Canada

Positron Emission Tomography (PET) Imaging of the brain might become the most effective imaging technique to predict Alzheimer&#39;s disease. However, the definition of the brain in PET images is low and the lesion area is not easy to define, so the accuracy of traditional machine learning algorithms in predicting Alzheimer&#39;s disease from PET images is low. Deep learning algorithms can effectively improve the accuracy of prediction. Here, a deep learning model based on multi-attention mechanisms is constructed for the prediction of Alzheimer&#39;s disease. Firstly, in order to improve the images and reduce the loss of spatial information caused by convolution, a soft-attention mechanism was introduced to embed non-local modules and CBAM modules into the prediction model, which effectively solved the problems of the lack of detail information and the lack of connection between channels in PET images after deep convolution. Secondly, a split-attention mechanism was introduced, and the influence of different parameters of feature images with different sizes on the prediction accuracy was enhanced by using a packet network. Finally, head movement correction, image registration, craniocerebral separation, Gaussian smoothing and other image preprocessing were performed on the acquired PET images to effectively improve the image features of Alzheimer&#39;s disease focal areas in brain PET images. The experimental results showed that the prediction accuracy, sensitivity and specificity of the model in the ADNI database for Alzheimer&#39;s disease were 90.5%, 86.1% and 93.5% respectively, which could provide more accurate diagnostic results compared with the existing methods.

CAP Virtual Congress 2021

Prediction of Alzheimer&#39;s Disease Based on Multi-Attentional Mechanisms using Brain PET Image

brain pet images; multi-attention mechanism; alzheimer's disease prediction

**WELCOME TO THE 2021 CAP VIRTUAL CONGRESS UNDERLINE PLATFORM**

The CAP staff and our organizing and program committees are pleased to welcome you to our 2021 Congress filled with plenary talks, technical sessions/talks, student presentations and competition sessions, workshops, and several symposia offered by the CAP divisions and its partners.  

You will also have opportunities to network with other delegates through the Gathertown platform accessed by selecting the lounge or posters button on the left.

Registered delegates should use the schedule button to call up the details of the session/event they wish to attend.  Each sessions includes a link to access the online platform where the event is taking place, either within Underline or through other Congress-related Zoom meetings.  

Invited speaker talks are identified with an (I), and student competitors include a (G*) or (U*) to identify graduate or undergraduate presenters.

Please take the time to check out the information about our sponsors and exhibitors, request further information, or leave a note of thanks for their support.

General information about the Congress can be found on the CAP website Congress pages at https://www.cap.ca/congress/2021.

Looking for a specific talk?   A full searchable list is available on our Indico Congress site at https://indico.cern.ch/event/985448/contributions/

We look forward to sharing this exciting week with you!


Positron Emission Tomography (PET) Imaging of the brain might become the most effective imaging technique to predict Alzheimer's disease. However, the definition of the brain in PET images is low and the lesion area is not easy to define, so the accuracy of traditional machine learning algorithms in predicting Alzheimer's disease from PET images is low. Deep learning algorithms can effectively improve the accuracy of prediction. Here, a deep learning model based on multi-attention mechanisms is constructed for the prediction of Alzheimer's disease. Firstly, in order to improve the images and reduce the loss of spatial information caused by convolution, a soft-attention mechanism was introduced to embed non-local modules and CBAM modules into the prediction model, which effectively solved the problems of the lack of detail information and the lack of connection between channels in PET images after deep convolution. Secondly, a split-attention mechanism was introduced, and the influence of different parameters of feature images with different sizes on the prediction accuracy was enhanced by using a packet network. Finally, head movement correction, image registration, craniocerebral separation, Gaussian smoothing and other image preprocessing were performed on the acquired PET images to effectively improve the image features of Alzheimer's disease focal areas in brain PET images. The experimental results showed that the prediction accuracy, sensitivity and specificity of the model in the ADNI database for Alzheimer's disease were 90.5%, 86.1% and 93.5% respectively, which could provide more accurate diagnostic results compared with the existing methods.

Prediction of Alzheimer's Disease Based on Multi-Attentional Mechanisms using Brain PET Image

technical paper

The behaviour of confined polymeric systems has been the topic of much recent research. The information from these studies may help guide the development and optimization of nanofluidic devices and understand the segregation of DNA chromosomes in prokaryotic cells. Some recent experiments examined two DNA molecules confined to a box-like cavity with strong confinement in one dimension. The researchers looked at the effect of using an anisotropic cross-sectional box geometry and of having non-identical polymers. In this study, we use computer simulations to examine the effect of such anisotropies on the behaviour of the confined-polymer systems. We employ Metropolis Monte Carlo simulations where the polymers are modelled as bead spring chains. We calculate probability distributions for the centre-of-mass positions of either polymer and for the centre-of-mass displacement between the polymers. These distributions are used to examine the effect of anisotropies in polymer length for two polymers confined to a square box-like cavity and anisotropies in box dimensions for two identical polymers confined to a box-like cavity. The information obtained from this study can then be used to have a better understanding of the trends seen in the experiments.

