The rapid advancement of medical imaging technologies requires the development of advanced, automated, and interpretable diagnostic tools for clinical decision-making. Although convolutional neural networks (CNNs) have shown significant promise in medical image analysis, they have limitations in capturing the global context and lack interpretability, thereby hindering their clinical adoption. This study presents EFFResNet-ViT, a novel hybrid deep learning (DL) model designed to address these challenges by combining EfficientNet-B0 and ResNet-50 CNN backbones with a vision transformer (ViT) module. The proposed architecture employs a feature fusion strategy to integrate the local feature extraction strengths of CNNs with the global dependency modeling capabilities of transformers. The extracted features are further refined through a post-transformer CNN and a global average pooling layer to enhance the classification performance. To improve interpretability, EFFResNet-ViT incorporates Grad-CAM visualization techniques to highlight regions contributing to classification decisions and employs t-distributed stochastic neighbor embedding for feature space analysis, providing insights into class separability. The proposed model was evaluated on two benchmark datasets: brain tumor (BT) CE-MRI for brain tumor classification and a retinal image dataset for ophthalmological diagnosis. EFFResNet-ViT achieved state-of-the-art performance, with accuracies of 99.31% and 92.54% on the BT CE-MRI and retinal datasets, respectively. Comparative analyses demonstrate the superior classification performance and interpretability of EFFResNet-ViT over existingViT and CNN-based hybrid models. The explainable design of EFFResNet-ViT addresses the critical need for transparency in artificial intelligence-driven medical diagnostics, facilitating its potential integration into clinical workflows to improve decision-making and patient outcomes.

Global warming is one of the main challenges of recent times. The glaciers are melting faster than expected which has resulted in global mean sea level rise and increased the risk of floods. The development of modern remote sensing technology has made it possible to obtain images more frequently than ever before. On the other side, the availability of high-performance computing hardware and processing techniques have made it possible to provide a cost-effective solution to monitor the temporal changes of glaciers at a large scale. In this study, supervised machine learning methods are investigated for automatic classification of glacier covers from multi-temporal Sentinel-2 imagery using texture, topographic, and spectral data. Three most commonly used supervised machine learning techniques were investigated: support vector machine (SVM), artificial neural network (ANN) and random forest (RF). The proposed method was employed on the data obtained from Passu watershed in Hunza Basin located along the Hunza river in Pakistan. Three main classes were considered: glaciers, debris-covered glaciers and non-glaciated areas. The data was split into training (70%) and testing datasets (30%). Finally, an area-based accuracy assessment was performed by comparing the results obtained for each classifier with the reference data. Experiments showed that the results produced for all classifiers were highly accurate and visually more consistent with the depiction of glacier cover types. For all experiments, random forest performed the best (Kappa= 0.95, f-measure =95.06%) on all three classes compared to ANN (Kappa= 0.92, f-measure =92.05) and SVM (Kappa= 0.89, f-measure =91.86% on average). The high classification accuracy obtained to distinguished debris-covered glaciers using our approach will be useful to determine the actual available water resources which can be further helpful for hazard and water resource management.

The most frequent and noticeable natural calamity in the Karakoram region is landslides. Extreme landslides have occurred frequently along Karakoram Highway, particularly during monsoons, causing a major loss of life and property. Therefore, it is necessary to look for a solution to increase growth and vigilance in order to lessen losses related to landslides caused by natural disasters. By utilizing contemporary technologies, an early warning system might be developed. Artificial neural networks (ANNs) are widely used nowadays across many industries. This paper’s major goal is to provide new integrative models for assessing landslide susceptibility in a prone area in the north of Pakistan. To achieve this, the training of an artificial neural network (ANN) was supervised using metaheuristic and Bayesian techniques: Particle Swarm Optimization (PSO) algorithm, Genetic algorithm (GA), Bayesian Optimization Gaussian Process (BO_GP), and Bayesian Optimization Tree-structured Parzen Estimator (BO_TPE). In total, 304 previous landslides and the eight most prevalent conditioning elements were combined to form a geospatial database. The models were hyperparameter optimized, and the best ones were employed to generate susceptibility maps. The obtained area under the curve (AUC) accuracy index demonstrated that the maps produced by both Bayesian and metaheuristic algorithms are highly accurate. The effectiveness and efficiency of applying ANNs for landslide mapping, susceptibility analysis, and forecasting were studied in this research, and it was observed from experimentation that the performance differences for GA, BO_GP, and PSO compared to BO_TPE were relatively small, ranging from 0.32% to 1.84%. This suggests that these techniques achieved comparable performance to BO_TPE in terms of AUC. However, it is important to note that the significance of these differences can vary depending on the specific context and requirements of the ML task. Additionally, in this study, we explore eight feature selection algorithms to determine the geospatial variable importance for landslide susceptibility mapping along the Karakoram Highway (KKH). The algorithms considered include Information Gain, Variance Inflation Factor, OneR Classifier, Subset Evaluators, principal components, Relief Attribute Evaluator, correlation, and Symmetrical Uncertainty. These algorithms enable us to evaluate the relevance and significance of different geospatial variables in predicting landslide susceptibility. By applying these feature selection algorithms, we aim to identify the most influential geospatial variables that contribute to landslide occurrences along the KKH. The algorithms encompass a diverse range of techniques, such as measuring entropy reduction, accounting for attribute bias, generating single rules, evaluating feature subsets, reducing dimensionality, and assessing correlation and information sharing. The findings of this study will provide valuable insights into the critical geospatial variables associated with landslide susceptibility along the KKH. These insights can aid in the development of effective landslide mitigation strategies, infrastructure planning, and targeted hazard management efforts. Additionally, the study contributes to the field of geospatial analysis by showcasing the applicability and effectiveness of various feature selection algorithms in the context of landslide susceptibility mapping.

