Photovoltaic effect by soft phonon excitation.

SignificanceThe quantum-mechanical geometric phase of electrons provides various phenomena such as the dissipationless photocurrent generation through the shift current mechanism. So far, the photocurrent generations are limited to above or near the band-gap photon energy, which contradicts the increasing demand of the low-energy photonic functionality. We demonstrate the photocurrent through the optical phonon excitations in ferroelectric BaTiO3 by using the terahertz light with photon energy far below the band gap. This photocurrent without electron-hole pair generation is never explained by the semiclassical treatment of electrons and only arises from the quantum-mechanical geometric phase. The observed photon-to-current conversion efficiency is as large as that for electronic excitation, which can be well accounted for by newly developed theoretical formulation of shift current.

GGC repeat expansion in NOTCH2NLC induces dysfunction in ribosome biogenesis and translation.

GGC repeat expansion in the 5' untranslated region (UTR) of NOTCH2NLC is associated with a broad spectrum of neurological disorders, especially neuronal intranuclear inclusion disease (NIID). Studies have found that GGC repeat expansion in NOTCH2NLC induces the formation of polyglycine (polyG)-containing protein, which is involved in the formation of neuronal intranuclear inclusions. However, the mechanism of neurotoxicity induced by NOTCH2NLC GGC repeats is unclear. Here, we used NIID patient-specific induced pluripotent stem cell (iPSC)-derived 3D cerebral organoids (3DCOs) and cellular models to investigate the pathophysiological mechanisms of NOTCH2NLC GGC repeat expansion. IPSC-derived 3DCOs and cellular models showed the deposition of polyG-containing intranuclear inclusions. The NOTCH2NLC GGC repeats could induce the upregulation of autophagic flux, enhance integrated stress response and activate EIF2α phosphorylation. Bulk RNA sequencing for iPSC-derived neurons and single-cell RNA sequencing (scRNA-seq) for iPSC-derived 3DCOs revealed that NOTCH2NLC GGC repeats may be associated with dysfunctions in ribosome biogenesis and translation. Moreover, NOTCH2NLC GGC repeats could induce the NPM1 nucleoplasm translocation, increase nucleolar stress, impair ribosome biogenesis and induce ribosomal RNA sequestration, suggesting dysfunction of membraneless organelles in the NIID cellular model. Dysfunctions in ribosome biogenesis and phosphorylated EIF2α and the resulting increase in the formation of G3BP1-positive stress granules may together lead to whole-cell translational inhibition, which may eventually cause cell death. Interestingly, scRNA-seq revealed that NOTCH2NLC GGC repeats may be associated with a significantly decreased proportion of immature neurons while 3DCOs were developing. Together, our results underscore the value of patient-specific iPSC-derived 3DCOs in investigating the mechanisms of polyG diseases, especially those caused by repeats in human-specific genes.

CircLDLRAD3 inhibits Oral squamous cell carcinoma progression by regulating miR-558/Smad4/TGF-ß.

Oral squamous cell carcinoma (OSCC) is a malignant neoplasm with high mortality and morbidity. The role of circRNA and its molecular mechanism in OSCC remains largely unknown. The study aims to explore the role of a novel circular RNA (circLDLRAD3) in OSCC and its underlying mechanism. PCR and fluorescence in situ hybridization were used to explore the expression features of circLDLRAD3 in OSCC. The effects of circLDLRAD3 on the behaviour of OSCC were investigated using CCK-8, colony formation assay, transwell and animal experiments. Bioinformatics analysis along with dual luciferase reporter assay and RIP assay were used to reveal the interaction between circLDLRAD3, miR-558 and Smad4. It was revealed that circLDLRAD3 exhibited low expression status in OSCC. CircLDLRAD3 inhibits proliferation, migration, and invasion of OSCC cells both in vitro and in vivo. Mechanistically, circLDLRAD3 could bind with miR-558 to positively regulate its target gene Smad4 expression. Rescue experiments further confirmed both miR-558 overexpression and Smad4 knockdown could reverse the influence of circLDLRAD3 on OSCC phenotypes. Moreover, circLDLRAD3 regulate the TGF-ß signalling pathways to influence EMT through miR-558/Smad4 axis. Our study found that circLDLRAD3 is downregulated in OSCC and verified its tumour suppressor function and mechanism in OSCC through sponging miR-558 to regulate miR-558/Smad4/TGF-ß axis. The characterization of such regulating network uncovers an important mechanism underlying OSCC progression, which could provide promising targets targeted therapy strategies for OSCC in the future.

