Breast cancer is a prevalent and life-threatening disease that affects women globally. Early detection and access to top-notch treatment are crucial in preventing fatalities from this condition. However, manual breast histopathology image analysis is time-consuming and prone to errors. This study proposed a hybrid deep learning model (CNN+EfficientNetV2B3). The proposed approach utilizes convolutional neural networks (CNNs) for the identification of positive invasive ductal carcinoma (IDC) and negative (non-IDC) tissue using whole slide images (WSIs), which use pre-trained models to classify breast cancer in images, supporting pathologists in making more accurate diagnoses. The proposed model demonstrates outstanding performance with an accuracy of 96.3%, precision of 93.4%, recall of 86.4%, F1-score of 89.7%, Matthew’s correlation coefficient (MCC) of 87.6%, the Area Under the Curve (AUC) of a Receiver Operating Characteristic (ROC) curve of 97.5%, and the Area Under the Curve of the Precision-Recall Curve (AUPRC) of 96.8%, which outperforms the accuracy achieved by other models. The proposed model was also tested against MobileNet+DenseNet121, MobileNetV2+EfficientNetV2B0, and other deep learning models, proving more powerful than contemporary machine learning and deep learning approaches.
Due to the exponential growth of video data, aided by rapid advancements in multimedia technologies. It became difficult for the user to obtain information from a large video series. The process of providing an abstract of the entire video that includes the most representative frames is known as static video summarization. This method resulted in rapid exploration, indexing, and retrieval of massive video libraries. We propose a framework for static video summary based on a Binary Robust Invariant Scalable Keypoint (BRISK) and bisecting K-means clustering algorithm. The current method effectively recognizes relevant frames using BRISK by extracting keypoints and the descriptors from video sequences. The video frames’ BRISK features are clustered using a bisecting K-means, and the keyframe is determined by selecting the frame that is most near the cluster center. Without applying any clustering parameters, the appropriate clusters number is determined using the silhouette coefficient. Experiments were carried out on a publicly available open video project (OVP) dataset that contained videos of different genres. The proposed method’s effectiveness is compared to existing methods using a variety of evaluation metrics, and the proposed method achieves a trade-off between computational cost and quality.
Early breast cancer diagnosis and detection are very important. It may greatly enhance treatment results and save a life. The absence of early cancer signs makes early identification challenging. Cancer continues to be one of the health subjects that many researchers work to advance. This study proposed a new hybrid model for classifying breast cancer images. The proposed framework consists of preprocessing stage and the proposed two models stage. For the preprocessing, we downsized every image from its original 50x50 to 32x32 pixel size, rotating and flipping all positive images for the Histology Images. The proposed hybrid model consists of a CNN model created from scratch and transfer learning based on EfficientNetB0 (CNN+ EfficientNetB0) to classify Invasive ductal carcinoma (IDC) into benign and malignant. According to tests, the CNN + EfficientNetB0 model has the highest accuracy compared to the other deep learning models. This model achieves 96% accuracy, 95% precision, 82.5% recall, and 88.3% F1- score.
Portable digital devices such as PDAs and camera phones are the easiest and most widely used methods to preserve and collect information. Capturing a document image using this method always has warping issues, especially when capturing pages from a book and rolled-up documents. In this article, we propose an effective method to correct the warping of the captured document image. The proposed method uses a checkerboard calibration pattern to calculate the world and image points. A radial distortion algorithm is used to handle the warping problem based on the computed image and world points. The proposed method obtained an error rate of 3% using a document de-warping dataset (CBDAR 2007). The proposed method achieved a high level of quality compared with other previous methods. Our method fixes the problem of warping in document images acquired with different levels of complexity, such as poor lighting, low quality, and different layouts.
1 July 2025 in IJCI. International Journal of Computers and Information
Data augmentation is a crucial technique for enhancing the generalization of deep learning registration models, especially in the medical imaging domain, where high-quality and diverse multimodal data are often scarce. Prior multimodal registration approaches faced multiple limitations, such as intricate implementation processes and overfitting, which reduce the generalizability of the models and impact registration accuracy. These limitations underscore the need for improved methodologies to enhance the effectiveness of medical image analysis by advancing the progressive GAN framework that synthesizes high-quality multimodal medical images. In this study, we propose an approach based on Generative Adversarial Networks (GANs) to improve the quality and diversity of multimodal medical images. Our methodology includes preprocessing steps, real-time modifications during GAN training, and post-processing techniques to enhance the generated images. The results demonstrate that the proposed method outperforms traditional registration techniques, achieving a mean Dice Similarity Coefficient of 0.78, indicating a significant improvement in registration accuracy. These findings support the potential application of our approach in clinical settings, enhancing the effectiveness of medical image analysis.