Deep Learning Classification Research Articles

Disorder-specific neurodynamic features in schizophrenia inferred by neurodynamic embedded contrastive variational autoencoder model

Neurodynamic models that simulate how micro-level alterations propagate upward to impact macroscopic neural circuits and overall brain function may offer valuable insights into the pathological mechanisms of schizophrenia (SCZ). In this study, we integrated a neurodynamic model with the classical Contrastive Variational Autoencoder (CVAE) to extract and evaluate macro-scale SCZ-specific features, including subject-level, region-level parameters, and time-varying states. Firstly, we demonstrated the robust fitting of the model within our multi-site dataset. Subsequently, by employing representational similarity analysis and a deep learning classifier, we confirmed the specificity and disorder-related information capturing ability of SCZ-specific features. Moreover, analysis of the attractor characteristics of the neurodynamic system revealed significant differences in attractor space patterns between SCZ-specific states and shared states. Finally, we utilized Partial Least Squares (PLS) regression to examine the multivariate mapping relationship between SCZ-specific features and symptoms, identifying two sets of correlated modes implicating unique molecular mechanisms: one mode corresponding to negative and general symptoms, and another mode corresponding to positive symptoms. Our results provide valuable insights into disorder-specific neurodynamic features and states associated with SCZ, laying the foundation for understanding the intricate pathophysiology of this disorder.

Leveraging AI and Data Visualization for Enhanced Policy-Making: Aligning Research Initiatives with Sustainable Development Goals

Scientists, research institutions, funding agencies, and policy-makers have all emphasized the need to monitor and prioritize research investments and outputs to support the achievement of the United Nations Sustainable Development Goals (SDGs). Unfortunately, many current and historic research publications, proposals, and grants were not categorized against the SDGs at the time of submission. Manual post hoc classification is time-consuming and prone to human biases. Even when classified, few tools are available to decision makers for supporting resource allocation. This paper aims to develop a deep learning classifier for categorizing research abstracts by the SDGs and a decision support system for research funding policy-makers. First, we fine-tune a Bidirectional Encoder Representations from Transformers (BERT) model using a dataset of 15,488 research abstracts from authors at leading Brazilian universities, which were preprocessed and balanced for training and testing. Second, we present a PowerBI dashboard that visualizes classifications for supporting informed resource allocation for sustainability-focused research. The model achieved an F1-score, precision, and recall exceeding 70% for certain classes and successfully classified existing projects, thereby enabling better tracking of Agenda 2030 progress. Although the model is capable of classifying any text, it is specifically optimized for Brazilian research due to the nature of its fine-tuning data.

Intersection of Performance, Interpretability, and Fairness in Neural Prototype Tree for Chest X-Ray Pathology Detection: Algorithm Development and Validation Study.