Meningeal immunity and neurological diseases: new approaches, new insights.

The meninges, membranes surrounding the central nervous system (CNS) boundary, harbor a diverse array of immunocompetent immune cells, and therefore, serve as an immunologically active site. Meningeal immunity has emerged as a key factor in modulating proper brain function and social behavior, performing constant immune surveillance of the CNS, and participating in several neurological diseases. However, it remains to be determined how meningeal immunity contributes to CNS physiology and pathophysiology. With the advances in single-cell omics, new approaches, such as single-cell technologies, unveiled the details of cellular and molecular mechanisms underlying meningeal immunity in CNS homeostasis and dysfunction. These new findings contradict some previous dogmas and shed new light on new possible therapeutic targets. In this review, we focus on the complicated multi-components, powerful meningeal immunosurveillance capability, and its crucial involvement in physiological and neuropathological conditions, as recently revealed by single-cell technologies.

Integrating single-nucleus sequence profiling to reveal the transcriptional dynamics of Alzheimer's disease, Parkinson's disease, and multiple sclerosis.

BACKGROUND: Alzheimer's disease (AD), Parkinson's disease (PD), and multiple sclerosis (MS) are three nervous system diseases that partially overlap clinically and genetically. However, bulk RNA-sequencing did not accurately detect the core pathogenic molecules in them. The availability of high-quality single cell RNA-sequencing data of post-mortem brain collections permits the generation of a large-scale gene expression in different cells in human brain, focusing on the molecular features and relationships between diseases and genes. We integrated single-nucleus RNA-sequencing (snRNA-seq) datasets of human brains with AD, PD, and MS to identify transcriptomic commonalities and distinctions among them. METHODS: The snRNA-seq datasets were downloaded from Gene Expression Omnibus (GEO) database. The Seurat package was used for snRNA-seq data processing. The uniform manifold approximation and projection (UMAP) were utilized for cluster identification. The FindMarker function in Seurat was used to identify the differently expressed genes. Functional enrichment analysis was carried out using the Gene Set Enrichment Analysis (GSEA) and Gene ontology (GO). The protein-protein interaction (PPI) analysis of differentially expressed genes (DEGs) was analyzed using STRING database ( http://string-db.org ). SCENIC analysis was performed using utilizing pySCENIC (v0.10.0) based on the hg19-tss-centered-10 kb-10species databases. The analysis of potential therapeutic drugs was analyzed on Connectivity Map ( https://clue.io ). RESULTS: The gene regulatory network analysis identified several hub genes regulated in AD, PD, and MS, in which HSPB1 and HSPA1A were key molecules. These upregulated HSP family genes interact with ribosome genes in AD and MS, and with immunomodulatory genes in PD. We further identified several transcriptional regulators (SPI1, CEBPA, TFE3, GRHPR, and TP53) of the hub genes, which has important implications for uncovering the molecular crosstalk among AD, PD, and MS. Arctigenin was identified as a potential therapeutic drug for AD, PD, and MS. CONCLUSIONS: Together, the integrated snRNA-seq data and findings have significant implications for unraveling the shared and unique molecular crosstalk among AD, PD, and MS. HSPB1 and HSPA1A as promising targets involved in the pathological mechanisms of neurodegenerative diseases. Additionally, the identification of arctigenin as a potential therapeutic drug for AD, PD, and MS further highlights its potential in treating these neurological disorders. These discoveries lay the groundwork for future research and interventions to enhance our understanding and treatment of AD, PD, and MS.

Time-Reversal-Even Nonlinear Current Induced Spin Polarization.

