-
IEEE Journal of Biomedical and Health... Mar 2024Medical imaging is a key component in clinical diagnosis, treatment planning and clinical trial design, accounting for almost 90% of all healthcare data. CNNs achieved...
Medical imaging is a key component in clinical diagnosis, treatment planning and clinical trial design, accounting for almost 90% of all healthcare data. CNNs achieved performance gains in medical image analysis (MIA) over the last years. CNNs can efficiently model local pixel interactions and be trained on small-scale MI data. Despite their important advances, typical CNN have relatively limited capabilities in modelling "global" pixel interactions, which restricts their generalisation ability to understand out-of-distribution data with different "global" information. The recent progress of Artificial Intelligence gave rise to Transformers, which can learn global relationships from data. However, full Transformer models need to be trained on large-scale data and involve tremendous computational complexity. Attention and Transformer compartments ("Transf/Attention") which can well maintain properties for modelling global relationships, have been proposed as lighter alternatives of full Transformers. Recently, there is an increasing trend to co-pollinate complementary local-global properties from CNN and Transf/Attention architectures, which led to a new era of hybrid models. The past years have witnessed substantial growth in hybrid CNN-Transf/Attention models across diverse MIA problems. In this systematic review, we survey existing hybrid CNN-Transf/Attention models, review and unravel key architectural designs, analyse breakthroughs, and evaluate current and future opportunities as well as challenges. We also introduced an analysis framework on generalisation opportunities of scientific and clinical impact, based on which new data-driven domain generalisation and adaptation methods can be stimulated.
Topics: Artificial Intelligence; Image Processing, Computer-Assisted; Computer Simulation
PubMed: 38157463
DOI: 10.1109/JBHI.2023.3348436 -
Proceedings. Biological Sciences Jun 2024Pesticides have been identified as major drivers of insect biodiversity loss. Thus, the study of their effects on non-pest insect species has attracted a lot of... (Review)
Review
Pesticides have been identified as major drivers of insect biodiversity loss. Thus, the study of their effects on non-pest insect species has attracted a lot of attention in recent decades. In general toxicology, the 'gold standard' to assess the toxicity of a substance is to measure mass-specific LD (i.e. median lethal dose per unit body mass). In entomology, reviews attempting to compare these data across all available studies are lacking. To fill this gap in knowledge, we performed a systematic review of the lethality of imidacloprid for adult insects. Imidacloprid is possibly the most extensively studied insecticide in recent times, yet we found that little is comparable across studies, owing to both methodological divergence and missing estimates of body mass. By accounting for body mass whenever possible, we show how imidacloprid sensitivity spans across an apparent range of approximately six orders of magnitude across insect species. Very high variability within species can also be observed owing to differences in exposure methods and observation time. We suggest that a more comparable and comprehensive approach has both biological and economic relevance. Ultimately, this would help to identify differences that could direct research towards preventing non-target species from being negatively affected.
Topics: Neonicotinoids; Nitro Compounds; Animals; Insecticides; Insecta; Imidazoles; Species Specificity; Lethal Dose 50
PubMed: 38864325
DOI: 10.1098/rspb.2023.2811 -
Journal of Environmental Management Feb 2024In Europe, agri-environment schemes (AES) are a key instrument to combat the ongoing decline of farmland biodiversity. AES aim is to support biodiversity and maintain... (Meta-Analysis)
Meta-Analysis
In Europe, agri-environment schemes (AES) are a key instrument to combat the ongoing decline of farmland biodiversity. AES aim is to support biodiversity and maintain ecosystem services, such as pollination or pest control. To what extent AES affect crop yield is still poorly understood. We performed a systematic review, including hierarchical meta-analyses, to investigate potential trade-offs and win-wins between the effectiveness of AES for arthropod diversity and agricultural yield on European croplands. Altogether, we found 26 studies with a total of 125 data points that fulfilled our study inclusion criteria. From each study, we extracted data on biodiversity (arthropod species richness and abundance) and yield for fields with AES management and control fields without AES. The majority of the studies reported significantly higher species richness and abundance of arthropods (especially wild pollinators) in fields with AES (31 % increase), but yields were at the same time significantly lower on fields with AES compared to control fields (21 % decrease). Aside from the opportunity costs, AES that promote out-of-production elements (e.g. wildflower strips), supported biodiversity (29-32 % increase) without significantly compromising yield (2-5 % increase). Farmers can get an even higher yield in these situations than in current conventional agricultural production systems without AES. Thus, our study is useful to identify AES demonstrating benefits for arthropod biodiversity with negligible or relatively low costs regarding yield losses. Further optimization of the design and management of AES is needed to improve their effectiveness in promoting both biodiversity and minimizing crop yield losses.
Topics: Animals; Ecosystem; Arthropods; Biodiversity; Agriculture; Crops, Agricultural
PubMed: 38325288
DOI: 10.1016/j.jenvman.2024.120277