Detection of Thymoma Disease Using mRMR Feature Selection and Transformer Models.

BACKGROUND: Thymoma is a tumor that originates in the thymus gland, a part of the human body located behind the breastbone. It is a malignant disease that is rare in children but more common in adults and usually does not spread outside the thymus. The exact cause of thymic disease is not known, but it is thought to be more common in people infected with the EBV virus at an early age. Various surgical methods are used in clinical settings to treat thymoma. Expert opinion is very important in the diagnosis of the disease. Recently, next-generation technologies have become increasingly important in disease detection. Today's early detection systems already use transformer models that are open to technological advances. METHODS: What makes this study different is the use of transformer models instead of traditional deep learning models. The data used in this study were obtained from patients undergoing treatment at Firat University, Department of Thoracic Surgery. The dataset consisted of two types of classes: thymoma disease images and non-thymoma disease images. The proposed approach consists of preprocessing, model training, feature extraction, feature set fusion between models, efficient feature selection, and classification. In the preprocessing step, unnecessary regions of the images were cropped, and the region of interest (ROI) technique was applied. Four types of transformer models (Deit3, Maxvit, Swin, and ViT) were used for model training. As a result of the training of the models, the feature sets obtained from the best three models were merged between the models (Deit3 and Swin, Deit3 and ViT, Deit3 and ViT, Swin and ViT, and Deit3 and Swin and ViT). The combined feature set of the model (Deit3 and ViT) that gave the best performance with fewer features was analyzed using the mRMR feature selection method. The SVM method was used in the classification process. RESULTS: With the mRMR feature selection method, 100% overall accuracy was achieved with feature sets containing fewer features. The cross-validation technique was used to verify the overall accuracy of the proposed approach and 99.22% overall accuracy was achieved in the analysis with this technique. CONCLUSIONS: These findings emphasize the added value of the proposed approach in the detection of thymoma.

Detection of retinopathy disease using morphological gradient and segmentation approaches in fundus images.

BACKGROUND AND OBJECTIVE: Diabetes-related cases can cause glaucoma, cataracts, optic neuritis, paralysis of the eye muscles, or various retinal damages over time. Diabetic retinopathy is the most common form of blindness that occurs with diabetes. Diabetic retinopathy is a disease that occurs when the blood vessels in the retina of the eye become damaged, leading to loss of vision in advanced stages. This disease can occur in any diabetic patient, and the most important factor in treating the disease is early diagnosis. Nowadays, deep learning models and machine learning methods, which are open to technological developments, are already used in early diagnosis systems. In this study, two publicly available datasets were used. The datasets consist of five types according to the severity of diabetic retinopathy. The objectives of the proposed approach in diabetic retinopathy detection are to positively contribute to the performance of CNN models by processing fundus images through preprocessing steps (morphological gradient and segmentation approaches). The other goal is to detect efficient sets from type-based activation sets obtained from CNN models using Atom Search Optimization method and increase the classification success. METHODS: The proposed approach consists of three steps. In the first step, the Morphological Gradient method is used to prevent parasitism in each image, and the ocular vessels in fundus images are extracted using the segmentation method. In the second step, the datasets are trained with transfer learning models and the activations for each class type in the last fully connected layers of these models are extracted. In the last step, the Atom Search optimization method is used, and the most dominant activation class is selected from the extracted activations on a class basis. RESULTS: When classified by the severity of diabetic retinopathy, an overall accuracy of 99.59% was achieved for dataset #1 and 99.81% for dataset #2. CONCLUSIONS: In this study, it was found that the overall accuracy achieved with the proposed approach increased. To achieve this increase, the application of preprocessing steps and the selection of the dominant activation sets from the deep learning models were implemented using the Atom Search optimization method.

Detection of COVID-19 findings by the local interpretable model-agnostic explanations method of types-based activations extracted from CNNs.