We propose a time-reversal-even spin generation in second order of electric fields, which dominates the current induced spin polarization in a wide class of centrosymmetric nonmagnetic materials, and leads to a novel nonlinear spin-orbit torque in magnets. We reveal a quantum origin of this effect from the momentum space dipole of the anomalous spin polarizability. First-principles calculations predict sizable spin generations in several nonmagnetic hcp metals, in monolayer TiTe_{2}, and in ferromagnetic monolayer MnSe_{2}, which can be detected in experiment. Our work opens up the broad vista of nonlinear spintronics in both nonmagnetic and magnetic systems.

Possible realization of optical Dirac points in woodpile photonic crystals.

The simulation of fermionic relativistic physics, e.g., Dirac and Weyl physics, has led to the discovery of many unprecedented phenomena in photonics, of which the optical-frequency realization is, however, still challenging. Here, surprisingly, we discover that the woodpile photonic crystals commonly used for optical frequency applications host exotic fermion-like relativistic degeneracies: a Dirac nodal line and a fourfold quadratic point, as protected by the nonsymmorphic crystalline symmetry. Deforming the woodpile photonic crystal leads to the emergence of type-II Dirac points from the fourfold quadratic point. Such type-II Dirac points can be detected by its anomalous refraction property which is manifested as a giant birefringence in a slab setup. Our findings provide a promising route towards 3D optical Dirac physics in all-dielectric photonic crystals.

Localized surface electromagnetic waves in CrI3-based magnetophotonic structures.

Resulting from strong magnetic anisotropy two-dimensional ferromagnetism was recently shown to be stabilized in chromium triiodide, CrI3, in the monolayer limit. While its properties remain largely unexplored, it provides a unique material-specific platform to unveil its electromagnetic properties associated with coupling of modes. Indeed, trigonal symmetry in the presence of out-of-plane magnetization results in a non-trivial structure of the conductivity tensor, including the off-diagonal terms. In this paper, we study the surface electromagnetic waves localized in a CrI3-based structure using the results of ab initio calculations for the CrI3 conductivity tensor. In particular, we provide an estimate for the critical angle corresponding to the surface plasmon polariton generation in the Kretschmann-Raether configuration by a detailed investigation of reflectance spectrum as well as the magnetic field distribution for different CrI3 layer thicknesses. We also study the bilayer structure formed by two CrI3 layers separated by a SiO2 spacer and show that the surface plasmon resonance can be achieved at the interface between CrI3 and air depending on the spacer thickness.

Higher-Order Weyl Semimetals.

Higher-order topology yields intriguing multidimensional topological phenomena, while Weyl semimetals have unconventional properties such as chiral anomaly. However, so far, Weyl physics remain disconnected with higher-order topology. Here, we report the theoretical discovery of higher-order Weyl semimetals and thereby the establishment of such an important connection. We demonstrate that higher-order Weyl semimetals can emerge in chiral materials such as chiral tetragonal crystals as the intermediate phase between the conventional Weyl semimetal and 3D higher-order topological phases. Higher-order Weyl semimetals manifest themselves uniquely by exhibiting concurrent chiral Fermi-arc surface states, topological hinge states, and the momentum-dependent fractional hinge charge, revealing a novel class of higher-order topological phases.

Efficacy and safety of anti-inflammatory agents for the treatment of major depressive disorder: a systematic review and meta-analysis of randomised controlled trials.

