Machine Learning (ML) and Deep Learning (DL) are these two approaches. ML approaches include Artificial Neural Networks - Support Vector Machines - Naïve Bayes - and Decision Tree. Both Convolutional Neural Networks and Pretrained Deep Neural Networks (PDNN) were employed in the DL approach.
Two methods for detecting and binary classifying dermoscopic skin cancer images into benign and malignant were proposed. The first proposed method employs K-Nearest Neighbor (KNN) as a classifier with several PDNN serving as feature extractors
- (KNN-PDNN). These networks include AlexNet - VGG-16 - VGG-19 - EfficientNet-B0 - ResNet-18 - ResNet-50 - ResNet-101 - DenseNet-201 - Inception-v3 - and MobileNet-v2. The second proposed method employs some PDNN with the Improved Grey Wolf Optimizer (I-GWO) - (PDNN-I-GWO). The PDNN used in this technique are AlexNet - ResNet-18 - SqueezeNet - ShuffleNet - and DarkNet-19.
The experiments of KNN-PDNN method used 4000 images from the ISIC archive dataset to train and test images. In certain PDNN
- the KNN-PDNN method’s accuracy exceeded 99%. The PDNN-I-GWO method investigated two datasets: MED-NODE and DermIS. The outcomes showed that the proposed methods outperformed the other tested approaches. The highest accuracy achieved by this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively. The highest accuracy achieved with this method is 100% and 97% in the MED-NODE and DermIS datasets - respectively.
The dissertation consists of five chapters as follows:
Chapter 1: Introduction
An introduction to the dissertation is given
- explaining the importance of the research point and the goals it seeks to achieve - and an explanation of the problems found in some of the old techniques that we seek to improve in this thesis and the extent of their impact on classifying skin cancer images. This chapter also summarizes what the other chapters contain and the order in which they are reviewed in the thesis.
Chapter 2: Literature Review
This chapter covers background on skin cancer image classification and presents some previous works and methods used and their features and characteristics.
Chapter 3: Proposed System
The third chapter presents the proposed algorithms that were represented and applied in the dissertation. It reviews them in detail and discusses the additions and modifications that were made to achieve high accuracy. This chapter also presents the preprocessing of images before using them in the proposed methods. In addition
- it includes different datasets for training and testing images.
Chapter 4: Experimental Results
It reviews all the experiments
- their accompanying results - and details of the images that were used in the experiments. This dissertation also includes many comparisons between the proposed and modified algorithms that were used during the image classification process. This included using several methods and methods to evaluate and compare the performance of these algorithms.
Chapter 5: Conclusions and Recommendations for Future Work
It presents a summary of the results reached as well as some recommended points for future work that can be used to develop the work presented in this dissertation or related works