While deep learning classifiers have shown remarkable results in detecting chest X-ray (CXR) pathologies, their adoption in clinical settings is often hampered by the lack of transparency. To bridge this gap, this study introduces the neural prototype tree (NPT), an interpretable image classifier that combines the diagnostic capability of deep learning models and the interpretability of the decision tree for CXR pathology detection. This study aimed to investigate the utility of the NPT classifier in 3 dimensions, including performance, interpretability, and fairness, and subsequently examined the complex interaction between these dimensions. We highlight both local and global explanations of the NPT classifier and discuss its potential utility in clinical settings. This study used CXRs from the publicly available Chest X-ray 14, CheXpert, and MIMIC-CXR datasets. We trained 6 separate classifiers for each CXR pathology in all datasets, 1 baseline residual neural network (ResNet)-152, and 5 NPT classifiers with varying levels of interpretability. Performance, interpretability, and fairness were measured using the area under the receiver operating characteristic curve (ROC AUC), interpretation complexity (IC), and mean true positive rate (TPR) disparity, respectively. Linear regression analyses were performed to investigate the relationship between IC and ROC AUC, as well as between IC and mean TPR disparity. The performance of the NPT classifier improved as the IC level increased, surpassing that of ResNet-152 at IC level 15 for the Chest X-ray 14 dataset and IC level 31 for the CheXpert and MIMIC-CXR datasets. The NPT classifier at IC level 1 exhibited the highest degree of unfairness, as indicated by the mean TPR disparity. The magnitude of unfairness, as measured by the mean TPR disparity, was more pronounced in groups differentiated by age (chest X-ray 14 0.112, SD 0.015; CheXpert 0.097, SD 0.010; MIMIC 0.093, SD 0.017) compared to sex (chest X-ray 14 0.054 SD 0.012; CheXpert 0.062, SD 0.008; MIMIC 0.066, SD 0.013). A significant positive relationship between interpretability (ie, IC level) and performance (ie, ROC AUC) was observed across all CXR pathologies (P<.001). Furthermore, linear regression analysis revealed a significant negative relationship between interpretability and fairness (ie, mean TPR disparity) across age and sex subgroups (P<.001). By illuminating the intricate relationship between performance, interpretability, and fairness of the NPT classifier, this research offers insightful perspectives that could guide future developments in effective, interpretable, and equitable deep learning classifiers for CXR pathology detection.

EEG-based action anticipation in human-robot interaction: a comparative pilot study.

As robots become integral to various sectors, improving human-robot collaboration is crucial, particularly in anticipating human actions to enhance safety and efficiency. Electroencephalographic (EEG) signals offer a promising solution, as they can detect brain activity preceding movement by over a second, enabling predictive capabilities in robots. This study explores how EEG can be used for action anticipation in human-robot interaction (HRI), leveraging its high temporal resolution and modern deep learning techniques. We evaluated multiple Deep Learning classification models on a motor imagery (MI) dataset, achieving up to 80.90% accuracy. These results were further validated in a pilot experiment, where actions were accurately predicted several hundred milliseconds before execution. This research demonstrates the potential of combining EEG with deep learning to enhance real-time collaborative tasks, paving the way for safer and more efficient human-robot interactions.

Deep Multiple Instance Learning Model to Predict Outcome of Pancreatic Cancer Following Surgery

Background/Objectives: Pancreatic ductal adenocarcinoma (PDAC) is a cancer with very poor prognosis despite early surgical management. To date, only clinical variables are used to predict outcome for decision-making about adjuvant therapy. We sought to generate a deep learning approach based on hematoxylin and eosin (H&amp;E) or hematoxylin, eosin and saffron (HES) whole slides to predict patients’ outcome, compare these new entities with known molecular subtypes and question their biological significance; Methods: We used as a training set a retrospective private cohort of 206 patients treated by surgery for PDAC cancer and a validation cohort of 166 non-metastatic patients from The Cancer Genome Atlas (TCGA) PDAC project. We estimated a multi-instance learning survival model to predict relapse in the training set and evaluated its performance in the validation set. RNAseq and exome data from the TCGA PDAC database were used to describe the transcriptomic and genomic features associated with deep learning classification; Results: Based on the estimation of an attention-based multi-instance learning survival model, we identified two groups of patients with a distinct prognosis. There was a significant difference in progression-free survival (PFS) between these two groups in the training set (hazard ratio HR = 0.72 [0.54;0.96]; p = 0.03) and in the validation set (HR = 0.63 [0.42;0.94]; p = 0.01). Transcriptomic and genomic features revealed that the poor prognosis group was associated with a squamous phenotype. Conclusions: Our study demonstrates that deep learning could be used to predict PDAC prognosis and offer assistance in better choosing adjuvant treatment.