Covid-19 is a disease that affects the upper and lower respiratory tract and has fatal consequences in individuals. Early diagnosis of COVID-19 disease is important. Datasets used in this study were collected from hospitals in Istanbul. The first dataset consists of COVID-19, viral pneumonia, and bacterial pneumonia types. The second dataset consists of the following findings of COVID-19: ground glass opacity, ground glass opacity, and nodule, crazy paving pattern, consolidation, consolidation, and ground glass. The approach suggested in this paper is based on artificial intelligence. The proposed approach consists of three steps. As a first step, preprocessing was applied and, in this step, the Fourier Transform and Gradient-weighted Class Activation Mapping methods were applied to the input images together. In the second step, type-based activation sets were created with three different ResNet models before the Softmax method. In the third step, the best type-based activations were selected among the CNN models using the local interpretable model-agnostic explanations method and re-classified with the Softmax method. An overall accuracy success of 99.15% was achieved with the proposed approach in the dataset containing three types of image sets. In the dataset consisting of COVID-19 findings, an overall accuracy success of 99.62% was achieved with the recommended approach.

Disease type detection in lung and colon cancer images using the complement approach of inefficient sets.

Lung and colon cancers are deadly diseases that can develop simultaneously in organs and adversely affect human life in some special cases. Although the frequency of simultaneous occurrence of these two types of cancer is unlikely, there is a high probability of metastasis between the two organs if not diagnosed early. Traditionally, specialists have to go through a lengthy and complicated process to examine histopathological images and diagnose cancer cases; yet, it is now possible to achieve this process faster with the available technological possibilities. In this study, artificial intelligence-supported model and optimization methods were used to realize the classification of lung and colon cancers' histopathological images. The used dataset has five classes of histopathological images consisting of two colon cancer classes and three lung cancer classes. In the proposed approach, the image classes were trained from scratch with the DarkNet-19 model, which is one of the deep learning models. In the feature set extracted from the DarkNet-19 model, selection of the inefficient features was performed by using Equilibrium and Manta Ray Foraging optimization algorithms. Then, the set containing the inefficient features was distinguished from the rest of the set features, creating an efficient feature set (complementary rule insets). The efficient features obtained by the two used optimization algorithms were combined and classified with the Support Vector Machine (SVM) method. The overall accuracy rate obtained in the classification process was 99.69%. Based on the outcomes of this study, it has been observed that using the complementary method together with some optimization methods improved the classification performance of the dataset.

Detection of segmented uterine cancer images by Hotspot Detection method using deep learning models, Pigeon-Inspired Optimization, types-based dominant activation selection approaches.

Uterine cancer consists of cells of a layer that forms the inside of the uterus. Sometimes, as a result of abnormal growth of normal cells, it can damage the surrounding tissues and cause the formation of cancerous cells. In the USA, according to the projections for 2021, approximately 66 thousand new cases of uterine cancer will be detected and approximately 13 thousand of these cancer patients are expected to die from uterine cancer. Early diagnosis of cancer is important. Recently, artificial intelligence-based technologies have been used in the diagnosis and treatment processes of various diseases. In this study, five categories of datasets including normal, abnormal, and benign cells were used. The dataset consists of cellular images and is publicly available. The proposed approach consists of three steps. In the first step, the Hotspot method was used to detect the tumor cells in the images. In the second step, tumor cells that were brought to the fore by segmentation were trained by deep learning models, and activation sets of five types from each deep learning model were created. In the last step, the best activation sets were selected among the activation sets obtained by deep learning models of each type (for five dataset types). Pigeon-Inspired Optimization was used for this selection. Thus, the activation sets with the best performance of the five types were classified by the Softmax method. The overall accuracy success achieved with the approach suggested as a result of the classification was 99.65%.

Tumor type detection in brain MR images of the deep model developed using hypercolumn technique, attention modules, and residual blocks.

