Enhancing breast cancer classification via histopathological image analysis: Leveraging self-supervised contrastive learning and transfer learning.

Breast cancer, a significant threat to women's health, demands early detection. Automating histopathological image analysis offers a promising solution to enhance efficiency and accuracy in diagnosis. This study addresses the challenge of breast cancer histopathological image classification by leveraging the ResNet architecture, known for its depth and skip connections. In this work, two distinct approaches were pursued, each driven by unique motivations. The first approach aimed to improve the learning process through self-supervised contrastive learning. It utilizes a small subset of the training data for initial model training and progressively expands the training set by incorporating confidently labeled data from the unlabeled pool, ultimately achieving a reliable model with limited training data. The second approach focused on optimizing the architecture by combining ResNet50 and Inception module to get a lightweight and efficient classifier. The dataset utilized in this work comprises histopathological images categorized into benign and malignant classes at varying magnification levels (40X, 100X, 200X, 400X), all originating from the same source image. The results demonstrate state-of-the-art performance, achieving 98% accuracy for images magnified at 40X and 200X, and 94% for 100X and 400X. Notably, the proposed architecture boasts a substantially reduced parameter count of approximately 3.6 million, contrasting with existing leading architectures, which possess parameter sizes at least twice as large.

Nordic socio-recreational ecosystem services in a hydropeaked river.

Fluctuating energy prices call for short-term river flow regulation at hydropower plants (HPPs), which can lead to hydropeaking - the pulsating water flow downstream from a HPP. Hydropeaking can affect land use areas of regulated rivers and subsequently their socio-recreational ecosystem services (SRESs). These areas often offer a range of services, such as swimming, boating, fishing, hiking, cycling, and berry picking. Such activities hold significant value in Nordic culture and for human wellbeing. We have examined how SRES land use areas are affected by hourly hydropeaking in a reach of the Kemijoki River in Finland. First, we determined the state of hydropeaking in the river by employing two indicators, normalized daily maximum flow difference and sub-daily flow ramping. Next, we looked at the spatiotemporal impacts of peaking hydrology using inundation maps derived from 2D-hydrodynamic modeling and a high-resolution land use map with clearly identified SRES areas. Finally, we examined the hazards to hydraulic safety in the river channel in the context of instream recreation. Our results show that hydropeaking levels in the study area remained consistently high throughout the entire study period, from 2010 to 2021. This was the case in all seasons except for the spring of 2013, 2016 and 2019. We determined that hydropeaking impacts on SRESs are mostly felt in the littoral zone (0.84 km2 i.e., 3.1 % of the study area) during the summer season as 25 % (0.21 km2) of this zone is influenced by hydropeaking. In addition, multiple recreational use areas in this zone, such as beaches, riparian forest, and summer cottages, were found to be affected by hydropeaking. The results show that most of the river channel becomes hydraulically unsafe during high ramping flows. The highest hazard to instream recreation opportunities is likely to occur during summer. Consequently, hydropeaking can threaten the social and recreational services of Nordic rivers.

Bio-activity prediction of drug candidate compounds targeting SARS-Cov-2 using machine learning approaches.

The SARS-CoV-2 3CLpro protein is one of the key therapeutic targets of interest for COVID-19 due to its critical role in viral replication, various high-quality protein crystal structures, and as a basis for computationally screening for compounds with improved inhibitory activity, bioavailability, and ADMETox properties. The ChEMBL and PubChem database contains experimental data from screening small molecules against SARS-CoV-2 3CLpro, which expands the opportunity to learn the pattern and design a computational model that can predict the potency of any drug compound against coronavirus before in-vitro and in-vivo testing. In this study, Utilizing several descriptors, we evaluated 27 machine learning classifiers. We also developed a neural network model that can correctly identify bioactive and inactive chemicals with 91% accuracy, on CheMBL data and 93% accuracy on combined data on both CheMBL and Pubchem. The F1-score for inactive and active compounds was 93% and 94%, respectively. SHAP (SHapley Additive exPlanations) on XGB classifier to find important fingerprints from the PaDEL descriptors for this task. The results indicated that the PaDEL descriptors were effective in predicting bioactivity, the proposed neural network design was efficient, and the Explanatory factor through SHAP correctly identified the important fingertips. In addition, we validated the effectiveness of our proposed model using a large dataset encompassing over 100,000 molecules. This research employed various molecular descriptors to discover the optimal one for this task. To evaluate the effectiveness of these possible medications against SARS-CoV-2, more in-vitro and in-vivo research is required.

YoNet: A Neural Network for Yoga Pose Classification.

