To effectively treat lung and colon cancer and save lives, early and accurate identification is essential. Conventional diagnosis takes a long time and requires the manual expertise of radiologists. The rising number of new cancer cases makes it challenging to process massive volumes of data quickly. Different machine learning approaches to the classification and detection of lung and colon cancer have been proposed by multiple research studies. However, when it comes to self-learning classification and detection tasks, deep learning (DL) excels. This paper suggests a novel DL convolutional neural network (CNN) model for detecting lung and colon cancer. The proposed model is lightweight and multi-scale since it uses only 1.1 million parameters, making it appropriate for real-time applications as it provides an end-to-end solution. By incorporating features extracted at multiple scales, the model can effectively capture both local and global patterns within the input data. The explainability tools such as gradient-weighted class activation mapping and Shapley additive explanation can identify potential problems by highlighting the specific input data areas that have an impact on the model’s choice. The experimental findings demonstrate that for lung and colon cancer detection, the proposed model was outperformed by the competition and accuracy rates of 99.20% have been achieved for multi-class (containing five classes) predictions.

In the field of medical imaging, deep learning (DL) techniques have made significant contributions to the detection and classification of various cancers. Identifying the precise regions in medical images containing cancerous cells plays a crucial role in the diagnostic process. Early and accurate cancer detection is essential for effective treatment and improved patient outcomes. However, manual diagnosis is labor-intensive, requiring the specialized expertise of radiologists, and the increasing number of cancer cases presents challenges in processing large volumes of image data efficiently. To address these challenges, an end-to-end concatenated Convolutional Neural Network (CNN) attention model has been proposed for automatic lung cancer classification. This approach integrates two distinct CNNs, followed by a multi-layer perceptron (MLP) and a multi-head attention (MHA) mechanism, to enhance performance. The model leverages explainable AI techniques, such as gradient-weighted class activation mapping (grad-CAM) and Shapley additive explanations (SHAP), to highlight critical regions within the images that influence the decision-making process. This model achieves impressive performance, with an accuracy of 99.54%, precision of 99.31%, recall of 99.95%, F1-score of 99.66%, and an AUC of 99.97%. These results demonstrate that this approach not only surpasses existing methods but also provides a highly accurate and interpretable solution. By reducing the need for extensive manual intervention, this model enables faster and more reliable lung cancer diagnosis, paving the way for timely and effective treatments.

Background: Examining chest radiograph images (CXR) is an intricate and time-consuming process, sometimes requiring the identification of many anomalies at the same time. Lung segmentation is key to overcoming this challenge through different deep learning (DL) techniques. Many researchers are working to improve the performance and efficiency of lung segmentation models. This article presents a DL-based approach to accurately identify the lung mask region in CXR images to assist radiologists in recognizing early signs of high-risk lung diseases. Methods: This paper proposes a novel technique, Lightweight Residual U-Net, combining the strengths of the convolutional block attention module (CBAM), the Atrous Spatial Pyramid Pooling (ASPP) block, and the attention module, which consists of only 3.24 million trainable parameters. Furthermore, the proposed model has been trained using both the RELU and LeakyReLU activation functions, with LeakyReLU yielding superior performance. The study indicates that the Dice loss function is more effective in achieving better results. Results: The proposed model is evaluated on three benchmark datasets: JSRT, SZ, and MC, achieving a Dice score of 98.72%, 97.49%, and 99.08%, respectively, outperforming the state-of-the-art models. Conclusions: Using the capabilities of DL and cutting-edge attention processes, the proposed model improves current efforts to enhance lung segmentation for the early identification of many serious lung diseases.

The detection of plant leaf diseases is essential for ensuring crop health and productivity. This study uses a comprehensive merged dataset, including the Mendeley Plant Leaf Dataset, which includes 22 classes, while the Jute and Mulberry datasets provide 2 classes each, representing healthy and diseased categories from various species. The final dataset, consisting of 26 classes, was augmented to ensure 500 training samples per class. In this study, the Segment Anything Model (SAM) was employed for zero-shot segmentation, enabling the automatic extraction of precise regions of interest (ROIs) from leaf images without the need for task-specific training for region-focused analysis. An EfficientNetB7- Multi-Head Attention (MHA) model that combines EfficientNetB7 with MHA was used to improve plant disease classification across 26 distinct classes. The proposed model is designed to handle the variety and diversity of leaf diseases, achieving a high classification accuracy of 98.01%, with precision, recall, and F1-scores all exceeding 97.9%. The integration of MHA allows the model to focus on disease-specific features, significantly enhancing its ability to generalize across diverse agricultural settings while maintaining scalability. Experimental results show that the EfficientNetB7-MHA model consistently outperforms several state-of-the-art (SOTA) models in terms of accuracy and robustness, making it a promising tool for precision agriculture. Explainable AI (XAI) including the use of Grad-CAM was utilized to detect the precise regions of interest for each class providing interpretability and insights into the model’s decision-making process. Finally, to demonstrate the model’s performance, a web application was developed that displays the predicted outputs when given different images of healthy and diseased leaves from the dataset. This comprehensive approach demonstrates significant potential for improving agricultural disease management through accurate, scalable, interpretable, and efficient disease detection.

Fused filament fabrication (FFF) printed polyetherimide (PEI), commercially available as ULTEM 9085, is qualified by the Federal Aviation Administration for aerospace structural applications. Its mechanical performance depends strongly on printing parameters, including build orientation, number of contours, raster angle, and air gap, whose nonlinear interactions govern void formation, interlayer bonding, and ultimate strength. Prior work using artificial neural networks (ANNs) predicted ULTEM 9085 properties with under 5% error from a single data split, leaving two questions open: whether that accuracy holds across repeated splits, and whether the model's reasoning can be physically interpreted. This study addresses both gaps. Eight regression algorithms (Ridge, Decision Tree, Random Forest, Extra Trees, Gradient Boosting, XGBoost, K‐Nearest Neighbors, and Support Vector Regression) are benchmarked to jointly predict complete stress–strain curves and three scalar mechanical properties (tensile strength, elastic modulus, yield strength) from the four printing parameters, evaluated across five random data splits. SHAP and LIME are then applied to the best‐performing model to attribute its predictions to individual printing parameters. Extra Trees Regressor achieves the lowest and most stable test mean absolute error (MAE) of 0.162 ± 0.041 in scaled space across five seeds, with scalar property errors of 4.0% (tensile strength), 4.4% (elastic modulus), and 8.5% (yield strength), accuracy comparable to the ANN baseline at a fraction of the computational cost. SHAP and LIME attributions agree that the number of contours and build orientation are the dominant drivers of tensile strength, with air gap acting as a smaller, suppressive factor. These results support Extra Trees with SHAP/LIME explanation as a repeatable, interpretable, and computationally inexpensive surrogate for reducing the physical testing burden in aerospace qualification of 3D‐printed ULTEM 9085.