Thesis 2024.
The application of anaerobic processes has tended to be restricted to
strong industrial wastewaters. The success of anaerobic processes as a
treatment technology for high strength
- industrial wastewater has meant
that the potential of these processes for the treatment of low strength
wastewater has been evaluated. However
- one of the main challenges to
anaerobic technology remains its applicability to low-strength wastewaters
like sewage. The up-flow anaerobic sludge blanket (UASB) reactor is the
most widely and successfully used high rate anaerobic system for
wastewater treatment. The aim of the thesis is to increase the efficiency of
the system by adding conductive materials. Ecofriendly bio-adsorbents
such as Rice Straw
- Phragmites australis - and Commercial Activated
Carbon were used for chemical oxygen demand (COD) removal and
Biogas production from wastewater. Experiments using a multilevel
complete factorial design were conducted to optimize the removal
effectiveness of COD (Chemical Oxygen Demand) while minimizing the
number of experiments required. To verify the structural characteristics
- elemental composition - and the existence of various functional groups - a
characterization investigation was conducted using X-ray diffractometry
(XRD)
- Fourier Transform InfraRed spectroscopy (FTIR) - Scanning
Electron Microscopy (SEM)
- and Brunner–Emmett–Teller (BET). Batch
experimental trails were operated to determine the optimum adsorpant
material
- its optimum dose - as well as the other operational parameters - such as solution pH - inoculation concentration - and their interactions
during COD removal and Biogas production were investigated. The
maximum removal of COD (99.63%) and the biogas production (5.16 mL
biogas/mg COD removed) of Rice Straw Biochar (RSB) were at pH value
- biochar dose - and buffalo sludge dose concentration were equal to 8 - 2 g/L -
IV
and 0%
- respectively. Commercial Activated Carbon (AC) has achieved
maximum removal of COD (95.55%)
- and the biogas production (6.08 mL
biogas/mg COD removed) at pH
- biochar dose - and buffalo sludge dose
concentration were equal to 5
- 2 g/L - and 0% - respectively. The maximum
removal of COD (98.88%) and the biogas production (4.08 mL biogas/mg
COD removed) of Phragmites australis Biochar (PaB) were at pH
- biochar dose - and buffalo sludge dose concentration were equal to 5 - 2 g/L - and 0% - respectively. These results revealed that rice straw biochar can be
used as an effective and low-cost adsorbent to remove COD from
wastewater. The surface properties of rice straw biochar substantially
affect its capability of removing metal ions from wastewater
- and fourier
transform infrared spectroscopy (FTIR) spectroscopy is a great tool to
observe this surface composition. Two identical pilot-scale models
simulating “Up-flow Anaerobic Sludge Blanket” reactors (UASBs) were
built and operated continuously within the work frame of the present work
to investigate its performance and efficiency in treating buffalo
wastewater treatment. The effect of supporting media on the UASB
efficiency will be also invistgated at the field. The two UASB reactors
were operated under the same operational conditions and scenario
- the
reators operated at HRT equals 4hr and ambiaint temperature. Both R4
(conventional UASB) and R3 (modified UASB) were fed by settleled
wastewater . The condutive media was not added to R3 at the beigning. After the start-up
- the modified UASB reactor (R3) was inoculated with
rice straw biochar through an inclined pipe. Samples were collected and
analyzed periodically twice weekly. The results indicated that; For the
conventional reactor
- the maximum removal efficiency of COD - TSS - TDS - Color - and Turbidity was 79.89% - 74.04% - 80.11% - 72.72% - and
75.70%
- respectively. Cumulative biogas production reached 0.028 mL - The application of anaerobic processes has tended to be restricted to
strong industrial wastewaters. The success of anaerobic processes as a
treatment technology for high strength
- industrial wastewater has meant
that the potential of these processes for the treatment of low strength
wastewater has been evaluated. However
- one of the main challenges to
anaerobic technology remains its applicability to low-strength wastewaters
like sewage. The up-flow anaerobic sludge blanket (UASB) reactor is the
most widely and successfully used high rate anaerobic system for
wastewater treatment. The aim of the thesis is to increase the efficiency of
the system by adding conductive materials. Ecofriendly bio-adsorbents
such as Rice Straw
- Phragmites australis - and Commercial Activated
Carbon were used for chemical oxygen demand (COD) removal and
Biogas production from wastewater. Experiments using a multilevel
complete factorial design were conducted to optimize the removal
effectiveness of COD (Chemical Oxygen Demand) while minimizing the
number of experiments required. To verify the structural characteristics
- elemental composition - and the existence of various functional groups - a
characterization investigation was conducted using X-ray diffractometry
(XRD)
- Fourier Transform InfraRed spectroscopy (FTIR) - Scanning
Electron Microscopy (SEM)
- and Brunner–Emmett–Teller (BET). Batch
experimental trails were operated to determine the optimum adsorpant
material
- its optimum dose - as well as the other operational parameters - such as solution pH - inoculation concentration - and their interactions
during COD removal and Biogas production were investigated. The
maximum removal of COD (99.63%) and the biogas production (5.16 mL
biogas/mg COD removed) of Rice Straw Biochar (RSB) were at pH value
- biochar dose - and buffalo sludge dose concentration were equal to 8 - 2 g/L -
IV
and 0%
- respectively. Commercial Activated Carbon (AC) has achieved
maximum removal of COD (95.55%)
- and the biogas production (6.08 mL
biogas/mg COD removed) at pH
- biochar dose - and buffalo sludge dose
concentration were equal to 5
- 2 g/L - and 0% - respectively. The maximum
removal of COD (98.88%) and the biogas production (4.08 mL biogas/mg
COD removed) of Phragmites australis Biochar (PaB) were at pH
- biochar dose - and buffalo sludge dose concentration were equal to 5 - 2 g/L - and 0% - respectively. These results revealed that rice straw biochar can be
used as an effective and low-cost adsorbent to remove COD from
wastewater. The surface properties of rice straw biochar substantially
affect its capability of removing metal ions from wastewater
- and fourier
transform infrared spectroscopy (FTIR) spectroscopy is a great tool to
observe this surface composition. Two identical pilot-scale models
simulating “Up-flow Anaerobic Sludge Blanket” reactors (UASBs) were
built and operated continuously within the work frame of the present work
to investigate its performance and efficiency in treating buffalo
wastewater treatment. The effect of supporting media on the UASB
efficiency will be also invistgated at the field. The two UASB reactors
were operated under the same operational conditions and scenario
- the
reators operated at HRT equals 4hr and ambiaint temperature. Both R4
(conventional UASB) and R3 (modified UASB) were fed by settleled
wastewater . The condutive media was not added to R3 at the beigning. After the start-up
- the modified UASB reactor (R3) was inoculated with
rice straw biochar through an inclined pipe. Samples were collected and
analyzed periodically twice weekly. The results indicated that; For the
conventional reactor
- the maximum removal efficiency of COD - TSS - TDS - Color - and Turbidity was 79.89% - 74.04% - 80.11% - 72.72% - and
75.70%
- respectively. Cumulative biogas production reached 0.028 mL