Yoga has become an integral part of human life to maintain a healthy body and mind in recent times. With the growing, fast-paced life and work from home, it has become difficult for people to invest time in the gymnasium for exercises. Instead, they like to do assisted exercises at home where pose recognition techniques play the most vital role. Recognition of different poses is challenging due to proper dataset and classification architecture. In this work, we have proposed a deep learning-based model to identify five different yoga poses from comparatively fewer amounts of data. We have compared our model's performance with some state-of-the-art image classification models-ResNet, InceptionNet, InceptionResNet, Xception and found our architecture superior. Our proposed architecture extracts spatial, and depth features from the image individually and considers them for further calculation in classification. The experimental results show that it achieved 94.91% accuracy with 95.61% precision.

Causes & effects of upstream-downstream flow regime alteration over Catchment-Estuary-Coastal systems.

The construction of large dams along rivers has significantly changed the natural flow regime, reducing the inflow into many lakes and terminal wetlands. However, the question of the impact of dam operation on downstream estuarine wetlands has less been taken into account. Spatio-temporal flow regime alteration in the Mond River shows the complexity of drivers affecting the estuary-coastal system named the Mond-Protected Area in southern Iran. To this end, we applied river impact (RI) and Indicator of hydrological alteration (IHA) methods on monthly and daily river flow data across the basin. Based on the river impact method, a "drastic" impact below two in-operation (Tangab and Salman Farsi) dams, with RI values of 0.02 and 0.08, diminish to a 'severe' impact with RI value of 0.35 at the last gauge (Ghantareh) on the main corridor of the Mond river due to the addition of flow from a large mid-basin (about 20,254 km2). Furthermore, the degree of hydrological alteration (daily flow analysis) at mid-stream (e.g., Dehram gauges) was similar to the unregulated upstream tributaries (e.g., Hanifaghan gauges). The remote sensing analysis in the Mond Protected Area showed the prevailing impact of sea-level rise in the Persian Gulf with the inundation of the coastal area and a shift of vegetation in a landward direction which complied with standardized precipitation index (SPI) values as a meteorological drought indicator. Thus, the consequence of climate change (e.g., sea-level rise, draught) has a higher impact on the protected area than the upstream river regulation and land-use change in the Mond basin. The holistic approach and the catchment-level study allowed us to see the complexity of the drivers influencing the estuary-coastal system.

An Analysis of Different Distance-Linkage Methods for Clustering Gene Expression Data and Observing Pleiotropy: Empirical Study.

BACKGROUND: Large amounts of biological data have been generated over the last few decades, encouraging scientists to look for connections between genes that cause various diseases. Clustering illustrates such a relationship between numerous species and genes. Finding an appropriate distance-linkage metric to construct clusters from diverse biological data sets has thus become critical. Pleiotropy is also important for a gene's expression to vary and create varied consequences in living things. Finding the pleiotropy of genes responsible for various diseases has become a major research challenge. OBJECTIVE: Our goal was to establish the optimal distance-linkage strategy for creating reliable clusters from diverse data sets and identifying the common genes that cause various tumors to observe genes with pleiotropic effect. METHODS: We considered 4 linking methods-single, complete, average, and ward-and 3 distance metrics-Euclidean, maximum, and Manhattan distance. For assessing the quality of different sets of clusters, we used a fitness function that combines silhouette width and within-cluster distance. RESULTS: According to our findings, the maximum distance measure produces the highest-quality clusters. Moreover, for medium data set, the average linkage method, and for large data set, the ward linkage method works best. The outcome is not improved by using ensemble clustering. We also discovered genes that cause 3 different cancers and used gene enrichment to confirm our findings. CONCLUSIONS: Accuracy is crucial in clustering, and we investigated the accuracy of numerous clustering techniques in our research. Other studies may aid related works if the data set is similar to ours.

Changes in short term river flow regulation and hydropeaking in Nordic rivers.

Quantifying short-term changes in river flow is important in understanding the environmental impacts of hydropower generation. Energy markets can change rapidly and energy demand fluctuates at sub-daily scales, which may cause corresponding changes in regulated river flow (hydropeaking). Due to increasing use of renewable energy, in future hydropower will play a greater role as a load balancing power source. This may increase current hydropeaking levels in Nordic river systems, creating challenges in maintaining a healthy ecological status. This study examined driving forces for hydropeaking in Nordic rivers using extensive datasets from 150 sites with hourly time step river discharge data. It also investigated the influence of increased wind power production on hydropeaking. The data revealed that hydropeaking is at high levels in the Nordic rivers and have seen an increase over the last decade and especially over the past few years. These results indicate that increased building for renewable energy may increase hydropeaking in Nordic rivers.