Ensuring reliability and interpretability in AI-driven medical imaging is vital for fostering trust in healthcare applications. Deep learning models, including Convolutional Neural Networks (CNNs), attention-based Transformers, and Mamba-based state-space models, have demonstrated significant advancements in classification performance. However, their black-box nature necessitates rigorous explainability analysis to ensure transparency and reliability. In this study, we systematically evaluate the classification performance of these architectures across three medical imaging datasets: BreakHis for histopathology (at multiple magnifications), Chest X-ray for pulmonary disease classification, and Retinal images for ocular disease diagnosis. We employ gradient-based and attention-based post-hoc explainability techniques, including Grad-CAM, Grad-CAM++, and attention rollout mechanisms, to interpret model decisions and visualize feature attributions. The results reveal that CNNs, particularly MobileNetv2 and ResNet50, excel in datasets requiring fine-grained feature extraction, whereas Transformers demonstrate superior performance in tasks emphasizing global dependencies, such as Chest X-ray analysis. Mamba-based models, such as EfficientViM, provide a balance between computational efficiency and interpretability, effectively capturing long-range dependencies in complex datasets like Retinal images. By analyzing decision-making patterns and feature attribution maps, we highlight the trade-offs between classification accuracy, computational efficiency, and model interpretability. Our findings offer actionable insights for selecting task-specific AI architectures, ensuring a balance between performance and transparency, and paving the way for the deployment of trustworthy AI in medical diagnostics.

Breast cancer is a devastating disease that affects women worldwide, and computer-aided algorithms have shown potential in automating cancer diagnosis. Recently Generative Artificial Intelligence (GenAI) opens new possibilities for addressing the challenges of labeled data scarcity and accurate prediction in critical applications. However, a lack of diversity, as well as unrealistic and unreliable data, have a detrimental impact on performance. Therefore, this study proposes an augmentation scheme to address the scarcity of labeled data and data imbalance in medical datasets. This approach integrates the concepts of the Gaussian-Laplacian pyramid and pyramid blending with similarity measures. In order to maintain the structural properties of images and capture inter-variability of patient images of the same category similarity-metric-based intermixing has been introduced. It helps to maintain the overall quality and integrity of the dataset. Subsequently, deep learning approach with significant modification, that leverages transfer learning through the usage of concatenated pre-trained models is applied to classify breast cancer histopathological images. The effectiveness of the proposal, including the impact of data augmentation, is demonstrated through a detailed analysis of three different medical datasets, showing significant performance improvement over baseline models. The proposal has the potential to contribute to the development of more accurate and reliable approach for breast cancer diagnosis.

Breast cancer is a significant global health concern, with Invasive Ductal Carcinoma (IDC) being a significant subtype. Detecting IDC is a challenging task that can be hindered by the oversight of important contextual cues within Whole Slide Images (WSIs). To address this issue, we present the SequentialLSTM Invasive Ductal Carcinoma Detection with EfficientNet (SLIDE-Net) framework. SLIDE-Net synchronizes tissue image patch locations within WSIs, allowing for comprehensive and subtle identification of IDC. The inherent issue of data imbalance in IDC datasets, particularly the variable density of positive IDC patches concerning WSI size, is effectively tackled through the introduction of Adaptive Fuzzy C-Mean as a Data Balancing Policy. This novel approach enhances model efficacy and robustness, resulting in superior performance across key metrics such as accuracy (86%), balanced accuracy (86%), sensitivity (87%), specificity (86%), and GMean (86%). Our findings were substantiated by a comprehensive analysis revealing the significant cross-patch dependencies observed among patches of similar classes. When tested on the balanced PatchCAM dataset, SLIDENet once again showcased superiority with accuracy (89%), balanced accuracy (89%), F1-Score (87%), sensitivity (87%), and specificity (90%). This underscores the efficient utilization of shared contextual information within WSIs. These results firmly establish SLIDE-Net as a robust and reliable solution for accurate IDC detection. Our work not only advances the precision of IDC detection but also contributes valuable insights into the intricate dynamics of histopathology image analysis, paving the way for enhanced diagnostic accuracy in the ongoing battle against breast cancer. Index Terms—Breast Cancer, Sequential-Modeling, Data Balancing Policy, WSI, Adaptive Fuzzy C-Mean, Deep Learning.