OBJECTIVES: To systematically review the efficacy and safety of anti-inflammatory agents for patients with major depressive disorders. METHODS: We searched the literature to identify potentially relevant randomised controlled trials (RCTs) up to 1 January 2019. The primary outcome was efficacy, measured by mean changes in depression score from baseline to endpoint. Secondary outcomes included response and remission rates and quality of life (QoL). Safety was evaluated by incidence of classified adverse events. Heterogeneity was examined using the I2 and Q statistic. Pooled standard mean differences (SMDs) and risk ratios (RRs) were calculated. Subgroup meta-analyses were conducted based on type of treatment, type of anti-inflammatory agents, sex, sponsor type and quality of studies. RESULTS: Thirty RCTs with 1610 participants were included in the quantitative analysis. The overall analysis pooling from 26 of the RCTs suggested that anti-inflammatory agents reduced depressive symptoms (SMD -0.55, 95% CI -0.75 to -0.35, I2=71%) compared with placebo. Higher response (RR 1.52, 95% CI 1.30 to 1.79, I2=29%) and remission rates (RR 1.79, 95% CI 1.29 to 2.49, I2=41%) were seen in the group receiving anti-inflammatory agents than in those receiving placebo. Subgroup analysis showed a greater reduction in symptom severity in both the monotherapy and adjunctive treatment groups. Subgroup analysis of non-steroidal anti-inflammatory drugs, omega-3 fatty acids, statins and minocyclines, respectively, disclosed significant antidepressant effects for major depressive disorder (MDD). For women-only trials, no difference in changes of depression severity was found between groups. Subanalysis stratified by sponsor type and study quality led to the same outcomes in favour of anti-inflammatory agents in both subgroups. Changes of QoL showed no difference between the groups. Gastrointestinal events were the only significant differences between groups in the treatment periods. CONCLUSIONS: Results of this systematic review suggest that anti-inflammatory agents play an antidepressant role in patients with MDD and are reasonably safe.

Efficacy and safety of calcitonin-gene-related peptide binding monoclonal antibodies for the preventive treatment of episodic migraine - an updated systematic review and meta-analysis.

BACKGROUND: Migraine is one of the most common neurological disorders that leads to disabilities. However, the conventional drug therapy for migraine might be unsatisfactory at times. Therefore, this meta-analysis aimed to evaluate the efficacy and safety of calcitonin-gene-related peptide binding monoclonal antibody (CGRP mAb) for the preventive treatment of episodic migraine, and provide high-quality clinical evidence for migraine therapy. METHODS: A systematic electronic database search was conducted to identify the potentially relevant studies. Two independent authors performed data extraction and quality appraisal. Mean difference (MD) and risk ratio (RR) were pooled for continuous and dichotomous data, respectively. The significance levels, weighted effect sizes and homogeneity of variance were calculated. RESULTS: Eleven high-quality randomized control trials that collectively included 4402 patients were included in this meta-analysis. Compared to placebo group, CGRP mAb therapy resulted in a reduction of monthly migraine days [weighted mean difference (WMD) = - 1.44, 95% CI = (- 1.68,- 1.19)] and acute migraine-specific medication days [WMD = - 1.28, 95% CI = (- 1.66,- 0.90)], with an improvement in 50% responder rate [RR = 1.51, 95% CI = (1.37,1.66)]. In addition, the adverse events (AEs) and treatment withdrawal rates due to AEs were not significantly different between CGRP mAb and placebo groups. Similar efficacy and safety results were obtained for erenumab, fremanezumab, and galcanezumab in subgroup analysis. CONCLUSIONS: The current body of evidence reveals that CGRP mAb is an effective and safe preventive treatment for episodic migraine.

Crystal Growth and Magnetic Properties of Topological Nodal-Line Semimetal GdSbTe with Antiferromagnetic Spin Ordering.

We report crystal growth, AC and DC magnetic susceptibilities [χ(T, H)], magnetization [M(T, H)], and heat capacity [CP(T, H)] measurement results of GdSbTe single crystal. GdSbTe is isostructural to the confirmed nonmagnetic nodal-line semimetal ZrSiS of noncentrosymmetric tetragonal crystal structure in space group P4/nmm (No. 129), but it shows additional long-range antiferromagnetic spin ordering for the Gd spins of S = 7/2 below TN. Both χ(T, H) and CP(T, H) measurements confirm the existence of a long-range antiferromagnetic (AFM) spin ordering of Gd spins below TN ∼ 12 K, and an additional spin reorientation/recovery (sr) behavior is identified from the change of on-site spin anisotropy between Tsr1 ∼ 7 and Tsr2 ∼ 4 K. The anisotropic magnetic susceptibilities of χ(T, H) below TN clearly demonstrate that the AFM long-range spin ordering of GdSbTe has an easy axis parallel to the ab-plane direction. The field- and orientation-dependent magnetization of M(T, H) at 2 K shows two plateaus to indicate the spin-flop transition for H||ab near ∼2.1 T and a metamagnetic state near ∼5.9 T having ∼1/3 of the fully polarized magnetization by the applied field. The heat capacity measurement results yield Sommerfeld coefficient of γ ∼ 7.6(4) mJ/mol K2 and θD ∼ 195(2) K being less than half of that for the nonmagnetic ZrSiS. A three-dimensional (3D) AFM spin structure is supported by the ab initio calculations for Gd having magnetic moment of 7.1 µB and the calculated AFM band structure indicates that GdSbTe is a semimetal with bare density of states (0.36 states/eV fu) at the Fermi level, which is 10 times smaller than the measured one to suggest strong spin-fluctuation.