Simulation of Two Polymers Confined to a Box-Like Cavity: The Effect of Anisotropies in Polymer Length and Confinement Dimensions

We elucidate the vector space (twisted relative cohomology) that is Poincar\'e dual to the vector space of Feynman integrals (twisted cohomology) in general dimension. These spaces are paired via an inner product called the intersection number -- an invariant which can be computed algebraically. In this language, reduction of an integrand modulo integration-by-part identities is simply its projection onto a chosen basis. The dual-forms turn out to be far simpler than their Feynman counterparts; they are uplifts of lower-dimensional forms supported on the maximal cuts of various sub-topologies (boundaries). The intersection numbers are then computed by taking residues around cuts where various subsets of propagators become on-shell, giving a systematic approach to generalized unitarity. Moreover, the dual integrals satisfy the transposed differential equation to their Feynman counterparts. Since the dual-forms are localized to cut the transposed differential equation is a much simpler to construct.

Generalized Unitarity and the Poincaré Duals of Feynman Integrals

Laser-induced rotational wavepacket of hydrogen molecules has been experimentally observed in real time by using two sequential 25-fs laser pulses (pump-probe scheme) and a COLTRIMs spectrometer. By measuring the time-dependent yield of the above-threshold dissociation and the enhanced ionization of the molecule, we observed a few-femtosecond time delay in between the two dissociation pathways for both H2 and D2. The delay was understood and interpreted by a classical model considering enhanced ionization requires extra interaction with the laser field. We demonstrate the molecule rotational clock in hydrogen molecule is a straightforward method for timing ultrafast molecular dissociation dynamics.

Clocking enhanced ionization of hydrogen molecule using molecular rotational wavepackets

Currently, plasmonic nanofibers doped with semiconductor quantum dots, organic dye quantum emitters (QEs), and metallic nanoparticles (MNPs) have attracted much attention due to their wide range of applications including waveguides, light-sources, and optical sensors. These nanofibers doped with QEs and MNPs have been fabricated using a variety of metals and emitters. For example, Hu et al. [1] have studied the fabrication of a plasmonic random fiber from gold MNPs and pyrromethene dye molecules (QEs) embedded in the liquid core optical fiber. They found that a narrower and sharper photoluminescence (PL) spectrum can be more easily obtained when there is greater overlap between the plasmonic resonance of the gold-MNP and the dye molecules. Here we have developed a theory of photoluminescence for plasmonic nanofibers [2]. When probe light propagates inside the nanofiber, it induces surface plasmon polariton (SPPs) and electric dipoles in metallic nanoparticles. These dipoles interact with each other via the dipole-dipole interaction (DDI) [3]. The energy of photonic bound states in the presence of the SPP and DDI fields is then calculated. We have demonstrated that the number of bound states can be controlled by changing the strength of the SPP and DDI couplings. The expression of photoluminescence has been calculated using the density matrix method in the presence of the DDI coupling. We found that the intensity of the PL spectrum depends on the quality called quantum efficiency, which depends on the radiative and non-radiative decay rates. We have found that the quantum efficiency is enhanced when the exciton energy is in resonance with the bound photon energy. Further, we predicted that the PL intensity is also enhanced due to the DDI coupling. The enhancement of the PL spectrum can be used to fabricate plasmonic nanosensors. 

[1] Hu, Z et al., Gold nanoparticle-based plasmonic random fiber laser. J. Opt. 2015, 17, 35001.
[2] Singh, M. R.; Brassem, G.; Yastrebov, S. G. Optical quantum yield in plasmonic
Nanowaveguide. Annalen der Physik in press, 2021.
[3] Singh, M. R. The effect of the dipole–dipole interaction in electromagnetically induced transparency in polaritonic band gap materials. Journal of Modern Optics 2007, 54, 1739.

Study of photoluminescence in plasmonic nanoparticles

Graphene is an interesting system for condensed matter physicists because of many potential technological applications. It is interesting to theorists as an abelian strongly coupled system that has some of the interesting properties of QCD.

We are particularly interested in studying the effects of anisotropic strain of the graphene
lattice on the critical coupling, which determines the transition point for graphene
from a conductor to an insulator. We use a low energy effective theory and a Schwinger-Dyson approach. The full set of coupled equations is numerically very difficult to solve, and so previous calculations have made many simplifying assumptions in order to make the procedure tractable. These simplifications have led to disagreement about the effect anisotropy has on the critical coupling constant. We have studied the effect of some common approximations including polarization tensor approximations and vertex ansaetze. In this talk we discuss the motivation for several of these approximations, and show that in some cases their effect on the critical coupling is significant. We discuss more accurate ways to solve the non-perturbative low energy effective theory, and its ability to predict the critical coupling of the phase transition.

