HSG-MGAF Net: Heterogeneous subgraph-guided multiscale graph attention fusion network for interpretable prediction of whole-slide image.

BACKGROUND AND OBJECTIVE: Pathological whole slide image (WSI) prediction and region of interest (ROI) localization are important issues in computer-aided diagnosis and postoperative analysis in clinical applications. Existing computer-aided methods for predicting WSI are mainly based on multiple instance learning (MIL) and its variants. However, most of the methods are based on instance independence and identical distribution assumption and performed at a single scale, which not fully exploit the hierarchical multiscale heterogeneous information contained in WSI. METHODS: Heterogeneous Subgraph-Guided Multiscale Graph Attention Fusion Network (HSG-MGAF Net) is proposed to build the topology of critical image patches at two scales for adaptive WSI prediction and lesion localization. The HSG-MGAF Net simulates the hierarchical heterogeneous information of WSI through graph and hypergraph at two scales, respectively. This framework not only fully exploits the low-order and potential high-order correlations of image patches at each scale, but also leverages the heterogeneous information of the two scales for adaptive WSI prediction. RESULTS: We validate the superiority of the proposed method on the CAMELYON16 and the TCGA- NSCLC, and the results show that HSG-MGAF Net outperforms the state-of-the-art method on both datasets. The average ACC, AUC and F1 score of HSG-MGAF Net can reach 92.7 %/0.951/0.892 and 92.2 %/0.957/0.919, respectively. The obtained heatmaps can also localize the positive regions more accurately, which have great consistency with the pixel-level labels. CONCLUSIONS: The results demonstrate that HSG-MGAF Net outperforms existing weakly supervised learning methods by introducing critical heterogeneous information between the two scales. This approach paves the way for further research on light weighted heterogeneous graph-based WSI prediction and ROI localization.

High-performance terahertz modulators induced by substrate field in Te-based all-2D heterojunctions.

High-performance active terahertz modulators as the indispensable core components are of great importance for the next generation communication technology. However, they currently suffer from the tradeoff between modulation depth and speed. Here, we introduce two-dimensional (2D) tellurium (Te) nanofilms with the unique structure as a new class of optically controlled terahertz modulators and demonstrate their integrated heterojunctions can successfully improve the device performances to the optimal and applicable levels among the existing all-2D broadband modulators. Further photoresponse measurements confirm the significant impact of the stacking order. We first clarify the direction of the substrate-induced electric field through first-principles calculations and uncover the unusual interaction mechanism in the photoexcited carrier dynamics associated with the charge transfer and interlayer exciton recombination. This advances the fundamental and applicative research of Te nanomaterials in high-performance terahertz optoelectronics.

Exploring terahertz spectral characteristics of L-sorbose and D-melibiose in solid and liquid states.

Saccharides are essential organic compounds that perform critical functions in sustaining life processes. As biomolecules, their vibrational frequencies predominantly fall in the terahertz (THz) range, making them amenable to analysis using THz techniques. In this study, L-sorbose and D-melibiose were measured using a THz time-domain spectroscopy system covering a frequency range of 0.1-2.0 THz, and their crystal structures were simulated using density functional theory. The experimental results demonstrated significant agreement with the simulation findings. In addition, the spectral properties of the two saccharides in solution were determined using microfluidic chip technology, thereby facilitating a comparison between the solid and aqueous states. The results demonstrate that the intramolecular and intermolecular interactions of the saccharides were weakened by the presence of water molecules, and the THz absorption spectrum of the same substance solution was found to be correlated with its concentration and temperature.

Vibrational spectroscopic detection and analysis of salicylic acid and aspirin binary cocrystal.

Salicylic acid is a raw material for preparing aspirin and holds an important position in medical history. Studying the crystallization of these two drugs is of great significance in improving their dissolution rate, bioavailability, and physical stability. Although various techniques have been used for structural characterization, there is still a lack of information on the collective vibrational behavior of aspirin and salicylic acid eutectic compounds. Firstly, two starting materials (salicylic acid and aspirin) were ground in a 1:1 M ratio to prepare eutectic compounds. The eutectic composition was studied using vibrational spectroscopy techniques, such as X-ray powder diffusion (XRPD), terahertz time-domain spectroscopy (THz-TDS), and Raman spectroscopy. Additionally, the structure of the aspirin and salicylic acid eutectic was simulated and optimized using density functional theory. It was found that the eutectic type II was the most consistent with the experiment, and the corresponding vibration modes of each peak were provided. These results offer a unique method for characterizing the structural composition of eutectic crystals, which can be utilized to enhance the physical and chemical properties, as well as the pharmacological activity, of specific drugs at the molecular level.