Hybrid feature extraction and integrated deep learning for cloud-based malware detection

The escalating prevalence of malware necessitates a proactive and vigilant approach to its detection and mitigation. The ramifications of a successful malware attack on cloud services can be severe, underscoring the critical importance of effective malware detection mechanisms in cloud environments. To address this pressing need, we propose a comprehensive methodology for creating a novel cloud-based malware dataset, namely the CMD_2024 dataset. This dataset integrates static and dynamic attributes, providing a robust framework for malware analysis. The CMD_2024 dataset, comprising 20,850 samples meticulously labeled into various malware categories such as Virus, Trojan, Worm, Ransomware, Adware, Miner, PUA, and Downloader, is designed to facilitate the testing and evaluation of diverse analysis tools, machine learning models, deep learning models, and security systems. We enhance the dataset’s utility and effectiveness by focusing on dynamic features, particularly system calls within the cloud, in conjunction with static attributes. To address the challenges of the imbalance towards less common malware categories in the dataset, we employed the Conditional Tabular Generative Adversarial Network to generate synthetic data, significantly improving the detection capability for these rare malware samples. The application of various machine learning and deep learning classifiers, including our proposed integrated deep learning models, yielded remarkable results, achieving 99.42% accuracy in binary classification and 86.97% in multi-class classification. These outcomes demonstrate the CMD_2024 dataset’s substantial efficacy in supporting robust malware detection within cloud environments.

Development and validation of the MRI-based deep learning classifier for distinguishing perianal fistulizing Crohn’s disease from cryptoglandular fistula: a multicenter cohort study

A singular reliable modality for early distinguishing perianal fistulizing Crohn's disease (PFCD) from cryptoglandular fistula (CGF) is currently lacking. We aimed to develop and validate an MRI-based deep learning classifier to effectively discriminate between them. The present study retrospectively enrolled 1054 patients with PFCD or CGF from three Chinese tertiary referral hospitals between January 1, 2015, and December 31, 2021. The patients were divided into four cohorts: training cohort (n=800), validation cohort (n=100), internal test cohort (n=100) and external test cohort (n=54). Two deep convolutional neural networks (DCNN), namely MobileNetV2 and ResNet50, were respectively trained using the transfer learning strategy on a dataset consisting of 44871MR images. The performance of the DCNN models was compared to that of radiologists using various metrics, including receiver operating characteristic curve (ROC) analysis, accuracy, sensitivity, and specificity. Delong testing was employed for comparing the area under curves (AUCs). Univariate and multivariate analyses were conducted to explore potential factors associated with classifier performance. A total of 532 PFCD and 522 CGF patients were included. Both pre-trained DCNN classifiers achieved encouraging performances in the internal test cohort (MobileNetV2 AUC: 0.962, 95% CI 0.903-0.990; ResNet50 AUC: 0.963, 95% CI 0.905-0.990), as well as external test cohort (MobileNetV2 AUC: 0.885, 95% CI 0.769-0.956; ResNet50 AUC: 0.874, 95% CI 0.756-0.949). They had greater AUCs than the radiologists (all p≤0.001), while had comparable AUCs to each other (p=0.83 and p=0.60 in the two test cohorts). None of the potential characteristics had a significant impact on the performance of pre-trained MobileNetV2 classifier in etiologic diagnosis. Previous fistula surgery influenced the performance of the pre-trained ResNet50 classifier in the internal test cohort (OR 0.157, 95% CI 0.025-0.997, p=0.05). The developed DCNN classifiers exhibited superior robustness in distinguishing PFCD from CGF compared to artificial visual assessment, showing their potential for assisting in early detection of PFCD. Our findings highlight the promising generalized performance of MobileNetV2 over ResNet50, rendering it suitable for deployment on mobile terminals. National Natural Science Foundation of China.

Automatic detection and classification of beluga whale calls in the St. Lawrence estuary.

