Browsing by Author "Akyol, K."

Now showing 1 - 7 of 7

An ensemble approach for classification of tympanic membrane conditions using soft voting classifier
(2024.01.01) Akyol, K.; Uçar, E.; Atila, U.; Uçar, M.
Otitis media is a medical concept that represents a range of inflammatory middle ear disorders. The high costs of medical devices utilized by field experts to diagnose the disease relevant to otitis media prevent the widespread use of these devices. This makes it difficult for field experts to make an accurate diagnosis and increases subjectivity in diagnosing the disease. To solve these problems, there is a need to develop computer-aided middle ear disease diagnosis systems. In this study, a deep learning-based approach is proposed for the detection of OM disease to meet this emerging need. This approach is the first that addresses the performance of a voting ensemble framework that uses Inception V3, DenseNet 121, VGG16, MobileNet, and EfficientNet B0 pre-trained DL models. All pre-trained CNN models used in the proposed approach were trained using the Public Ear Imagery dataset, which has a total of 880 otoscopy images, including different eardrum cases such as normal, earwax plug, myringosclerosis, and chronic otitis media. The prediction results of these models were evaluated with voting approaches to increase the overall prediction accuracy. In this context, the performances of both soft and hard voting ensembles were examined. Soft voting ensemble framework achieved highest performance in experiments with 98.8% accuracy, 97.5% sensitivity, and 99.1% specificity. Our proposed model achieved the highest classification performance so far in the current dataset. The results reveal that our voting ensemble-based DL approach showed quite high performance for the diagnosis of middle ear disease. In clinical applications, this approach can provide a preliminary diagnosis of the patient's condition just before field experts make a diagnosis on otoscopic images. Thus, our proposed approach can help field experts to diagnose the disease quickly and accurately. In this way, clinicians can make the final diagnosis by integrating automatic diagnostic prediction with their experience.
An ensemble approach for classification of tympanic membrane conditions using soft voting classifier
(Springer, 2024) Akyol, K.; Uçar, E.; Atila, Ü.; Uçar, M.
Otitis media is a medical concept that represents a range of inflammatory middle ear disorders. The high costs of medical devices utilized by field experts to diagnose the disease relevant to otitis media prevent the widespread use of these devices. This makes it difficult for field experts to make an accurate diagnosis and increases subjectivity in diagnosing the disease. To solve these problems, there is a need to develop computer-aided middle ear disease diagnosis systems. In this study, a deep learning-based approach is proposed for the detection of OM disease to meet this emerging need. This approach is the first that addresses the performance of a voting ensemble framework that uses Inception V3, DenseNet 121, VGG16, MobileNet, and EfficientNet B0 pre-trained DL models. All pre-trained CNN models used in the proposed approach were trained using the Public Ear Imagery dataset, which has a total of 880 otoscopy images, including different eardrum cases such as normal, earwax plug, myringosclerosis, and chronic otitis media. The prediction results of these models were evaluated with voting approaches to increase the overall prediction accuracy. In this context, the performances of both soft and hard voting ensembles were examined. Soft voting ensemble framework achieved highest performance in experiments with 98.8% accuracy, 97.5% sensitivity, and 99.1% specificity. Our proposed model achieved the highest classification performance so far in the current dataset. The results reveal that our voting ensemble-based DL approach showed quite high performance for the diagnosis of middle ear disease. In clinical applications, this approach can provide a preliminary diagnosis of the patient's condition just before field experts make a diagnosis on otoscopic images. Thus, our proposed approach can help field experts to diagnose the disease quickly and accurately. In this way, clinicians can make the final diagnosis by integrating automatic diagnostic prediction with their experience.
Comprehensive comparison of modified deep convolutional neural networks for automated detection of external and middle ear conditions
(2024.01.01) Akyol, K.
