Brain tumor segmentation from Magnetic Resonance Images (MRI) presents significant challenges due to the complex nature of brain tumor tissues. This complexity poses a significant challenge in distinguishing tumor tissues from healthy tissues, particularly when radiologists rely on manual segmentation. Reliable and accurate segmentation is crucial for effective tumor grading and treatment planning. In this paper, we proposed a novel ensemble dual-modality approach for 3D brain tumor segmentation using MRI. Initially, individual U-Net models are trained and evaluated on single MRI modalities (T1, T2, T1ce, and FLAIR) to establish each modality's performance. Subsequently, we trained U-net models using combinations of the best-performing modalities to exploit the complementary information and improve segmentation accuracy. Finally, we introduced the ensemble dual-modality by combining the two best-performing pre-trained dual-modalities models to enhance segmentation performance. Experimental results show that the proposed model enhanced the segmentation result and achieved a Dice Coefficient of 97.73% and a Mean IoU of 60.08%. The results illustrate that the ensemble dual-modality approach outperforms single-modality and dual-modality models. Grad-CAM visualizations are implemented, generating heat maps that highlight tumor regions and provide useful information to clinicians about how the model made the decision, increasing their confidence in using deep learning-based systems. Our code publicly available at: https://github.com/Ahmeed-Suliman-Farhan/Ensemble-Dual-Modality-Approach.

In recent years, deep learning’s use in medical imaging has grown exponentially. However, one of the biggest problems with training deep learning models is the unavailability of large amounts of data, which leads to overfitting. Collecting large quantities of labelled medical images is expensive, time-consuming, and depends on specialists’ availability. In this paper, we proposed a novel method namely Oriented Combination MRI (OCMRI) for augmenting brain MRI dataset. The proposed method helps CNN models overcome overfitting and class imbalance problems by combining Brain MRI images to generate new images. The image fusion is performed by selecting two images of the same tumor class if the Mean Squared Error (MSE) between these two images is greater than threshold 1 and lower than threshold 2. Both thresholds are variable, initially set by the user and automatically fine-tuned by the algorithm to control the number of images produced for each class, thus helping to address the data imbalance problem. The proposed approach was evaluated by training and testing the PRCnet model on four publicly available datasets before and after applying the proposed method to the datasets. Where the classification accuracy without data augmentation was 85.19% for dataset A, 90.12% for dataset B, 94.77% for dataset C, and 90% for dataset D respectively. After adding the synthetic data; the accuracy improved to 92.7% for dataset A, 95.37% for dataset B, 96.51% for dataset C and 98% for dataset D respectively.

Brain tumors are the most prevalent and life-threatening cancer; an early and accurate diagnosis of brain tumors increases the chances of patient survival and treatment planning. However, manual tumor detection is a complex, cumbersome and time-consuming task and is prone to errors, which relies on the radiologist’s experience. As a result, the development of an accurate and automatic tumor detection system is critical. In this paper, we proposed a new model called Parallel Residual Convolutional Network (PRCnet) model to classify brain tumors from Magnetic Resonance Imaging. The PCRnet model uses several techniques (such as filters of different sizes with parallel layers, connections between layers, batch normalization layer, and ReLU) and dropout layer to overcome the over-fitting problem, for achieving accurate and automatic classification of brain tumors. Our methodology used data augmentation techniques such as rotation, flipping, and scaling. These enhanced the diversity and quantity of the training dataset, contributing significantly to the model’s improved performance. The PRCnet model is trained and tested on two different datasets and obtained an accuracy of 94.77% and 97.1% for dataset A and dataset B, respectively which is way better as compared to the state-of-the-art models. Our PRCnet code publicly available at: https://github.com/Ahmeed-Suliman-Farhan/PRCnet-Model

Brain tumors are the most prevalent and life-threatening cancer; an early and accurate diagnosis of brain tumors increases the chances of patient survival and treatment planning. However, manual tumor detection is a complex, cumbersome and time-consuming task and is prone to errors, which relies on the radiologist’s experience. As a result, the development of accurate and automatic system for tumor detection is critical. In this paper, we proposed a new model called Parallel Residual Convolutional Network (PRCnet) model to classify brain tumors from Magnetic Resonance Imaging. The PCRnet model uses several techniques (such as filters of different sizes with parallel layers, connections between layers, batch normalization layer, and ReLU) and dropout layer to overcome the over-fitting problem, for achieving accurate and automatic classification of brain tumors. The PRCnet model is trained and tested on two different datasets and obtained an accuracy of 94.77% and 97.1% for dataset A and dataset B, respectively which is way better as compared to the state-of-the-art models.

Brain tumor segmentation from Magnetic Resonance Images (MRI) presents significant challenges due to the complex nature of brain tumor tissues. This complexity makes distinguishing tumor tissues from healthy tissues difficult, mainly when radiologists perform manual segmentation. Reliable and accurate segmentation is crucial for effective tumor grading and treatment planning. In this paper, we proposed a novel ensemble dual-modality approach for 3D brain tumor segmentation using MRI. Initially, individual U-Net models are trained and evaluated on single MRI modalities (T1, T2, T1ce, and FLAIR) to establish each modality’s performance. Subsequently, we trained U-net models using combinations of the best-performing modalities to exploit the complementary information and improve segmentation accuracy. Finally, we suggested the ensemble dual-modality by combining the best-performing two pre-trained dual-modalities models to enhance segmentation performance. Experimental results show that the proposed model enhanced the segmentation result and achieved a Dice Coefficient of 97.73% and a Mean IoU of 60.08% on the BraTS2020 dataset. The results illustrate that the ensemble dual-modality approach outperforms single-modality and dual-modality models. This study shows that ensemble dual-modality models can help improve the accuracy and dependability of brain tumor segmentation based on MRI. Our code publicly available at: https://github.com/Ahmeed-Suliman-Farhan/Ensemble-Dual-Modality-Approach

Chest X-ray (CXR) interpretation is essential for diagnosing pulmonary diseases, yet manual reading remains slow and prone to human error, especially in high-volume or resource-limited settings. To address delayed diagnoses and improve clinical efficiency, this study introduces (HyRA-CXR), a hybrid residual–attention convolutional neural network for automated CXR classification. The proposed model integrates residual blocks to enhance gradient stability and dual attention mechanisms to focus on significant lung regions. Experiments are conducted on the publicly available Lung X-Ray Image dataset. Hyperparameters are optimized using KerasTuner as well as model evaluation is carried out using five-fold stratified cross-validation. HyRA-CXR achieved an average accuracy of 90.39%, outperforming DenseNet121 (89.38%) and Xception (89.12%) models. Also, the experimental results confirmed that both residual and attention modules contribute, as removing either reduced accuracy below 90%. Overall, the proposed model achieves competitive accuracy with maintaining a compact architecture (0.52M parameters), indicating its suitability for deployment in resource-constrained settings. Our code is publicly available at: https://github.com/Ahmeed-Suliman-Farhan/HyRA-CXR-A-Hybrid-Residual-Attention-Deep-Network-for-Chest-X-Ray-Classification .