Applying Transfer Learning Using DenseNet121 in Radiographic Image Classification
Nahla Saeed Saad Aldeen
Yosser Mohammad Marwan Atassi
Faculty of Information Technology || AlBaath University || Syria
DOI ” type=”big” color=”orange”] DOI[/button]
Tab title The study aims to apply one of the fully connected convolutional neural networks, DenseNet121 network, to a data sample that includes a large group of radiographs through transfer learning technology. Radiography technology is a very important technique in the medical community to detect diseases and abnormalities that may be present, but the interpretation of these images may take a long time and it is subject to error by radiologists who are exposed to external practical factors (such as fatigue resulting from working for long hours, or exhaustion, or thinking about other life matters). To assist radiologists, we have worked on developing a diagnostic model with the help of a deep learning technique to classify radiographic images into two classes: (Normal and Abnormal images), by transferring the selected deep convolutional neural network between a large group of available networks that we studied on the basis of the regions that possibly abnormalities provided by the radiologists for the study sample. We also studied the feasibility of using the well-known VGG16 model on the same data sample and its performance through transfer learning technology and compared its results with the results of the DenseNet121 network. At the end of the research, we obtained a set of good results, which achieved a high diagnostic accuracy of 87.5% in some studied cases, using the DenseNet121 network model, which is considered satisfactory results in the case studied compared to the performance of other models. As for the VGG16 model, it did not give any of the satisfactory results in this field, the accuracy of the classification did not exceed 55% in most cases, and in only two cases it reached about 60% and 62%. The model presented during the research – DenseNet121 model – can be used in the diagnostic process and help in obtaining accurate results in terms of diagnostic results. As for the VGG16 model, it does not give satisfactory results according to the results also obtained during the research, so it is excluded in this type of applications. Keywords: Transfer learning, Machine Learning, Radiographic Image Classification, DenseNet121, Convolutional Neural Networks.
Tab title
The study aims to apply one of the fully connected convolutional neural networks, DenseNet121 network, to a data sample that includes a large group of radiographs through transfer learning technology. Radiography technology is a very important technique in the medical community to detect diseases and abnormalities that may be present, but the interpretation of these images may take a long time and it is subject to error by radiologists who are exposed to external practical factors (such as fatigue resulting from working for long hours, or exhaustion, or thinking about other life matters). To assist radiologists, we have worked on developing a diagnostic model with the help of a deep learning technique to classify radiographic images into two classes: (Normal and Abnormal images), by transferring the selected deep convolutional neural network between a large group of available networks that we studied on the basis of the regions that possibly abnormalities provided by the radiologists for the study sample. We also studied the feasibility of using the well-known VGG16 model on the same data sample and its performance through transfer learning technology and compared its results with the results of the DenseNet121 network. At the end of the research, we obtained a set of good results, which achieved a high diagnostic accuracy of 87.5% in some studied cases, using the DenseNet121 network model, which is considered satisfactory results in the case studied compared to the performance of other models. As for the VGG16 model, it did not give any of the satisfactory results in this field, the accuracy of the classification did not exceed 55% in most cases, and in only two cases it reached about 60% and 62%. The model presented during the research – DenseNet121 model – can be used in the diagnostic process and help in obtaining accurate results in terms of diagnostic results. As for the VGG16 model, it does not give satisfactory results according to the results also obtained during the research, so it is excluded in this type of applications. Keywords: Transfer learning, Machine Learning, Radiographic Image Classification, DenseNet121, Convolutional Neural Networks.
تطبيق التعلم بالنقل باستخدام شبكة DenseNet121 في تصنيف الصور الشعاعية
نهله سعيد سعد الدين
يسر محمد مروان الأتاسي
كلية الهندسة المعلوماتية || جامعة البعث || سوريا
Tab title
هدفت الدراسة إلى تطبيق إحدى الشبكات العصبونية التلافيفية المتصلة كلياً وهي شبكة DenseNet121 على عينة بيانات تضم مجموعة كبيرة من الصور الشعاعية من خلال تقنية التعلم بالنقل. تعد تقنية التصوير الشعاعي تقنية مهمة جداً في المجتمع الطبي للكشف عن الأمراض والتشوهات التي من الممكن وجودها، إلا أن تفسير هذه الصور قد يستغرق وقتاً طويلاً كما أنه معرّض للخطأ من قبل أخصائيي الأشعة الذين يتعرضون للعوامل العملية الخارجية (مثل التعب الناتج عن العمل لساعات طويلة، أو الإرهاق، أو التفكير بأمور الحياة الأخرى). ولمساعدة أخصائيي الأشعة، قمنا بالعمل على تطوير نموذج تشخيص بمساعدة إحدى تقنيات التعلم العميق لتصنيف الصور الشعاعية إلى تصنيفين: (صور طبيعية وصور غير طبيعية)، وذلك من خلال نقل الشبكة العصبونية التلافيفية العميقة المختارة بين مجموعة شبكات كبيرة متوفرة قمنا بدراستها على أساس المناطق التي من المحتمل أن تكون غير طبيعية والتي يوفرها اختصاصيو الأشعة لعينة الدراسة. كما قمنا بدراسة جدوى استخدام نموذج VGG16 المعروف على نفس عينة البيانات ودراسة أدائه أيضاَ من خلال تقنية التعلم بالنقل ومقارنة نتائجه مع نتائج شبكة DenseNet121. حصلنا في نهاية البحث على مجموعة من النتائج الجيدة، والتي حققت دقة تشخيص عالية بلغت 87.5% في بعض الحالات المدروسة بالاستفادة من نموذج شبكة DenseNet121 والتي تعتبر نتائج مرضية في الحالة المدروسة مقارنة مع أداء النماذج الأخرى، أما نموذج VGG16 فلم يعطي أي من النتائج المرضية في هذا المجال المدروس ولم تتجاوز دقة التصنيف 55% في معظم الحالات، وصلت في حالتين فقط إلى حوالي 60% و62%. يمكن الاستفادة من النموذج الذي تم تقديمه خلال البحث- نموذج DenseNet121 – في عملية التشخيص المطروحة والمساعدة بالحصول على نتائج دقيقة من ناحية نتائج التشخيص، أما نموذج VGG16 فلا يعطي نتائج مرضية وفق ما تم الحصول عليه من نتائج أيضا خلال البحث، لذلك يتم استبعاده في هذا النوع من التطبيقات. الكلمات المفتاحية: التعلم بالنقل، تعلم الآلة، تصنيف الصور الشعاعية، DenseNet121 , شبكة عصبية تلافيفية.