Otitis media disease, a frequent childhood ailment, could have severe repercussions, including mortality. This disease induces permanent hearing loss, commonly seen in developing countries with limited medical resources. It is estimated that approximately 21,000 people worldwide die from reasons related to this disease each year. The main aim of this study is to develop a model capable of detecting external and middle ear conditions. Experiments were conducted to find the most successful model among the modified deep convolutional neural networks within two scenarios. According to the results, the modified EfficientNetB7 model could detect normal, chronic otitis media, earwax, myringosclerosis cases with high accuracy in Scenario 2. This model offers average values of 99.94% accuracy, 99.86% sensitivity, 99.95% specificity, and 99.86% precision. An expert system based on this model is expected to provide a second opinion to doctors in detecting external and middle ear conditions, particularly in primary healthcare institutions and hospitals lacking field specialists.
Comprehensive comparison of modified deep convolutional neural networks for automated detection of external and middle ear conditions
(Springer Science and Business Media Deutschland GmbH, 2024) Akyol, K.
Otitis media disease, a frequent childhood ailment, could have severe repercussions, including mortality. This disease induces permanent hearing loss, commonly seen in developing countries with limited medical resources. It is estimated that approximately 21,000 people worldwide die from reasons related to this disease each year. The main aim of this study is to develop a model capable of detecting external and middle ear conditions. Experiments were conducted to find the most successful model among the modified deep convolutional neural networks within two scenarios. According to the results, the modified EfficientNetB7 model could detect normal, chronic otitis media, earwax, myringosclerosis cases with high accuracy in Scenario 2. This model offers average values of 99.94% accuracy, 99.86% sensitivity, 99.95% specificity, and 99.86% precision. An expert system based on this model is expected to provide a second opinion to doctors in detecting external and middle ear conditions, particularly in primary healthcare institutions and hospitals lacking field specialists.
ETSVF-COVID19: efficient two-stage voting framework for COVID-19 detection
(Springer Science and Business Media Deutschland GmbH, 2024) Akyol, K.
COVID-19 disease, an outbreak in the spring of 2020, reached very alarming dimensions for humankind due to many infected patients during the pandemic and the heavy workload of healthcare workers. Even though we have been saved from the darkness of COVID-19 after about three years, the importance of computer-aided automated systems that support field experts in the fight against with global threat has emerged once again. This study proposes a two-stage voting framework called ETSVF-COVID19 that includes transformer-based deep features and a machine learning approach for detecting COVID-19 disease. ETSVF-COVID19, which offers 99.2% and 98.56% accuracies on computed tomography scan and X-radiation images, respectively, could compete with the related works in the literature. The findings demonstrate that this framework could assist field experts in making informed decisions while diagnosing COVID-19 with its fast and accurate classification role. Moreover, ETSVF-COVID19 could screen for chest infections and help physicians, particularly in areas where test kits and specialist doctors are inadequate.
Robust stacking-based ensemble learning model for forest fire detection
(2023.01.01) Akyol, K.
Forests reduce soil erosion and prevent drought, wind, and other natural disasters. Forest fires, which threaten millions of hectares of forest area yearly, destroy these precious resources. This study aims to design a deep learning model with high accuracy to intervene in forest fires at an early stage. A stacked-based ensemble learning model is proposed for fire detection from forest landscape images in this context. This model offers high test accuracies of 97.37%, 95.79%, and 95.79% with hold-out validation, fivefold cross-validation, and tenfold cross-validation experiments, respectively. The artificial intelligence model developed in this study could be used in real-time systems run on unmanned aerial vehicles to prevent potential disasters in forest areas.Graphical abstractBlock diagram of the proposed model.
Robust stacking-based ensemble learning model for forest fire detection
(2023.01.01) Akyol, K.
Forests reduce soil erosion and prevent drought, wind, and other natural disasters. Forest fres, which threaten millions of hectares of forest area yearly, destroy these precious resources. This study aims to design a deep learning model with high accuracy to intervene in forest fres at an early stage. A stacked-based ensemble learning model is proposed for fre detection from forest landscape images in this context. This model ofers high test accuracies of 97.37%, 95.79%, and 95.79% with hold-out validation, fvefold cross-validation, and tenfold cross-validation experiments, respectively. The artifcial intelligence model developed in this study could be used in real-time systems run on unmanned aerial vehicles to prevent potential disasters in forest areas. Graphical abstract Block diagram of the proposed model