Semi-metal insulator phase transitions in graphene - the role of anisotropy

Organic molecules capable of forming strongly bound self-assembled monolayers (SAMs) offer a promising route to surface modification and functionalization. Technological advances including the protection and stabilization of nanoparticles and the development of lab-on-chip sensors have been largely realized by the adsorption of alkanethiols. *N*-heterocyclic carbenes (NHCs) were recently introduced as alternative anchors in the preparation of SAMs on metal surfaces and may be superior to thiol-based systems. NHCs have important advantages compared to thiol-based systems, most notably their chemical tunability and ability to form strongly bound monolayers with improved stability. Given their novelty in the field of materials science, many fundamental properties of NHC-based SAMs are not well-understood, including factors that control the adsorption and self-assembly processes.

By means of scanning tunneling microscopy and further supplemented by *ab initio* (DFT) and Monte Carlo simulations, we investigate the mechanisms by which NHCs bind and self-assemble on Au(111). Our investigation demonstrates a critical dependence of the binding mode on NHC structure, surface coverage, and substrate temperature. The dependence on these factors arises due to the minimization of molecule-surface steric interactions, the production of Au adatoms by the effect of NHC adsorption on the underlying gold substrate, and the reactivity and stability of the ligand, respectively. The main effect on the adsorption process is a competition between upright attachment geometries and the formation of flat-lying bis-NHC-metal complexes. These species form a rich variety of self-assembled structures, including in particular a zig-zag lattice formed by adatom-bound NHCs and a non-porous chiral Kagome lattice containing trapped NHC monomers. These lattices are analyzed within our three-pronged methodology, providing valuable insights into molecule--molecule and molecule-substate interactions, as well as the role of surface modifications and on-surface chemical reactions in the formation of NHC SAMs. Efforts to understand these processes were crucial in advancing the chemistry and applications of thiol-based SAMs, and will similarly offer critical strategies for the design of upright NHC SAMs and their applications.

N-heterocyclic carbene adsorption and self-assembly on Au(111): Fine-tuning the binding mode

One of the more exciting things to emerge from black hole thermodynamics in the past 10 years is the understanding that black holes can undergo a broad range of chemical-like phase transitions, including liquid-gas phase transitions, triple points, superfluid transitions, polymer-type transitions, and exhibit critical behaviour. It is even possible to consider black holes as the working material for heat engines. The efficiencies for a variety of black holes can be calculated and compared against each other.

In this talk I will discuss the connection between critical behaviour and the efficiency of black hole heat engines. I first consider the heat capacity of static black holes at constant volume such that Cv=0 Using the near critical expansion of the equation of state, the coefficients appearing in this expansion can be found from an engine cycle placed along a critical point on a PV plot. 

I will discuss the importance and applications of the simplifications made, along with how this result allows one to go from the near critical expansion of the equation of state directly to a conclusion about the behaviour of a heat engine near the critical point.

Black Hole Heat Engines and Critical Behaviour

One of the more exciting things to emerge from black hole thermodynamics is that black holes can form the working material for heat engines. I explore the connection between the critical behaviour of black holes and their efficiency as heat engines over a range of dimensions and for a variety of theories of gravity. 

I first show that their efficiency as heat engines near the critical point can be written in general dimensions in terms of the variables characterizing the geometry of the cycle and the critical exponents. Engines near the critical point approach the Carnot efficiency, with the rate of approach determined by the universality class of the black hole. I will specifically consider a broad range of charged black holes, Lovelock black holes, and black holes with isolated critical points. I will then discuss work in progress exploring this formalism for black holes whose specific heat at constant volume is nonzero, applying it to examples such as rotating black holes and STU black holes.

Efficiency of Black Hole Heat Engines and Universality

The common picture of a binary black hole merger is the “pair of pants” diagram for the event horizon. However, in many circumstances, such as those encountered in numerical simulations, the event horizon may be ill-suited and it is more practical to work with quasi-local definitions of black hole boundaries, such as marginally outer trapped surfaces (MOTS). The analog of the pair of pants diagram for the apparent horizons remains to be fully understood. In this talk, I will discuss the complete picture for the merger of two axisymmetric black holes. I will begin by introducing new classes of MOTS present in Brill-Lindquist initial data. I will then discuss the role played by these and related surfaces in understanding the final fate of the apparent horizons of the initial two participants in the merger.

The fate of apparent horizons in a binary black hole merger

I describe the first investigation of the holographic complexity conjectures for rotating black holes. Exploiting a simplification that occurs for equal-spinning odd dimensional black holes, I demonstrate a relationship between the complexity of formation and the thermodynamic volume associated with the black hole. This result suggests that it is thermodynamic volume and not entropy that governs the complexity of formation in both the Complexity Equals Volume and Complexity Equals Action proposals. This proposal reduces to known results involving the entropy in settings where the thermodynamic volume and entropy are not independent, but has much broader scope. Assuming the validity of a conjectured inequality for thermodynamic volume, this result suggests the complexity of formation is bounded from below by the entropy for large black holes.

Premium content

Prediction of Alzheimer's Disease Based on Multi-Attentional Mechanisms using Brain PET Image

Downloads

Next from CAP Virtual Congress 2021

Simulation of Two Polymers Confined to a Box-Like Cavity: The Effect of Anisotropies in Polymer Length and Confinement Dimensions

Similar lecture

Moving toward faster measurements of micron-sized axon diameters in vivo