Background-The incidence of thyroid cancer is rising steadily because of overdiagnosis and overtreatment conferred by widespread use of sensitive imaging techniques for screening. This overall incidence growth is especially driven by increased diagnosis of indolent and welldifferentiated papillary subtype and early-stage thyroid cancer, whereas the incidence of advancedstage thyroid cancer has increased marginally. Thyroid ultrasound is frequently used to diagnose thyroid cancer. The aim of this study was to use deep convolutional neural network (DCNN) models to improve the diagnostic accuracy of thyroid cancer by analysing sonographic imaging data from clinical ultrasounds.Methods-We did a retrospective, multicohort, diagnostic study using ultrasound images sets from three hospitals in China. We developed and trained the DCNN model on the training set, 131 731 ultrasound images from 17 627 patients with thyroid cancer and 180 668 images from 25 325 controls from the thyroid imaging database at Tianjin Cancer Hospital. Clinical diagnosis of the training set was made by 16 radiologists from Tianjin Cancer Hospital. Images from anatomical sites that were judged as not having cancer were excluded from the training set and only individuals with suspected thyroid cancer underwent pathological examination to confirm diagnosis. The model's diagnostic performance was validated in an internal validation set from Tianjin Cancer Hospital (8606 images from 1118 patients) and two external datasets in China (the
Pollination networks are usually constructed and assessed by direct field observations which commonly assume that all flower visitors are true pollinators. However, this assumption is often invalid and the use of data based on mere visitors to flowers may lead to a misunderstanding of intrinsic pollination networks. Here, using a large dataset by both sampling floral visitors and analyzing their pollen loads, we constructed 32 networks pairs (visitation versus pollen transport) across one flowering season at four elevation sites in the Himalaya–Hengduan Mountains region. Pollen analysis was conducted to determine which flower visitors acted as potential pollinators (pollen vectors) or as cheaters (those not carrying pollen of the visited plants). We tested whether there were topological differences between visitation and pollen transport networks and whether different taxonomic groups of insect visitors differed in their ability to carry pollen of the visited plants. Our results indicated that there was a significantly higher degree of specialization at both the network and species levels in the pollen transport networks in contrast to the visitation networks. Modularity was lower but nestedness was higher in the visitation networks compared to the pollen transport networks. All the cheaters were identified as peripheral species and most of them contributed positively to the nested structure. This may explain in part the differences in modularity and nestedness between the two network types. Bees carried the highest proportion of pollen of the visited plants. This was followed by Coleoptera, other Hymenoptera and Diptera. Lepidoptera carried the lowest proportion of pollen of the visited plants. Our study shows that the construction of pollen transport networks could provide a more in‐depth understanding of plant–pollinator interactions. Moreover, it suggests that detecting and removing cheater interactions when studying the topology of other mutualistic networks might be also important.
BackgroundHow floral traits and community composition influence plant specialization is poorly understood and the existing evidence is restricted to regions where plant diversity is low. Here, we assessed whether plant specialization varied among four species-rich subalpine/alpine communities on the Yulong Mountain, SW China (elevation from 2725 to 3910 m). We analyzed two factors (floral traits and pollen vector community composition: richness and density) to determine the degree of plant specialization across 101 plant species in all four communities. Floral visitors were collected and pollen load analyses were conducted to identify and define pollen vectors. Plant specialization of each species was described by using both pollen vector diversity (Shannon’s diversity index) and plant selectiveness (d’ index), which reflected how selective a given species was relative to available pollen vectors.ResultsPollen vector diversity tended to be higher in communities at lower elevations, while plant selectiveness was significantly lower in a community with the highest proportion of unspecialized flowers (open flowers and clusters of flowers in open inflorescences). In particular, we found that plant species with large and unspecialized flowers attracted a greater diversity of pollen vectors and showed higher selectiveness in their use of pollen vectors. Plant species with large floral displays and high flower abundance were more selective in their exploitation of pollen vectors. Moreover, there was a negative relationship between plant selectiveness and pollen vector density.ConclusionsThese findings suggest that flower shape and flower size can increase pollen vector diversity but they also increased plant selectiveness. This indicated that those floral traits that were more attractive to insects increased the diversity of pollen vectors to plants while decreasing overlap among co-blooming plant species for the same pollen vectors. Furthermore, floral traits had a more important impact on the diversity of pollen vectors than the composition of anthophilous insect communities. Plant selectiveness of pollen vectors was strongly influenced by both floral traits and insect community composition. These findings provide a basis for a better understanding of how floral traits and community context shape interactions between flowers and their pollen vectors in species-rich communities.Electronic supplementary materialThe online version of this article (doi:10.1186/s12898-016-0080-1) contains supplementary material, which is available to authorized users.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.