Skeletal Extracellular Matrix Supports Cardiac Differentiation of Embryonic Stem Cells: a Potential Scaffold for Engineered Cardiac Tissue.

BACKGROUND/AIMS: Decellularized cardiac extracellular matrix (cECM) has been widely considered as an attractive scaffold for engineered cardiac tissue (ECT), however, its application is limited by immunogenicity and shortage of organ donation. Skeletal ECM (sECM) is readily available and shows similarities with cECM. Here we hypothesized that sECM might be an alternative scaffold for ECT strategies. METHODS: Murine ventricular tissue and anterior tibial muscles were sectioned into 300 mm-thick, and then cECM and sECM were acquired by pretreatment/SDS/TritonX-100 three-step-method. Acellularity and morphological properties of ECM was assessed. SECM was recellularized with murine embryonic stem cells (mESCs) or mESC-derived cardiomyocytes (mESC-CMs), and was further studied by biocompatibility assessment, immunofluorescent staining, quantitative real-time PCR and electrophysiological experiment. RESULTS: The relative residual contents of DNA, protein and RNA of sECM were comparable with cECM. The morphological properties and microstructure of sECM were similar to cECM. SECM supported mESCs to adhere, survive, proliferate and differentiate into functional cardiac microtissue with both electrical stimulated response and normal adrenergic response. Purified mESC-CMs also could adhere, survive, proliferate and form a sECM-based ECT with synchronized contraction within 6 days of recellularization. CONCLUSION: ECMs from murine skeletal muscle support survival and cardiac differentiation of mESCs, and are suitable to produce functional ECT patch. This study highlights the potential of patient specific of sECM to replace cECM for bioengineering ECT.

Tuning spontaneous radiation of chiral molecules by asymmetric chiral nanoparticles.

We have obtained analytical expressions for the radiative decay rate of the spontaneous emission of a chiral molecule located near a dielectric spherical particle with a chiral nonconcentric spherical shell made of a bi-isotropic material. Our numerical and graphical analyses show that material composition, thickness and degree of non-concentricity of the shell can influence significantly the spontaneous radiation of the chiral molecule. In particular, the radiative decay rates can differ in orders of magnitude for a chiral molecule located near the thin and thick parts of a nonconcentric shell as well as near a concentric shell made of chiral metamaterial. We also find that the radiative decay rates of the "right" and "left" chiral molecule enantiomers located near a nanoparticle with a chiral metamaterial shell can differ pronouncedly from each other. Our findings therefore suggest a way to tune the spontaneous emission of chiral molecules by varying the material composition, thickness and degree of non-concentricity of the shell in the nearby composite nanoparticle and also to enhance the chirality selection of chiral molecules in racemic mixtures.

Predicted Quantum Topological Hall Effect and Noncoplanar Antiferromagnetism in K_{0.5}RhO_{2}.

The quantum anomalous Hall (QAH) phase is a two-dimensional bulk ferromagnetic insulator with a nonzero Chern number in the presence of spin-orbit coupling (SOC) but in the absence of applied magnetic fields. Associated metallic chiral edge states host dissipationless current transport in electronic devices. This intriguing QAH phase has recently been observed in magnetic impurity-doped topological insulators, albeit, at extremely low temperatures. Based on first-principles density functional calculations, here we predict that layered rhodium oxide K_{0.5}RhO_{2} in the noncoplanar chiral antiferromagnetic state is an unconventional three-dimensional QAH insulator with a large band gap and a Néel temperature of a few tens of Kelvins. Furthermore, this unconventional QAH phase is revealed to be the exotic quantum topological Hall effect caused by nonzero scalar spin chirality due to the topological spin structure in the system and without the need of net magnetization and SOC.