The Indus River, which flows through China, India, and Pakistan, is mainly fed by melting snow and glaciers that are spread across the Hindukush–Karakoram–Himalaya Mountains. The downstream population of the Indus Plain heavily relies on this water resource for drinking, irrigation, and hydropower generation. Therefore, its river runoff variability must be properly monitored. Gilgit Basin, the northwestern part of the Upper Indus Basin, is selected for studying cryosphere dynamics and its implications on river runoff. In this study, 8-day snow products (MOD10A2) of moderate resolution imaging spectroradiometer, from 2001 to 2015 are selected to access the snow-covered area (SCA) in the catchment. A non-parametric Mann–Kendall test and Sen’s slope are calculated to assess whether a significant trend exists in the SCA time series data. Then, data from ground observatories for 1995–2013 are analyzed to demonstrate annual and seasonal signals in air temperature and precipitation. Results indicate that the annual and seasonal mean of SCA show a non-significant decreasing trend, but the autumn season shows a statistically significant decreasing SCA with a slope of −198.36 km²/year. The annual mean temperature and precipitation show an increasing trend with highest values of slope 0.05 °C/year and 14.98 mm/year, respectively. Furthermore, Pearson correlation coefficients are calculated for the hydro-meteorological data to demonstrate any possible relationship. The SCA is affirmed to have a highly negative correlation with mean temperature and runoff. Meanwhile, SCA has a very weak relation with precipitation data. The Pearson correlation coefficient between SCA and runoff is −0.82, which confirms that the Gilgit River runoff largely depends on the melting of snow cover rather than direct precipitation. The study indicates that the SCA slightly decreased for the study period, which depicts a possible impact of global warming on this mountainous region.

Gravity Recovery and Climate Experiment (GRACE) and satellite altimetry are suitable for the precise measurement of terrestrial water storage (TWS) and lake water level variations from space. In this study, two GRACE solutions, namely, spherical harmonics (SH) and mascon (MSC), are utilized with the Global Land Data Assimilation System (GLDAS) model to estimate the spatial and temporal variations of TWS in the Upper Indus Basin (UIB) for the study period of January 2003 to December 2016. The TWS estimated by SH, MSC, and the GLDAS model are consistent and generally show negative trends of -4.47 ± 0.38 mm/year, -4.81 ± 0.49 mm/year, and -3.77 ± 0.46 mm/year, respectively. Moreover, we use the GLDAS model data to understand the roles of variations in land surface state variables (snow water equivalent (SWE), soil moisture, and canopy water storage) in enhancing or dissipating the TWS in the region. Results indicate that SWE, which has a significant contribution to GRACE TWS variability, is an important parameter. Spearman's rank correlations are calculated to demonstrate the relationship of the GLDAS land surface state variables and the GRACE signals. A highly positive correlation between SWE with TWS is estimated by SH and MSC as 0.691 and 0.649, respectively, indicating that the TWS signal is mainly reliant on snow water in the study region. The ground water storages estimated by SH and MSC solutions are nearly stable with slight increasing trends of 0.63 ± 0.48 mm/year and 0.29 ± 0.51 mm/year, respectively. We also take advantage of the potential of satellite altimetry in measuring lake water level variations, and our result indicates that Cryosat-2 SARin mode altimetry data can be used in estimating small water bodies accurately in the high mountainous region of the UIB. Moreover, the climate indices data of El-Niño Southern Oscillation and Pacific Decadal Oscillation are analyzed to determine the influence of pacific climatic variability on TWS.