32 innovations from Bar-Ilan University, available for licensing, co-investment, or spin-out through BIRAD.
Lindell Yehuda
Impersonation Detection
Tischler Yaakov Raphael
We previously introduced a method and device for significantly enhancing the resolution of Raman spectroscopy measurements by using angle tuning of a Fabry-Perot (F-P) etalon in the beam path of a standard grating-based dispersive Raman spectrometer. Building on this innovation, we propose a novel configuration where Raman filters are placed after the F-P etalon. This configuration allows the F-P etalon to interact with both the coherent laser line and the excited Raman signal, enabling simultaneous measurement of the laser's spectral peak position and linewidth along with the stimulated Raman peak. By leveraging this dual measurement, we achieve simultaneous super-spectral-resolution of the Raman peak and the laser's spectral position, leading to an enhanced-resolution and determination of the absolute Raman peak shift, with both the laser and Raman peaks being super-resolved simultaneously. The method relies on computationally reconstructing the peak positions and linewidths of both the laser excitation and Raman scattering by comparing their angle-dependent intensity spectra to a physical model. The reconstruction provides ultra-high resolution for the linewidths and positions of both laser and Raman peaks in the same measurement, which can be achieved either by using a wide-enough range dispersive grating or by separately measuring the reflected or scattered laser signal with a photodetector. When analyzing a substance with a known Raman spectrum, this dual-measurement approach enables highly precise scattering spectroscopy using much more compact instrumentation. By combining the angle-dependent spectra from the Fabry-Perot etalon with a physical model, this method offers a streamlined and cost-effective solution for high-resolution spectral analysis, making it particularly advantageous for applications requiring compact and efficient setups. When analyzing a substance with a known single Raman peak, this dual-measurement approach enables highly precise scattering spectroscopy using just a pair of photodetectors, effectively eliminating the need for a full spectrometer. By combining the angle-dependent spectra from the Fabry-Perot etalon with a physical model, this method offers a streamlined and cost-effective solution for high-resolution spectral analysis, making it particularly advantageous for applications requiring super-compact and efficient setups.
Garini Yuval
Analyzing stained tissue sections is of major importance for pathological diagnostics, and it forms a bottleneck in various clinical procedures, including cancer detection. Although there are now systems that can scan whole biopsies, they only measure color (RGB data) that provides limited information for reliable and accurate analysis. We invented a new modality for cancer analysis that is based on rapid spectral imaging measurement of whole biopsy, followed by adequate analysis. The spectral information at each pixel of the image is valuable and it allows to perform accurate analysis by using adequate algorithms. Using the system, we also identified the spectra of normal and cancerous cells from a lymph node sample of breast cancer biopsy. Using this information, we developed an algorithm that allows to identify cancer cells with high sensitivity and specificity. The method combines hardware for rapid scan of biopsies followed by specific way for detecting cancer from the measured spectral images.
Salomon Adi
The META-SERS-NET substrates described herein provide a versatile, scalable, and highly sensitive platform for Surface-Enhanced Raman Scattering (SERS) detection of a wide range of organic and inorganic analytes in aqueous environments. Owing to their three-dimensional nanostructured architecture, broad electromagnetic enhancement, negligible background, and compatibility with lightweight machine-learning (ML) models, these metasurfaces are suitable for multiple commercial and industrial applications, including but not limited to: Environmental Monitoring and Water Quality Control META-SERS-NET substrates are designed for rapid, label-free detection of organic contaminants (e.g., dyes, pesticides, pharmaceuticals) and inorganic ions (e.g., Li⁺, Mg²⁺, B⁺, Na⁺) directly in water. The high enhancement factor, combined with the ability to detect analytes at concentrations down to 10⁻⁹ M, enables: Salomon_SERS_Spec_V3-M Online and offline monitoring of drinking water quality. Surveillance of industrial effluents, wastewater treatment plants, and surface waters. Early detection of persistent organic pollutants and heavy-metal–related species when coupled with appropriate ion-selective polymers. Portable and Field-Deployable Sensing Devices The META-SERS-NET architecture maintains strong SERS performance even when used with low-NA optics (e.g., NA = 0.15), which are typical in portable and handheld Raman instruments. Salomon_SERS_Spec_V3-M This optical tolerance enables: Integration into handheld Raman probes for in-field environmental monitoring. Compact sensors for on-site industrial process control, pipeline monitoring, and spill detection. Low-cost, battery-operated point-of-use devices for municipalities, utilities, and emergency response teams. Industrial Process Control and Quality Assurance The reproducible enhancement and negligible substrate background of META-SERS-NET allow robust quantification of analytes across several orders of magnitude in concentration. Salomon_SERS_Spec_V3-M This makes the platform suitable for: Real-time tracking of dyes, intermediates, and by-products in chemical and pharmaceutical production. Quality control in manufacturing processes requiring precise monitoring of residual contaminants. Inline or