The endangered beluga whale (Delphinapterus leucas) of the St. Lawrence Estuary (SLEB) faces threats from a variety of anthropogenic factors. Since belugas are a highly social and vocal species, passive acoustic monitoring has the potential to deliver, in a non-invasive and continuous way, real-time information on SLEB spatiotemporal habitat use, which is crucial for their monitoring and conservation. In this study, we introduce an automatic pipeline to analyze continuous passive acoustic data and provide standard and accurate estimations of SLEB acoustic presence and vocal activity. An object detector extracted vocalizations of beluga whales from an acoustic recording of beluga vocal activity. Then, two deep learning classifiers discriminated between high-frequency call types (40-120 kHz) and the presence of low-frequency components (0-20 kHz), respectively. Different algorithms were tested for each step and their main combinations were compared in time and performance. We focused our work on a high residency area, Baie Sainte-Marguerite (BSM), used for socialization and feeding by SLEB. Overall, this project showed that accurate continuous analysis of SLEB vocal activity at BSM could provide valuable information to estimate habitat use, link beluga behavior and acoustic activity within and between herds, and quantify beluga presence and abundance.

Predicting antimicrobial resistance in E. coli with discriminative position fused deep learning classifier

Escherichia coli (E. coli) has become a particular concern due to the increasing incidence of antimicrobial resistance (AMR) observed worldwide. Using machine learning (ML) to predict E. coli AMR is a more efficient method than traditional laboratory testing. However, further improvement in the predictive performance of existing models remains challenging. In this study, we collected 1,937 high-quality whole genome sequencing (WGS) data from public databases with an antimicrobial resistance phenotype and modified the existing workflow by adding an attention mechanism to enable the modified workflow to focus more on core single nucleotide polymorphisms (SNPs) that may significantly lead to the development of AMR in E. coli. While comparing the model performance before and after adding the attention mechanism, we also performed a cross-comparison among the published models using random forest (RF), support vector machine (SVM), logistic regression (LR), and convolutional neural network (CNN). Our study demonstrates that the discriminative positional colors of Chaos Game Representation (CGR) images can selectively influence and highlight genome regions without prior knowledge, enhancing prediction accuracy. Furthermore, we developed an online tool (https://github.com/tjiaa/E.coli-ML/tree/main) for assisting clinicians in the rapid prediction of the AMR phenotype of E. coli and accelerating clinical decision-making.

Multimodal Explainability Using Class Activation Maps and Canonical Correlation for MI-EEG Deep Learning Classification

Brain–computer interfaces (BCIs) are essential in advancing medical diagnosis and treatment by providing non-invasive tools to assess neurological states. Among these, motor imagery (MI), in which patients mentally simulate motor tasks without physical movement, has proven to be an effective paradigm for diagnosing and monitoring neurological conditions. Electroencephalography (EEG) is widely used for MI data collection due to its high temporal resolution, cost-effectiveness, and portability. However, EEG signals can be noisy from a number of sources, including physiological artifacts and electromagnetic interference. They can also vary from person to person, which makes it harder to extract features and understand the signals. Additionally, this variability, influenced by genetic and cognitive factors, presents challenges for developing subject-independent solutions. To address these limitations, this paper presents a Multimodal and Explainable Deep Learning (MEDL) approach for MI-EEG classification and physiological interpretability. Our approach involves the following: (i) evaluating different deep learning (DL) models for subject-dependent MI-EEG discrimination; (ii) employing class activation mapping (CAM) to visualize relevant MI-EEG features; and (iii) utilizing a questionnaire–MI performance canonical correlation analysis (QMIP-CCA) to provide multidomain interpretability. On the GIGAScience MI dataset, experiments show that shallow neural networks are good at classifying MI-EEG data, while the CAM-based method finds spatio-frequency patterns. Moreover, the QMIP-CCA framework successfully correlates physiological data with MI-EEG performance, offering an enhanced, interpretable solution for BCIs.

Comparing neural networks against click train detectors to reveal temporal trends in passive acoustic sperm whale detections.