The industrial sector has experienced a tremendous advancement in deep supervised learning due to its representation ability, but it comes with high computing and labeled data demands. Recently, the demand for intelligent IoT devices on assembly and disassembly lines has surged. This necessitates algorithms that can use data to make intelligent decisions and a framework that can enable multiple IoT devices to learn collaboratively. Further, huge image generation using IoTs also needs an efficient data annotation scheme for classification problems. WeCollab is one such framework in federated learning that significantly reduces human efforts in data annotation with breakthroughs in self-supervised learning. The proposed framework is generic and can be adapted to any specific image data generated by industrial robots involved in assembly and disassembly lines. Our method outperforms supervised learning by 25% and 20% on the CIFAR-10 and CINIC-10 datasets, respectively, for the labeling task. We generate pseudo labels for the unlabeled part of the data and train a model to achieve 30% better test accuracy on CIFAR-10 and 20% better test accuracy on the CINIC-10 dataset as compared to supervised learning. Extensive experiments unveil the effectiveness of the method and proposed combination of loss functions used by WeCollab.

Low hemolytic therapeutic peptides have gained an edge over small molecule-based medicines. However, finding low hemolytic peptides in laboratory is time-consuming, costly and necessitates the use of mammalian red blood cells. Therefore, wet-lab researchers often performin-silicoprediction to select low hemolytic peptides before proceeding with in-vitro testing. Thein-silicotools available for this purpose have following limitations: (i) They do not provide predictions for peptides having N/C terminal modifications. (ii) Data is food for AI; however, datasets used to create existing tools do not contain peptide data generated over past eight years. (iii) Performance of available tools is also low. Therefore, a novel framework has been proposed in current work, which utilizes recent dataset and uses ensemble learning technique to combine the decisions produced by bidirectional long short-term memory, bidirectional temporal convolutional network, and 1-dimensional convolutional neural network deep learning algorithms. Deep learning algorithms are capable of extracting features themselves from data. However, instead of relying solely on deep learning-based features (DLF), handcrafted features (HCF) were also provided so that deep learning algorithms can learn features that are missing from HCF, and a better feature vector can be constructed by concatenating HCF and DLF. Additionally, ablation studies were carried out to understand the roles of an ensemble algorithm, HCF, and DLF in the proposed framework. Ablation studies found that the ensemble algorithm, HCF and DLF are crucial components of proposed framework, and there is a decrease in performance on eliminating any of them. Mean value of performance metrics, namely$A_{cc}$,$S_{n}$,$P_{r}$,$F_{s}$,$S_{p}$,$B_{a}$, and$M{cc}$obtained by proposed framework for test data is$\approx$87, 85, 86, 86, 88, 87, and 73, respectively. To aid scientific community, model developed from proposed framework has been deployed as a web server athttps://endl-hemolyt.anvil.app/.

In response to environmental threats, pathogens make several changes in their genome, leading to antimicrobial resistance (AMR). Due to AMR, the pathogens do not respond to antibiotics. Amongst drug-resistant pathogens, the ESKAPEE group of bacteria poses a major threat to humans, and therefore World Health Organization has given them the highest priority status. Antibacterial peptides (ABPs) are a family of peptides found in nature that play a crucial role in the innate immune systems of organisms. These ABPs offer several advantages over widely used antibiotics. As a result, they have recently received a lot of attention as potential replacements for currently available antibiotics. But it is expensive and time-consuming to identify ABPs from natural sources. Thus, wet lab researchers employ various tools to screen promising ABPs rapidly. However, the main limitation of the existing tools is that they do not provide the minimum inhibitory concentration values against the ESKAPEE pathogens for the identified ABP. To address this, in the current work, we developed ESKAPEE-MICpred, a two-input model that utilizes transfer learning and ensemble learning techniques. The concept of ensemble learning was realized by combining the decisions provided by deep learning algorithms, whereas the concept of transfer learning was realized by utilizing pretrained amino acid embeddings. The proposed model has been deployed as a web server at https://eskapee-micpred.anvil.app/ to aid the scientific community.

Fungal infections or mycosis cause a wide range of diseases in humans and animals. The incidences of community acquired; nosocomial fungal infections have increased dramatically after the emergence of COVID-19 pandemic. The increase in number of patients with immunodeficiency / immunosuppression related diseases, resistance to existing antifungal compounds and availability of limited therapeutic options has triggered the search for alternative antifungal molecules. In this direction, antifungal peptides (AFPs) have received a lot of interest as an alternative to currently available antifungal drugs. Although the AFPs are produced by diverse population of living organisms, identifying effective AFPs from natural sources is time-consuming and expensive. Therefore, there is a need to develop a robust in silico model capable of identifying novel AFPs in protein sequences. In this paper, we propose Deep-AFPpred, a deep learning classifier that can identify AFPs in protein sequences. We developed Deep-AFPpred using the concept of transfer learning with 1DCNN-BiLSTM deep learning algorithm. The findings reveal that Deep-AFPpred beats other state-of-the-art AFP classifiers by a wide margin and achieved approximately 96% and 94% precision on validation and test data, respectively. Based on the proposed approach, an online prediction server is created and made publicly available at https://afppred.anvil.app/. Using this server, one can identify novel AFPs in protein sequences and the results are provided as a report that includes predicted peptides, their physicochemical properties and motifs. By utilizing this model, we identified AFPs in different proteins, which can be chemically synthesized in lab and experimentally validated for their antifungal activity.