at-line sensors for continuous verification of feedstocks, solvents, and process streams. Food and Beverage Safety By enabling sensitive detection of trace dyes, adulterants, and ionic species, META-SERS-NET can be incorporated into: Screening tools for contaminants in beverages and liquid food matrices (e.g., juices, dairy, brewing lines). Rapid verification of cleaning and sanitation processes via detection of residual chemicals in rinse water. Biomedical and Clinical Research Tools (Non-diagnostic Use) In research settings, META-SERS-NET can serve as a high-performance SERS platform for: Studying drug–polymer and ion–polymer interactions using the polymer-assisted detection mode. Salomon_SERS_Spec_V3-M Investigating model bio-relevant ions and small molecules in simulated physiological media. Serving as a robust reference substrate for SERS method development in analytical and bioanalytical laboratories. AI-Augmented Analytical Platforms The integration of the metasurface with a dedicated machine-learning spectral reconstruction module provides enhanced peak separation, noise suppression, and analyte classification. Salomon_SERS_Spec_V3-M This joint optical–computational architecture enables: Automated, high-accuracy detection and quantification of multiple analytes in complex mixtures. Cloud-connected or on-device AI-SERS platforms for routine monitoring tasks operated by non-experts. “Smart” SERS instruments that self-calibrate using the internal Si reference and adapt to device-specific spectral distortions. Calibration Standards and Reference Substrates Due to their ligand-free, additive-free fabrication and high reproducibility over large areas, META-SERS-NET substrates can function as: Salomon_SERS_Spec_V3-M Standard SERS reference substrates for instrument calibration and inter-laboratory comparisons. Internal standards in commercial Raman instruments, ensuring consistent performance across devices and over time. Long-Lifetime, Low-Maintenance Sensing Modules The metasurfaces exhibit lifetimes of at least one year under standard storage and operating conditions and demonstrate efficient heat dissipation and stability at low excitation powers. Salomon_SERS_Spec_V3-M This durability supports: Long-term deployment in remote or difficult-to-access locations. Low-maintenance sensor cartridges for subscription-based monitoring services. Replacement-ready “plug-and-measure” chips that can be exchanged in field devices without complex recalibration.
Louzoun Yoram
The effect of microbes on their human host is often mediated through changes in metabolite concentrations. As such, multiple tools have been proposed to predict metabolitc profiles from microbial taxa frequencies, assuming a direct relation between the gut microbiome composition and blood metabolite concentrations. However, the microbiome-metabolite relation may depend on host demographics or condition. We show that the relation between microbiome and metabolites is best predicted at the log concentration level. We further develop LOCATE (Latent Of miCrobiome And meTabolites rElations), a machine learning (ML) tool based on latent representation which predicts the log normalized metabolites composition based on the log normalized microbiome composition. LOCATE has a higher overall accuracy than all current state-of-the-art predictors in both 16S rRNA gene and shotgun gene sequencing. The accuracy of LOCATE and all other predictors significantly decreases when predicting on one dataset and testing on a different dataset, or on a different condition in the same dataset, especially in 16S rRNA gene sequence based data. We propose an intermediate representation between the microbiome and the metabolite concentrations and show that this representation can be used to predict the host phenotype better than either the microbiome or the metabolome. This representation is strongly correlated with host demographics, including age, gender and diet and can be used to improve ML predictions of host phenotypes in comparison with either microbiome or metabolome using a large microbiome sample combined with a small number of metabolome samples (~ 50)
Shamay Meir
Distal cis-regulatory elements, such as enhancers and silencers, dictate tissue-specific complex transcriptional repertoire in an orientation- and position-independent manner. Herpesviruses show programmed latent and lytic gene expression based on the infected tissue and physiological cell state. In a recent study we systematically identified the enhancers within the Kaposi’s Sarcoma-associated Herpesvirus (KSHV) genome. Here, we present ENHAvir, an NLM-based tool that can successfully predict the enhancers in a viral genome. We used the DeBERTa v3 language model56 in our framework. DeBERTa v3 is an encoder-style language model, making it a suitable candidate for extracting representations that could be used for new tasks. ENHAvir successfully identified known and novel enhancer elements in the herpesviruses, namely, EBV, HSV1, HSV2, VZV, HCMV, HHV-6, HSV-7, and MHV68. ENHAvir learned the minute patterns of previously published KSHV enhancers and their adjacent sequences responsible for enhancer looping. The activity of the predicted enhancers was validated by cloning the ENHAvir predicted sequences on the EBV genome downstream to the luciferase gene in a reporter plasmid with a weak promoter and performing dual-luciferase reporter assays in EBV-infected and uninfected cell lines. Interestingly, ENHAvir also precisely identified enhancers with the human genome, and examples for Fos, Jun, DPPA3, and Myc genes are presented. The ability of ENHAvir to predict both viral and cellular enhancers, provides an additional layer to the complex gene regulation via viral enhancers but also points out the evolutionary conservation of enhancer micro-signatures between a virus and its host.