Passive acoustic monitoring (PAM) is an increasingly popular tool to study vocalising species. The amount of data generated by PAM studies calls for robust automatic classifiers. Deep learning (DL) techniques have been proven effective in identifying acoustic signals in challenging datasets, but due to their black-box nature their underlying biases are hard to quantify. This study compares human analyst annotations, a multi-hypothesis tracking (MHT) click train classifier and a DL-based acoustic classifier to classify acoustic recordings based on the presence or absence of sperm whale (Physeter macrocephalus) click trains and study the temporal and spatial distributions of the Mediterranean sperm whale subpopulation around the Balearic Islands. The MHT and DL classifiers showed agreements with human labels of 85.7% and 85.0%, respectively, on data from sites they were trained on, but both saw a drop in performance when deployed on a new site. Agreement rates between classifiers surpassed those between human experts. Modeled seasonal and diel variations in sperm whale detections for both classifiers showed compatible results, revealing an increase in occurrence and diurnal activity during the summer and autumn months. This study highlights the strengths and limitations of two automatic classification algorithms to extract biologically useful information from large acoustic datasets.

A Comparative Study of Deep Learning and Transfer Learning Approaches for Image Forgery Detection

The rapid growth of online activities such as commerce, education, research, and virtual conferences has led to a greater reliance on digital images as primary information sources on social media and other platforms. The extensive use, combined with the ease of modification via image-editing software, highlights the crucial need for effective image forgery detection tools. Traditional detection methods based on handcrafted features have grown less efficient, prompting the introduction of deep learning-based approaches, many of which combine transfer learning with pre-trained models to improve detection efficiency and shorten training time. This research presents a comprehensive evaluation of image forgery detection algorithms, categorizing them as classical, deep learning, and transfer learning frameworks. The study compares deep learning with transfer learning methods, assessing their strengths in feature extraction, classification, and detection accuracy. The findings indicate that, while transfer learning models are particularly effective at feature extraction using pre-trained architectures, deep learning remains superior for classification tasks. This insight intends to help academics construct high-accuracy, efficient models for detecting various forms of forgeries. Combining pre-trained models for feature extraction and deep learning for classification is the best option for real-time digital forensics, increasing detection accuracy and processing speed.

Deep Metric Learning with Augmented Latent Fusion and Response-Based Knowledge Distillation on Edge Device for Paddy Pests and Disease Identification

The health of paddy fields significantly impacts rice yields and the economic stability of farmers. Limited number of experts available to watch these issues poses a challenge. Consequently, a reliable diagnostic system is necessary to find pests and diseases in rice crops. In this study, we propose deep metric learning with augmented latent fusion (FADMAKA) combined with a response-based knowledge distillation (KD) approach. The student model, which processes single RGB input images, is trained using soft latent labels derived from four augmented input from the teacher model. Our method delivers a high validation accuracy of 0.973, keeps an accuracy of 0.782 on the unseen data, and with rapid inference time of 38.911 milliseconds. This approach’s accuracy outperforms SoftMax deep learning classification with fine-tuning, which only has a maximum accuracy of 0.739 on the unseen data with computation time of 36.224 ms, and the DML with augmented latent fusion with k-NN classifier on the same base model, which achieves an accuracy of 0.78 with computation time of 124.977 ms. Our proposed model has 0.12 giga floating point operations per second (GFLOPs) that is suitable for edge devices with low computational resources. Following the modeling phase, we deployed the highest-accuracy student model to a Raspberry Pi 4B device equipped with a camera. This system can provide biological agent-based recommendations for identified pest and disease threats in rice fields. Our approach not only improved accuracy but also proved efficiency, enabling farmers to identify pests and disease without relying on internet connectivity.