Due to the rapid emergence of multi-drug resistant (MDR) bacteria, existing antibiotics are becoming ineffective. So, researchers are looking for alternatives in the form of antibacterial peptides (ABPs) based medicines. The discovery of novel ABPs using wet-lab experiments is time-consuming and expensive. Many machine learning models have been proposed to search for new ABPs, but there is still scope to develop a robust model that has high accuracy and precision. In this work, we present StaBle-ABPpred, a stacked ensemble technique-based deep learning classifier that uses bidirectional long-short term memory (biLSTM) and attention mechanism at base-level and an ensemble of random forest, gradient boosting and logistic regression at meta-level to classify peptides as antibacterial or otherwise. The performance of our model has been compared with several state-of-the-art classifiers, and results were subjected to analysis of variance (ANOVA) test and its post hoc analysis, which proves that our model performs better than existing classifiers. Furthermore, a web app has been developed and deployed at https://stable-abppred.anvil.app to identify novel ABPs in protein sequences. Using this app, we identified novel ABPs in all the proteins of the Streptococcus phage T12 genome. These ABPs have shown amino acid similarities with experimentally tested antimicrobial peptides (AMPs) of other organisms. Hence, they could be chemically synthesized and experimentally validated for their activity against different bacteria. The model and app developed in this work can be further utilized to explore the protein diversity for identifying novel ABPs with broad-spectrum activity, especially against MDR bacterial pathogens.

The application of machine intelligence in biological sciences has led to the development of several automated tools, thus enabling rapid drug discovery. Adding to this development is the ongoing COVID-19 pandemic, due to which researchers working in the field of artificial intelligence have acquired an active interest in finding machine learning-guided solutions for diseases like mucormycosis, which has emerged as an important post-COVID-19 fungal complication, especially in immunocompromised patients. On these lines, we have proposed a temporal convolutional network-based binary classification approach to discover new antifungal molecules in the proteome of plants and animals to accelerate the development of antifungal medications. Although these biomolecules, known as antifungal peptides (AFPs), are part of an organism's intrinsic host defense mechanism, their identification and discovery by traditional biochemical procedures is arduous. Also, the absence of a large dataset on AFPs is also a considerable impediment in building a robust automated classifier. To this end, we have employed the transfer learning technique to pre-train our model on antibacterial peptides. Subsequently, we have built a classifier that predicts AFPs with accuracy and precision of 94%. Our classifier outperforms several state-of-the-art models by a considerable margin. The results of its performance were proven as statistically significant using the Kruskal-Wallis H test, followed by a post hoc analysis performed using the Tukey honestly significant difference (HSD) test. Furthermore, we identified potent AFPs in representative animal (Histatin) and plant (Snakin) proteins using our model. We also built and deployed a web app that is freely available at https://tcn-afppred.anvil.app/ for the identification of AFPs in protein sequences.

Deep learning demonstrates its efficacy and potential to solve challenging computer vision problems in medical and other industrial applications. Federated learning is a learning paradigm that facilitates collaborative learning in a federation of users without exchanging actual data with a single authority like a server. However, federated learning provides only a basic level of privacy and robustness and is vulnerable to model poisoning and model inversion attacks in hostile training environments. Hence, in this article, we propose MediSecFed—a secure framework for federated learning in a hostile environment. Compared to the widely used FedAvg, our method relies on simple and practical ideas from knowledge distillation and model inversion to ensure additional security and privacy features. Our approach achieves knowledge exchange among participating entities without sharing model parameters as FedAvg does, thus protecting the privacy of the local data from the server and significantly reducing communication costs. We evaluate our method on two chest X-ray datasets. Our method outperforms FedAvg by 15% on both datasets in a hostile environment. Our method will also continue to maintain good performance even if the number of malicious participating entities increases. Robustness to learn in a malicious environment while preserving privacy with reduced communication costs makes our method more desirable and efficient than that of FedAvg.