Dual-Band and Multi-State Polarization Conversion Using aTerahertz Symmetry-Breaking Metadevice.

We numerically and experimentally demonstrate a terahertz metadevice consisting of split-ring resonators (SRRs) present within square metallic rings. This device can function as a dual-band polarization converter by breaking the symmetry of SRRs. Under x-polarized incidence, the metastructure is able to convert linearly polarized (LP) light into a left-hand circular-polarized (LCP) wave. Intriguingly, under y-polarized incidence, frequency-dependent conversion from LP to LCP and right-hand circular-polarized (RCP) states can be achieved at different frequencies. Furthermore, reconfigurable LCP-to-LP and RCP-to-LP switching can be simulated by integrating the device with patterned graphene and changing its Fermi energy. This dual-band and multi-state polarization control provides an alternative solution to developing compact and multifunctional components in the terahertz regime.

Interpretable Inference and Classification of Tissue Types in Histological Colorectal Cancer Slides Based on Ensembles Adaptive Boosting Prototype Tree.

Digital pathology images are treated as the "gold standard" for the diagnosis of colorectal lesions, especially colon cancer. Real-time, objective and accurate inspection results will assist clinicians to choose symptomatic treatment in a timely manner, which is of great significance in clinical medicine. However, Manual methods suffers from long inspection cycle and serious reliance on subjective interpretation. It is also a challenging task for existing computer-aided diagnosis methods to obtain models that are both accurate and interpretable. Models that exhibit high accuracy are always more complex and opaque, while interpretable models may lack the necessary accuracy. Therefore, the framework of ensemble adaptive boosting prototype tree is proposed to predict the colorectal pathology images and provide interpretable inference by visualizing the decision-making process in each base learner. The results showed that the proposed method could effectively address the "accuracy-interpretability trade-off" issue by ensemble of m adaptive boosting neural prototype trees. The superior performance of the framework provides a novel paradigm for interpretable inference and high-precision prediction of pathology image patches in computational pathology.

Magnetic induced terahertz modulation characteristics based on ferromagnetic nematic liquid crystals.

In recent years, solid state terahertz (THz) modulators have obtained rapid progress with the widespread use of two-dimensional (2D) materials in the field of THz; however, challenges remain in preparing flexible THz modulators. In this study, flexible ferromagnetic nematic materials were prepared by doping thermotropic nematic liquid crystals 5CB into magnetic fluids, and the influence of water was reduced by a self-made cyclic olefin copolymer (COC) microfluidic chip. THz modulation characteristics of magnetic fluid and ferromagnetic nematic liquid crystal (FNLC) under the induction of external magnetic field were compared using a THz time domain spectroscopy system. Under the action of a 91 mT magnetic field, the magnetic fluid has a maximum modulation depth (MD) of 54%. Under the same magnetic field, the maximum MD of the FNLC materials increase to 78% because of the rearrangement of Fe3O4 nanoparticles induced by the topological defect of the liquid crystal. We demonstrate that the magneto-optical effect is significantly enhanced in the ferromagnetic nematic liquid crystal hybrid system. This strategy of doping thermotropic nematic liquid crystals to enhance the magneto-optical effect has great potential for THz filtering, modulation, and sensing applications.

Interpretable classification of pathology whole-slide images using attention based context-aware graph convolutional neural network.

BACKGROUND AND OBJECTIVE: Whole slide image (WSI) classification and lesion localization within giga-pixel slide are challenging tasks in computational pathology that requires context-aware representations of histological features to adequately infer nidus. The existing weakly supervised learning methods mainly treat different locations in the slide as independent regions and cannot learn potential nonlinear interactions between instances based on i.i.d assumption, resulting in the model unable to effectively utilize context-ware information to predict the labels of WSIs and locate the region of interest (ROI). METHODS: Here, we propose an interpretable classification model named bidirectional Attention-based Multiple Instance Learning Graph Convolutional Network (ABMIL-GCN), which hierarchically aggregates context-aware features of instances into a global representation in a topology fashion to predict the slide labels and localize the region of lymph node metastasis in WSIs. RESULTS: We verified the superiority of this method on the Camelyon16 dataset, and the results show that the average predicted ACC and AUC of the proposed model after flooding optimization can reach 90.89% and 0.9149, respectively. The average accuracy and ACC score are improved by more than 7% and 4% compared with the existing state-of-the-art algorithms. CONCLUSIONS: The results demonstrate that context-aware GCN outperforms existing weakly supervised learning methods by introducing spatial correlations between the neighbor image patches, which also addresses the 'accuracy-interpretability trade-off' problem. The framework provides a novel paradigm for the clinical application of computer-aided diagnosis and intelligent systems.