Producing plankton classifiers that are robust to dataset shift

AbstractModern plankton high‐throughput monitoring relies on deep learning classifiers for species recognition in water ecosystems. Despite satisfactory nominal performances, a significant challenge arises from dataset shift, which causes performances to drop during deployment. In our study, we integrate the ZooLake dataset, which consists of dark‐field images of lake plankton (Kyathanahally et al. 2021a), with manually annotated images from 10 independent days of deployment, serving as test cells to benchmark out‐of‐dataset (OOD) performances. Our analysis reveals instances where classifiers, initially performing well in in‐dataset conditions, encounter notable failures in practical scenarios. For example, a MobileNet with a 92% nominal test accuracy shows a 77% OOD accuracy. We systematically investigate conditions leading to OOD performance drops and propose a preemptive assessment method to identify potential pitfalls when classifying new data, and pinpoint features in OOD images that adversely impact classification. We present a three‐step pipeline: (i) identifying OOD degradation compared to nominal test performance, (ii) conducting a diagnostic analysis of degradation causes, and (iii) providing solutions. We find that ensembles of BEiT vision transformers, with targeted augmentations addressing OOD robustness, geometric ensembling, and rotation‐based test‐time augmentation, constitute the most robust model, which we call BEsT. It achieves an 83% OOD accuracy, with errors concentrated on container classes. Moreover, it exhibits lower sensitivity to dataset shift, and reproduces well the plankton abundances. Our proposed pipeline is applicable to generic plankton classifiers, contingent on the availability of suitable test cells. By identifying critical shortcomings and offering practical procedures to fortify models against dataset shift, our study contributes to the development of more reliable plankton classification technologies.

An efficient plant disease prediction model based on machine learning and deep learning classifiers

An efficient plant disease prediction model based on machine learning and deep learning classifiers

Event Classification on Subsea Pipeline Inspection Data Using an Ensemble of Deep Learning Classifiers

Subsea pipelines are the backbone of the modern oil and gas industry, transporting a total of 28% of global oil production. Due to several factors, such as corrosion or deformations, the pipelines might degrade over time, which might lead to serious economic and environmental damages if not addressed promptly. Therefore, it is crucial to detect any serious damage to subsea pipelines before they cause dangerous catastrophes. Inspections of subsea pipelines are usually made using a Remote Operating Vehicle and the inspection data is usually processed manually, which is subject to human errors, and requires experienced Remote Operating Vehicle operators. It is thus necessary to automate the inspection process to enable more efficiency as well as reduce costs. Besides, it is recognised that specific challenges of noisy and low-quality inspection data arising from the underwater environment prevent the industry from taking full advantage of the recent development in the Artificial Intelligence field to the problem of subsea pipeline inspection. In this paper, we developed an ensemble of deep learning classifiers to further improve the performance of single deep learning models in classifying anomalous events on the subsea pipeline inspection data. The output of the proposed ensemble was combined based on a weighted combining method. The weights of base classifiers were found by minimising the difference between the weighted combining result and the given associated ground truth annotation information. Three inspection datasets, gathered from different oil and gas companies in the United Kingdom, were analysed. These datasets were recorded under varying conditions and include a range of anomalies. The results showed that the proposed ensemble achieves around 78% accuracy on two datasets and more than 99% accuracy on one dataset, which is better compared to base classifiers and two popular ensembles.

Survey on Focused Web Crawling Approaches for Ayurvedic Plant Information Retrieval

The Ayurvedic medical system is heavily reliant on medicinal plants, demanding correct information retrieval from the web. This study examines specific web crawling strategies for collecting useful information about Ayurvedic botanicals, with a focus on deep learning methodologies. Crawlers optimized with machine learning models retrieve domain-specific content while filtering out unnecessary data. The paper looks at approaches such as the TRES framework, which is a reinforcement learning-based crawler that discretizes vast state and action spaces in order to effectively choose ideal URLs. Furthermore, convolutional neural networks (CNN) and natural language processing (NLP) have been used in crawlers to improve categorization, as demonstrated by successful Turkish language processing applications. The paper "Learning to Crawl: Comparing Classification Schemes" conducts a comparative comparison of old rule-based approaches and newer deep learning classifiers, demonstrating the latter's superiority. In addition, a Naive Bayes classifier is employed in an Ayurvedic plant-focused crawler, which employs query expansion via a carefully curated thesaurus to improve relevancy in retrieved web pages. This poll emphasizes the need for more efficient, adaptive, and focused crawlers powered by deep learning to progress Ayurvedic research.