Thesis 2022.
تعتبر مشكلة االحتباس الحراري وارتفاع درجة حرارة األرض دافع للعمل في مجال الطاقة المتجددة وذلك بتقليل
استخدام الوقود الحفري. يعتبر الطاقة الشمسية والهيدروجين من مصادر الطاقة النظيفة والمتجددة والتي لها تأثير في
فروع مختلفة مثل التغييرات المناخية، البناء، وسائل النقل. تمثل الطاقة الشمسية مصدرا اساسيا للطاقة المتجددة والتي
تستخدم في توليد الطاقة الكهربية من خالل األنظمة الشمسية حيث تتأثر الطاقة الكهربية المولدة من االنظمة الشمسية
بعوامل الطقس والتغييرات غير المتوقعة للعوامل المناخية والتي تؤثر بالسلب على استمرارية وانتاجية هذه الطاقة
الكهربية لذلك فإن التوقع المسبق والدقيق إلنتاج الطاقة الكهربية ضرورة حتمية لضمان ثبات انتاجية الطاقة المتوقعة.
بناء على ذلك هذه الدراسة تتناول دراسة التنبؤ للطاقة الكهربية المتولدة من الخاليا الشمسية والمستخدمة في توقع انتاج
الهيدروجين في منطقة االسماعيلية – جمهورية مصر العربية.
تم تجميع بيانات الطقس لمحافظة االسماعيلية خالل الفترة من 2010 إلى2016 من خالل قاعدة البيانات لموقع نظام
المعلومات الضوئية والجغرافية للمفوضية األوروبية باالتحاد األوروبي، والتي تم استخدامها على النحو التالي:
بيئة البرمجة تتكون من لغة البايثون كلغة برمجة في دراسة وتجهيز البيانات وايضا في بناء نماذج تعليم اآللة. استخدام
جوبيتر نوت من خالل جوجل كوالب لتنفيذ جميع العمليات وذلك تحت بيئة منصة جوجل السحابية لتقليل زمن التنفيذ.
تم بناء واختبار ثالثة نماذج من خوارزميات تعليم اآللة لتنفيذ عملية التنبؤ بالطاقة الكهربية ولتوليد الهيدروجين وهم:
الغابة العشوائية، فيسبوك بروفت وشبكة الذاكرة ذات المدى القصير والطويل.تم عمل مقارنة بين الخوارزميات الثالثة
بناء على مقياس األداء والتي اثبتت تفوق الفيسبوك بروفت كأفضل أداء بنسبة 45.93 %ونسبة خطأ 76.10 وذلك
في عملية التنبؤ بالطاقة الكهربية المولدة من النظام الشمسي باإلضافة الى استخدامه في التنبؤ بالهيدروجين كل ساعة
3 بمتوسط يومي لإلنتاج في شهر مايو يبلغ 76.203×10
2 كجم /كم
كوقود متجدد ونظيف

Articles 2024.
Vol. 96, No. 1 (June 2024) /


من 766
 







Powered by Future Library Software.All rights reserved © CITC - Mansoura University. Sponsored by Mansoura University Privacy Policy