Terahertz Kerr Effect of Liquids.

In recent years, tremendous advancements have been made in various technologies such as far-infrared, low-frequency Raman, and two-dimensional (2D) Raman terahertz (THz) spectroscopies. A coherent method has emerged from numerous experimental and theoretical investigations of molecular dynamics in liquids by comparing linear and non-linear spectroscopic techniques. Intermolecular hydrogen bond vibration, molecular reorientation motion, and interaction between molecule/ionic solute and hydrogen bonds have been demonstrated to occur in the THz region, which are closely related to their physical/chemical properties and structural dynamics. However, precise probing of various modes of motion is difficult because of the complexity of the collective and cooperative motion of molecules and spectral overlap of related modes. With the development of THz science and technology, current state-of-the-art THz sources can generate pulsed electric fields with peak intensities of the order of microvolts per centimeter (MV/cm). Such strong fields enable the use of THz waves as the light source for non-linear polarization of the medium and in turn leads to the development of the emerging THz Kerr effect (TKE) technique. Many low-frequency molecular motions, such as the collective directional motion of molecules and cooperative motion under the constraint of weak intermolecular interactions, are resonantly excited by an intense THz electric field. Thus, the TKE technique provides an interesting prospect for investigating low-frequency dynamics of different media. In view of this, this paper first summarizes the research work on TKE spectroscopy by taking a solid material without low-frequency molecular motions as an example. Starting from the principle of TKE technology and its application in investigating the properties of solid matter, we have explored the low-frequency molecular dynamics of liquid water and aqueous solutions using TKE. Liquid water is a core of life and possesses many extraordinary physical and biochemical properties. The hydrogen bond network plays a crucial role in these properties and is the main reason for its various kinetic and thermodynamic properties, which differ from those of other liquids. However, the structure of the hydrogen bond network between water and solutes is not well known. Therefore, evaluating the hydrogen bond-related kinetic properties of liquid water is important.

Terahertz spectroscopic detection of amino acid molecules under magnetic field.

Terahertz (THz) waves can cover the characteristic spectra of substances such as plasma, organisms, and biomolecules, whereas THz photons have low energy and do not damage biological tissues. Therefore, its absorption characteristics in the THz region can be used to characterize the internal structure of biomolecules. In this study, we designed a microfluidic chip and combined it with THz technology. The spectral intensity in descending order was found to be deionized water, phenylalanine, histidine, glycine and glutamic acid by observing the THz wave transmission in the range of 0.1-1.0 THz, comparing the frequency domain spectra of four amino acid solutions with volume fraction of 2% and deionized water. It is inferred that different molecular structures of amino acids resulted in different numbers of hydrogen bonds formed between them and water molecules, leading to different degrees of absorption of THz waves. In addition, magnetic fields parallel to the THz wave transmission were used to study the variation of different amino acids with magnetic field intensity. It is found that increasing the magnetic field strength decrease the transmission of THz waves. This is because under the action of the magnetic field, on the one hand, the hydrogen bonds formed by water molecules are strengthened and the absorption of THz waves is enhanced; on the other hand, amino acid molecules aggregate and the radius of molecular clusters increases, thus blocking the transmission of THz waves. Finally, we also calculated the electric conductivity of the solutions to prove the accuracy of the experimental results from a theoretical point of view.

Highly sensitive detection of broadband terahertz waves using aqueous salt solutions.

Water-based coherent detection of broadband terahertz (THz) wave has been recently proposed with superior performances, which can alleviate the limited detection bandwidth and high probe laser energy requirement in the solid- and air-based detection schemes, respectively. Here, we demonstrate that the water-based detection method can be extended to the aqueous salt solutions and the sensitivity can be significantly enhanced. The THz coherent detection signal intensity scales linearly with the third-order nonlinear susceptibility χ(3) or quadratically with the linear refractive index η0 of the aqueous salt solutions, while the incoherent detection signal intensity scales quadratically with χ(3) or quartically with η0, proving the underlying mechanism is the four-wave mixing. Both the coherent and incoherent detection signal intensities appear positive correlation with the solution concentration. These results imply that the liquid-based THz detection scheme could provide a new technique to measure χ(3) and further investigate the physicochemical properties in the THz band for various liquids.

Terahertz modulation characteristics of three nanosols under external field control based on microfluidic chip.