Improved facial emotion recognition model based on a novel deep convolutional structure

Facial Emotion Recognition (FER) is a very challenging task due to the varying nature of facial expressions, occlusions, illumination, pose variations, cultural and gender differences, and many other aspects that cause a drastic degradation in quality of facial images. In this paper, an anti-aliased deep convolution network (AA-DCN) model has been developed and proposed to explore how anti-aliasing can increase and improve recognition fidelity of facial emotions. The AA-DCN model detects eight distinct emotions from image data. Furthermore, their features have been extracted using the proposed model and numerous classical deep learning algorithms. The proposed AA-DCN model has been applied to three different datasets to evaluate its performance: The Cohn-Kanade Extending (CK+) database has been utilized, achieving an ultimate accuracy of 99.26% in (5 min, 25 s), the Japanese female facial expressions (JAFFE) obtained 98% accuracy in (8 min, 13 s), and on one of the most challenging FER datasets; the Real-world Affective Face (RAF) dataset; reached 82%, in low training time (12 min, 2s). The experimental results demonstrate that the anti-aliased DCN model is significantly increasing emotion recognition while improving the aliasing artifacts caused by the down-sampling layers.

Improved prediction of post-translational modification crosstalk within proteins using DeepPCT.

Post-translational modification (PTM) crosstalk events play critical roles in biological processes. Several machine learning methods have been developed to identify PTM crosstalk within proteins, but the accuracy is still far from satisfactory. Recent breakthroughs in deep learning and protein structure prediction could provide a potential solution to this issue. We proposed DeepPCT, a deep learning algorithm to identify PTM crosstalk using AlphaFold2-based structures. In this algorithm, one deep learning classifier was constructed for sequence-based prediction by combining the residue and residue pair embeddings with cross-attention techniques, while the other classifier was established for structure-based prediction by integrating the structural embedding and a graph neural network. Meanwhile, a machine learning classifier was developed using novel structural descriptors and a random forest model to complement the structural deep learning classifier. By integrating the three classifiers, DeepPCT outperformed existing algorithms in different evaluation scenarios and showed better generalizability on new data owing to its less distance dependency. Datasets, codes, and models of DeepPCT are freely accessible at https://github.com/hzau-liulab/DeepPCT/. Supplementary data are available at Bioinformatics online.

High-Throughput Multiplexed Plasmonic Color Encryption of Microgel Architectures via Programmable Dithering-Mask Flow Microlithography.

Silver nanoparticles (AgNPs) are known for their unique plasmonic colors and interaction with light, making them ideal for color printing and data encoding. Traditional methods like electron beam lithography (EBL) and focused ion beam (FIB) milling, however, suffer from low throughput and high costs. In this paper, a scalable and cost-efficient method is introduced for producing multiplexed plasmonic colors by in situ photoreducing AgNPs within microgel architectures with controlled porosity. Utilizing a digital micro-mirror device (DMD)-based flow microlithography system combined with a programmable dithering-mask technique, the high-throughput synthesis of shape or barcoded microparticles is facilitated, along with large-scale, high-resolution images embedded with hidden multiplexed plasmonic colors. This approach allows for a hidden multiplexed plasmonic color code library, significant enhancing the encoding capacity of barcode microparticles from 33 to 303 (a 1000-fold increase). Additionally, quantitative agreement is achieved between chemically encrypted and optically decrypted plasmonic colors using a deep learning classifier. Moreover, the method also supports the production of large scale (>5.6 × 5.6cm2), high-resolution (>300 dpi) microgel arrays encrypted with multiple plasmonic colors in under 30min. The multiplexed plasmonic coloration strategy in microgel architectures paves a new way for hidden data storage, secure optical labeling, and anti-counterfeiting technologies.