Magnetic Orderings in Li2Cu(WO4)2 with Tungstate-Bridged Quasi-1D Spin-1/2 Chains.

By both experimental measurements and theoretical calculations, we investigated the magnetic and electronic properties of Li2Cu(WO4)2 as a tungstate-bridged quasi-one-dimensional (1D) copper spin-(1/2) chain system. Interestingly, magnetic susceptibility χ(T) and specific heat measurements show that the system undergoes a three-dimensional antiferromagnetic (AF)-like ordering at TN ≈ 3.7 K, below a broad χ(T) maximum at ∼8.9 K indicating a low-dimensional short-range AF spin correlation. Bonner-Fisher model fitting of χ(T) leads to an AF intrachain exchange constant of J/kB = 15.8 ± 0.1 K, and mean-field theory estimation gives an interchain coupling constant of J⊥/kB = 1.6 K, which supports the quasi-1D nature of this spin system. Theoretical evaluation of exchange coupling constants within the generalized gradient approximation (GGA) plus on-site Coulomb interaction (U) shows that the dominant AF exchange interaction is of ∼13.9 K along the a-axis with weak interchain coupling, in agreement with the experimental result of a quasi-1D spin-(1/2) chain system. The GGA+U calculations also predict that Li2Cu(WO4)2 is a charge transfer-type AF semiconductor with a direct band gap of 1.5 eV.

Stable isotope metabolic labeling-based quantitative phosphoproteomic analysis of Arabidopsis mutants reveals ethylene-regulated time-dependent phosphoproteins and putative substrates of constitutive triple response 1 kinase.

Ethylene is an important plant hormone that regulates numerous cellular processes and stress responses. The mode of action of ethylene is both dose- and time-dependent. Protein phosphorylation plays a key role in ethylene signaling, which is mediated by the activities of ethylene receptors, constitutive triple response 1 (CTR1) kinase, and phosphatase. To address how ethylene alters the cellular protein phosphorylation profile in a time-dependent manner, differential and quantitative phosphoproteomics based on (15)N stable isotope labeling in Arabidopsis was performed on both one-minute ethylene-treated Arabidopsis ethylene-overly-sensitive loss-of-function mutant rcn1-1, deficient in PP2A phosphatase activity, and a pair of long-term ethylene-treated wild-type and loss-of-function ethylene signaling ctr1-1 mutants, deficient in mitogen-activated kinase kinase kinase activity. In total, 1079 phosphopeptides were identified, among which 44 were novel. Several one-minute ethylene-regulated phosphoproteins were found from the rcn1-1. Bioinformatic analysis of the rcn1-1 phosphoproteome predicted nine phosphoproteins as the putative substrates for PP2A phosphatase. In addition, from CTR1 kinase-enhanced phosphosites, we also found putative CTR1 kinase substrates including plastid transcriptionally active protein and calcium-sensing receptor. These regulatory proteins are phosphorylated in the presence of ethylene. Analysis of ethylene-regulated phosphosites using the group-based prediction system with a protein-protein interaction filter revealed a total of 14 kinase-substrate relationships that may function in both CTR1 kinase- and PP2A phosphatase-mediated phosphor-relay pathways. Finally, several ethylene-regulated post-translational modification network models have been built using molecular systems biology tools. It is proposed that ethylene regulates the phosphorylation of arginine/serine-rich splicing factor 41, plasma membrane intrinsic protein 2A, light harvesting chlorophyll A/B binding protein 1.1, and flowering bHLH 3 proteins in a dual-and-opposing fashion.

Absolute quantitation of isoforms of post-translationally modified proteins in transgenic organism.