Brain cancer is a disease caused by the growth of abnormal aggressive cells in the brain outside of normal cells. Symptoms and diagnosis of brain cancer cases are producing more accurate results day by day in parallel with the development of technological opportunities. In this study, a deep learning model called BrainMRNet which is developed for mass detection in open-source brain magnetic resonance images was used. The BrainMRNet model includes three processing steps: attention modules, the hypercolumn technique, and residual blocks. To demonstrate the accuracy of the proposed model, three types of tumor data leading to brain cancer were examined in this study: glioma, meningioma, and pituitary. In addition, a segmentation method was proposed, which additionally determines in which lobe area of the brain the two classes of tumors that cause brain cancer are more concentrated. The classification accuracy rates were performed in the study; it was 98.18% in glioma tumor, 96.73% in meningioma tumor, and 98.18% in pituitary tumor. At the end of the experiment, using the subset of glioma and meningioma tumor images, it was determined which at brain lobe the tumor region was seen, and 100% success was achieved in the analysis of this determination. In this study, a hybrid deep learning model is presented to determine the detection of the brain tumor. In addition, open-source software was proposed, which statistically found in which lobe region of the human brain the brain tumor occurred. The methods applied and tested in the experiments have shown promising results with a high level of accuracy, precision, and specificity. These results demonstrate the availability of the proposed approach in clinical settings to support the medical decision regarding brain tumor detection.

COVID-19 detection using deep learning models to exploit Social Mimic Optimization and structured chest X-ray images using fuzzy color and stacking approaches.

Coronavirus causes a wide variety of respiratory infections and it is an RNA-type virus that can infect both humans and animal species. It often causes pneumonia in humans. Artificial intelligence models have been helpful for successful analyses in the biomedical field. In this study, Coronavirus was detected using a deep learning model, which is a sub-branch of artificial intelligence. Our dataset consists of three classes namely: coronavirus, pneumonia, and normal X-ray imagery. In this study, the data classes were restructured using the Fuzzy Color technique as a preprocessing step and the images that were structured with the original images were stacked. In the next step, the stacked dataset was trained with deep learning models (MobileNetV2, SqueezeNet) and the feature sets obtained by the models were processed using the Social Mimic optimization method. Thereafter, efficient features were combined and classified using Support Vector Machines (SVM). The overall classification rate obtained with the proposed approach was 99.27%. With the proposed approach in this study, it is evident that the model can efficiently contribute to the detection of COVID-19 disease.

Application of breast cancer diagnosis based on a combination of convolutional neural networks, ridge regression and linear discriminant analysis using invasive breast cancer images processed with autoencoders.

Invasive ductal carcinoma cancer, which invades the breast tissues by destroying the milk channels, is the most common type of breast cancer in women. Approximately, 80% of breast cancer patients have invasive ductal carcinoma and roughly 66.6% of these patients are older than 55 years. This situation points out a powerful relationship between the type of breast cancer and progressed woman age. In this study, the classification of invasive ductal carcinoma breast cancer is performed by using deep learning models, which is the sub-branch of artificial intelligence. In this scope, convolutional neural network models and the autoencoder network model are combined. In the experiment, the dataset was reconstructed by processing with the autoencoder model. The discriminative features obtained from convolutional neural network models were utilized. As a result, the most efficient features were determined by using the ridge regression method, and classification was performed using linear discriminant analysis. The best success rate of classification was achieved as 98.59%. Consequently, the proposed approach can be admitted as a successful model in the classification.

BrainMRNet: Brain tumor detection using magnetic resonance images with a novel convolutional neural network model.

A brain tumor is a mass that grows unevenly in the brain and directly affects human life. This mass occurs spontaneously because of the tissues surrounding the brain or the skull. Surgical methods are generally preferred for the treatment of the brain tumor. Recently, models of deep learning in the diagnosis and treatment of diseases in the biomedical field have gained intense interest. In this study, we propose a new convolutional neural network model named BrainMRNet. This architecture is built on attention modules and hypercolumn technique; it has a residual network. Firstly, image is preprocessed in BrainMRNet. Then, this step is transferred to attention modules using image augmentation techniques for each image. Attention modules select important areas of the image and the image is transferred to convolutional layers. One of the most important techniques that the BrainMRNet model uses in the convolutional layers is hypercolumn. With the help of this technique, the features extracted from each layer of the BrainMRNet model are retained by the array structure in the last layer. The aim is to select the best and the most efficient features among the features maintained in the array. Accessible magnetic resonance images were used to detect brain tumor with the BrainMRNet model. BrainMRNet model is more successful than the pre-trained convolutional neural network models (AlexNet, GoogleNet, VGG-16) used in this study. The classification success achieved with the BrainMRNet model was 96.05%.