Ozana, Nisan
Vocal fold paralysis (VFP) is characterized by impaired vocal fold movement, commonly resulting from nerve damage during surgical procedures. Current diagnostic methods rely on endoscopic examinations requiring specialized physicians, reducing accessibility and potentially delaying treatment. We propose a non-contact optical sensing method using speckle pattern analysis for VFP identification. Our approach uses external laser illumination and a camerathatcapturesspecklepatterns,providinganon-invasiveandreal-timeassessment. The techniqueusesspectralanalysisenhancedbyslidingwindowscanningtoextractamplitudepeaks across vocal fold regions.
Naveh Doron
A computational spectrometer system based on a voltage-tunable GeSe-InSe heterojunction device that addresses nonlinear photoresponses through piecewise linearization using classification and hierarchical clustering models for accurate spectral reconstruction.
Magnezi Racheli
We have developed a model designed to predict re hospitalization in premature infants on the day of their discharge from the neonatal intensive care unit (NICU). The model's objective is to identify preterm infants who are likely to be readmitted during their first year of life from NICU discharge. This early identification enables a reduction in the multiple hospitalization days associated with these infants, estimated at a total of $3.3 million annually for a single hospital. The model aims to reduce the economic burden and focus on preventive medicine in Israel. The uniqueness of this model lies in its ability to present approximately 20 new parameters that have not been previously shown in parallel models in the literature. The predictive capability of the developed model has an accuracy of 0.79.
Zalevsky Zeev
Intraocular pressure (IOP) measurements comprise an essential tool in modern medicine for the early diagnosis of glaucoma, the second leading cause of human blindness. The world's highest prevalence of glaucoma is in low-income countries. Current diagnostic methods require experience in running expensive equipment as well as the use of anesthetic eye drops. We present herein a remote photonic IOP biomonitoring method based on deep learning of secondary speckle patterns, captured by a fast camera, that are reflected from eye sclera stimulated by an external sound wave. By combining speckle pattern analysis with deep learning, high precision measurements are possible. The method was tested under artificially varying eye pressures on a series of 24 pig eyeballs, found to be similar to human eyes. As a low-cost procedure, it has the potential to meet clinical needs in low- and middle-income countries and at points of care everywhere.
Singer Gonen
Transformers play a central role in modern artificial intelligence, yet their susceptibility to adversarial perturbations raises serious reliability concerns. Current defenses, such as adversarial training, are often computationally expensive and tailored to specific attack vectors, while formal verification methods remain difficult to scale. In this work, we propose RAHP (Robustness-Aware Head Pruning), a framework that enhances the intrinsic robustness of Transformers by selectively removing attention heads that contribute to model adversarial vulnerability. Unlike standard pruning, which often degrades robustness, RAHP guides the pruning process using a composite score of two complementary signals: (i) Fisher Information, which preserves task accuracy, and (ii) CLEVER, a sensitivity-based proxy derived from the local Lipschitz constant that estimates the model’s vulnerability to perturbations. By pruning attention heads that exhibit high adversarial sensitivity, RAHP steers the model toward a more stable decision boundary without the need for costly adversarial retraining. Extensive experiments demonstrate that RAHP yields compact models that are not only efficient but also more resistant to a wide variety of attacks compared to standard pruning and regularization baselines. These results suggest that incorporating local stability criteria into the pruning process provides a scalable and attack-agnostic pathway toward robust and efficient Transformer models.
Cohen Cyrille
The invention involves the computational design and development of novel constant regions for human T-cell receptors (TCRs), termed "Structurally Enhanced TCR" (SET). By introducing a set of strategic mutations into the TCR constant domains, the SET design improves receptor stability, enhances surface expression, increases functional avidity, and ensures preferential pairing of the α and β chains, minimizing mispairing with endogenous TCRs. This innovation offers a universal platform to optimize T-cell therapies for cancer and infectious diseases without requiring additional gene editing.