Recently, with the widespread application of metamaterials in the terahertz (THz) modulation field, solid-state THz modulators have made breakthrough progress; however, there are still challenges in preparing flexible THz modulators with wide modulation bandwidths. In this study, a THz microfluidic chip was fabricated using cycloolefin copolymers with high transmission (90%) of THz waves. The THz modulation characteristics of TiO2, Ag, and Fe3O4 nanosols under the control of an optical field, electric field, and magnetic field, respectively, were investigated. Under the action of photogenerated carrier migration, colloidal electrophoresis, and magneto-optical effect, all three nanosols exhibit broadband modulation performance in the frequency range of 0.3-2.4 THz, and the maximum modulation depth is 24%, 33%, and 54%, respectively. Contrary to previous studies based on traditional solid-state materials, this study innovatively explores the possibility of modulating THz waves with liquid materials, laying the foundation for the application of flexible liquid-film THz modulators.

Terahertz absorption characteristics of ammonium salt solution based on self-sampling microfluidic chip.

With the continuous development of terahertz (THz) detection technology, the use of terahertz spectroscopy to study chemical samples has become one of the indispensable tools in the field of biochemistry. While most biomolecules biological activity can only be expressed in aqueous solutions, water as a polar molecule has strong absorption properties for terahertz waves, making it difficult to use terahertz technology to study the activity of biological samples in aqueous solutions. In this study, a sandwich-type terahertz microfluidic chip with high terahertz wave transmission was designed and combined with a terahertz time domain spectroscopy (THz-TDS) system to test the terahertz spectra of distilled water, 0.9 mol/L NH4Cl, (NH4)2SO4, (NH4)2CO3 and CH3COONH4 solutions, respectively, and to investigate the effect of the electric field action time on the hydrogen bond in the solution under the action of an external electric field. The experimental results show that the terahertz spectra of different ammonium solutions at the same concentration differ significantly, indicating that the ion hydration process affects the intermolecular hydrogen bonding in water, while the applied electric field also affects the hydrogen bonding in water, resulting in a change in the terahertz waves water absorption.

Terahertz spectral properties of glucose and two disaccharides in solid and liquid states.

The vibrational and rotational frequencies of most biological macromolecules fall within the terahertz (THz) band; therefore, the THz wave has a strong ability to distinguish substances. Saccharides are important organic substances and the main source of life-sustaining activities. In this study, the spectral characteristics of D-glucose, α-lactose hydrate, and ß-maltose hydrate were measured in the solid state through THz time-domain spectroscopy in the frequency range of 0.1-2.5 THz. The crystal configurations of these three saccharides were then simulated using solid-state density functional theory, and the experimental results were found to be in good agreement with the simulation results. Furthermore, the spectral characteristics of the three saccharides in solutions were measured. Each saccharide was found to have unique spectral characteristics, and a correlation existed between the THz absorption spectra of the same substance in the solid state and aqueous solution.

Water-Based Coherent Detection of Broadband Terahertz Pulses.

Both solids and gases have been demonstrated as the materials for terahertz (THz) coherent detection. The gas-based coherent detection methods require a high-energy probe laser beam and the detection bandwidth is limited in the solid-based methods. Whether liquids can be used for THz detection and relax these problems has not yet been reported, which becomes a timely and interesting topic due to the recent observation of efficient THz wave generation in liquids. Here, we propose a THz coherent detection scheme based on liquid water. When a THz pulse and a fundamental laser beam are mixed on a free-flowing water film, a second harmonic (SH) beam is generated as the plasma is formed. Combining this THz-induced SH beam with a control SH beam, we successfully achieve the time-resolved waveform of the THz field with the frequency range of 0.1-18 THz. The required probe laser energy is as low as a few microjoules. The sensitivity of our scheme is 1 order of magnitude higher than that of the air-based method under comparable detection conditions. The scheme is sensitive to the THz polarization and the phase difference between the fundamental and control SH beams, which brings direct routes for optimization and polarization sensitive detection. Energy scaling and polarization properties of the THz-induced beam indicate that its generation can be attributed to a four-wave mixing process. This generation mechanism makes simple relationships among the probe laser, THz-induced SH, and THz field, favorable for robustness and flexibility of the detection device.

A novel strategy regarding geometric product for liquids discrimination based on THz reflection spectroscopy.

A novel expression of geometric product that associated with the geometric relationship from geometric algebra constructed by the vectorized refractive index and absorption coefficient in THz region is proposed, which could provide a new insight into the THz properties of materials. From the novel expression, the candidate characteristic parameters are extracted for liquids discrimination and present the abundant second order correlation information of optical parameters with the consideration of dimension rising. Three groups of liquids, containing C-reactive protein calibrators and alpha fetoprotein calibrators, were selected as examples to validate the feasibility of the proposed strategy. Comparing with the traditional THz parameters, including refractive index, absorption coefficient, and complex permittivity, the novel approach exhibits notable superiority for differentiation with the evaluation of statistical differences and effect sizes. The proposed geometric product expression could have a large potential on promoting the substance identification in some applications of THz technology.