Post-translational modification isoforms of a protein are known to play versatile biological functions in diverse cellular processes. To measure the molar amount of each post-translational modification isoform (P(isf)) of a target protein present in the total protein extract using mass spectrometry, a quantitative proteomic protocol, absolute quantitation of isoforms of post-translationally modified proteins (AQUIP), was developed. A recombinant ERF110 gene overexpression transgenic Arabidopsis plant was used as the model organism for demonstration of the proof of concept. Both Ser-62-independent (14)N-coded synthetic peptide standards and (15)N-coded ERF110 protein standard isolated from the heavy nitrogen-labeled transgenic plants were employed simultaneously to determine the concentration of all isoforms (T(isf)) of ERF110 in the whole plant cell lysate, whereas a pair of Ser-62-dependent synthetic peptide standards were used to quantitate the Ser-62 phosphosite occupancy (R(aqu)). The P(isf) was finally determined by integrating the two empirically measured variables using the following equation: P(isf) = T(isf) · R(aqu). The absolute amount of Ser-62-phosphorylated isoform of ERF110 determined using AQUIP was substantiated with a stable isotope labeling in Arabidopsis-based relative and accurate quantitative proteomic approach. The biological role of the Ser-62-phosphorylated isoform was demonstrated in transgenic plants.

Pixel Distillation: Cost-flexible Distillation across Image Sizes and Heterogeneous Networks.

Previous knowledge distillation (KD) methods mostly focus on compressing network architectures, which is not thorough enough in deployment as some costs like transmission bandwidth and imaging equipment are related to the image size. Therefore, we propose Pixel Distillation that extends knowledge distillation into the input level while simultaneously breaking architecture constraints. Such a scheme can achieve flexible cost control for deployment, as it allows the system to adjust both network architecture and image quality according to the overall requirement of resources. Specifically, we first propose an input spatial representation distillation (ISRD) mechanism to transfer spatial knowledge from large images to student's input module, which can facilitate stable knowledge transfer between CNN and ViT. Then, a Teacher-Assistant-Student (TAS) framework is further established to disentangle pixel distillation into the model compression stage and input compression stage, which significantly reduces the overall complexity of pixel distillation and the difficulty of distilling intermediate knowledge. Finally, we adapt pixel distillation to object detection via an aligned feature for preservation (AFP) strategy for TAS, which aligns output dimensions of detectors at each stage by manipulating features and anchors of the assistant. Comprehensive experiments on image classification and object detection demonstrate the effectiveness of our method.

Position-based anchor optimization for point supervised dense nuclei detection.

Nuclei detection is one of the most fundamental and challenging problems in histopathological image analysis, which can localize nuclei to provide effective computer-aided cancer diagnosis, treatment decision, and prognosis. The fully-supervised nuclei detector requires a large number of nuclei annotations on high-resolution digital images, which is time-consuming and needs human annotators with professional knowledge. In recent years, weakly-supervised learning has attracted significant attention in reducing the labeling burden. However, detecting dense nuclei of complex crowded distribution and diverse appearances remains a challenge. To solve this problem, we propose a novel point-supervised dense nuclei detection framework that introduces position-based anchor optimization to complete morphology-based pseudo-label supervision. Specifically, we first generate cellular-level pseudo labels (CPL) for the detection head via a morphology-based mechanism, which can help to build a baseline point-supervised detection network. Then, considering the crowded distribution of the dense nuclei, we propose a mechanism called Position-based Anchor-quality Estimation (PAE), which utilizes the positional deviation between an anchor and its corresponding point label to suppress low-quality detections far from each nucleus. Finally, to better handle the diverse appearances of nuclei, an Adaptive Anchor Selector (AAS) operation is proposed to automatically select positive and negative anchors according to morphological and positional statistical characteristics of nuclei. We conduct comprehensive experiments on two widely used benchmarks, MO and Lizard, using ResNet50 and PVTv2 as backbones. The results demonstrate that the proposed approach has superior capacity compared with other state-of-the-art methods. In particularly, in dense nuclei scenarios, our method can achieve 95.1% performance of the fully-supervised approach. The code is available at https://github.com/NucleiDet/DenseNucleiDet.