Ultrahigh Modulation Enhancement in All-Optical Si-Based THz Modulators Integrated with Gold Nanobipyramids.

Optical regulation strategy with the aid of hybrid materials can significantly optimize the performance of terahertz devices. Gold nanobipyramids (AuNBPs) with synthetical tunability to the near-infrared band show strong local field enhancement, which improves optical coupling at the interface and benefits the modulation performance. We design AuNBPs-integrated terahertz modulators with multiple structured surfaces and demonstrate that introducing AuNBPs can effectively enhance their modulation depths. In particular, an ultrahigh modulation enhancement of 1 order of magnitude can be achieved in the AuNBPs hybrid metamaterials accompanied by the multifunctional modulation characteristics. By application of the coupled Lorentz oscillator model, the theoretical calculation suggests that the optical regulation with AuNBPs originates from increased damping rate and higher coupling coefficient under pump excitation. Additionally, a terahertz spatial light modulator is constructed to demonstrate multiple imaging display and consume extremely low power, which is promising for the potential application in spatial and frequency selective imaging.

Quantitative determination of glycerol concentration in aqueous glycerol solutions by metamaterial-based terahertz spectroscopy.

Glycerol is an important quality indicator for foodstuffs. There is an increasing request for one more accurate, reliable, and convenient detection of the glycerol concentrations. Terahertz radiation is highly sensitive to the low-frequency intermolecular interactions between the glycerol and waters. Considering the enhancement property of localized field from the metamaterials, terahertz spectroscopy has been utilized for the determination of glycerol content with metamaterial-based biosensor, where the interaction between the analyte and the terahertz wave can be greatly increased. But the quantitative sensing performance was poor due to the sensitivity limitation of single-mode resonance of metamaterial and the lack of appropriate modeling methods. We propose the optimized structural design with internal coupling and multiple resonances. The induced remarkable changes in the lineshape of different transmitted dip regions imply that our metastructure biosensor is of high sensitivity to the change of surrounding environment on the surface dielectric constant, which has been also verified by coupled Lorentz oscillator theory. Furthermore, the optimal partial least squares regression model with variables of spectral lineshape for the first dip region covering the frequency range of 0.45-0.85 THz was established. It shows more accurate and reliable predictions of glycerol concentrations with residual predictive deviation value of 6.095. Metamaterial-based terahertz spectroscopy combined with statistical modeling with lineshape features can provide one new strategy for quantitative sensing. It has great potential for the improvement of determination of analyte concentrations in the practical applications of food, pharmaceutical or cosmetic area.

Layer-Resolving Terahertz Light-Field Imaging Based on Angular Intensity Filtering Method.

Terahertz focal plane array imaging methods, direct camera imaging and conventional light field imaging methods are incapable of resolving and separating layers of multilayer objects. In this paper, for the purpose of fast, high-resolution and layer-resolving imaging of multilayer structures with different reflection characteristics, a novel angular intensity filtering (AIF) method based on terahertz light-field imaging is purposed. The method utilizes the extra dimensional information from the 4D light field and the reflection characteristics of the imaging object, and the method is capable to resolve and reconstruct layers individually. The feasibility of the method is validated by experiment on both "idealized" and "practical" multilayer samples, and the advantages in performance of the method are proven by quantitative comparison with conventional methods.

Molecular dynamic investigation of ethanol-water mixture by terahertz-induced Kerr effect.

The terahertz Kerr effect (TKE) spectroscopy provides time-resolved measurement of low-frequency molecular motions of liquids. Here, the intense broadband terahertz (THz) pulses resonantly excite multiple molecular modes in pure ethanol and ethanol-water mixtures. For pure ethanol, the obtained unipolar TKE response contains the molecular relaxation information extending over tens of picoseconds, which originates from the coupling between the permanent molecular dipole moment of ethanol and the THz electric field. For ethanol-water mixtures with different molar proportions, the results observed on the sub-picosecond time scale can always be divided into the linear superposition of the TKE signals of pure ethanol and water. Under the observation time window over tens of picoseconds (after 1 picosecond), the relative molecular contribution of ethanol in the mixture changes nonlinearly with the increase of water molecules, implying the complex structural perturbation of ethanol hydrogen bond network in the mixture. This work provides a new perspective for further investigation on the hydrogen bond network structure and dynamics in aqueous amphiphilic solutions.