Rapid increase in viral outbreaks has resulted in the spread of viral diseases in diverse species and across geographical boundaries. The zoonotic viral diseases have greatly affected the well-being of humans, and the COVID-19 pandemic is a burning example. The existing antivirals have low efficacy, severe side effects, high toxicity, and limited market availability. As a result, natural substances have been tested for antiviral activity. The host defense molecules like antiviral peptides (AVPs) are present in plants and animals and protect them from invading viruses. However, obtaining AVPs from natural sources for preparing synthetic peptide drugs is expensive and time-consuming. As a result, an in-silico model is required for identifying new AVPs. We proposed Deep-AVPpred, a deep learning classifier for discovering AVPs in protein sequences, which utilises the concept of transfer learning with a deep learning algorithm. The proposed classifier outperformed state-of-the-art classifiers and achieved approximately 94% and 93% precision on validation and test sets, respectively. The high precision indicates that Deep-AVPpred can be used to propose new AVPs for synthesis and experimentation. By utilising Deep-AVPpred, we identified novel AVPs in human interferons- α family proteins. These AVPs can be chemically synthesised and experimentally verified for their antiviral activity against different viruses. The Deep-AVPpred is deployed as a web server and is made freely available at https://deep-avppred.anvil.app, which can be utilised to predict novel AVPs for developing antiviral compounds for use in human and veterinary medicine.

With advancements in genomics, there has been substantial reduction in the cost and time of genome sequencing and has resulted in lot of data in genome databases. Antimicrobial host defense proteins provide protection against invading microbes. But confirming the antimicrobial function of host proteins by wet-lab experiments is expensive and time consuming. Therefore, there is a need to develop an in silico tool to identify the antimicrobial function of proteins. In the current study, we developed a model AniAMPpred by considering all the available antimicrobial peptides (AMPs) of length $\in $[10 200] from the animal kingdom. The model utilizes a support vector machine algorithm with deep learning-based features and identifies probable antimicrobial proteins (PAPs) in the genome of animals. The results show that our proposed model outperforms other state-of-the-art classifiers, has very high confidence in its predictions, is not biased and can classify both AMPs and non-AMPs for a diverse peptide length with high accuracy. By utilizing AniAMPpred, we identified 436 PAPs in the genome of Helobdella robusta. To further confirm the functional activity of PAPs, we performed BLAST analysis against known AMPs. On detailed analysis of five selected PAPs, we could observe their similarity with antimicrobial proteins of several animal species. Thus, our proposed model can help the researchers identify PAPs in the genome of animals and provide insight into the functional identity of different proteins. An online prediction server is also developed based on the proposed approach, which is freely accessible at https://aniamppred.anvil.app/.

The overuse of antibiotics has led to emergence of antimicrobial resistance, and as a result, antibacterial peptides (ABPs) are receiving significant attention as an alternative. Identification of effective ABPs in lab from natural sources is a cost-intensive and time-consuming process. Therefore, there is a need for the development of in silico models, which can identify novel ABPs in protein sequences for chemical synthesis and testing. In this study, we propose a deep learning classifier named Deep-ABPpred that can identify ABPs in protein sequences. We developed Deep-ABPpred using bidirectional long short-term memory algorithm with amino acid level features from word2vec. The results show that Deep-ABPpred outperforms other state-of-the-art ABP classifiers on both test and independent datasets. Our proposed model achieved the precision of approximately 97 and 94% on test dataset and independent dataset, respectively. The high precision suggests applicability of Deep-ABPpred in proposing novel ABPs for synthesis and experimentation. By utilizing Deep-ABPpred, we identified ABPs in the tail protein sequences of Streptococcus bacteriophages, chemically synthesized identified peptides in lab and tested their activity in vitro. These ABPs showed potent antibacterial activity against selected Gram-positive and Gram-negative bacteria, which confirms the capability of Deep-ABPpred in identifying novel ABPs in protein sequences. Based on the proposed approach, an online prediction server is also developed, which is freely accessible at https://abppred.anvil.app/. This web server takes the protein sequence as input and provides ABPs with high probability (>0.95) as output.

Recent advances in artificial intelligence (AI), especially convolutional neural networks (CNNs), alongside the digitization of histopathological images, have made the computer-aided diagnosis of breast cancer a reality. However, deep learning-based approaches are computationally expensive and have huge parameters, which makes them less affordable for edge devices. In order to make them affordable for edge devices, the whole classification model needs to be compressed while maintaining accuracy. Providing a low-cost solution for histopathological diagnosis in the recent edge-computing world is of utmost importance. Therefore, in this study, we propose “MobiHisNet,” an efficient and lightweight CNN model for histopathological image classification (HIC) based on MobileNet. MobiHisNet was successfully deployed on a Raspberry Pi, as well as three mobile devices, demonstrating its ability to run on a lightweight and portable processor. Our studies indicated that a depth parameter ($\gamma = 0.5$) and a 16-bit quantization are the optimum parameters for the proposed model while balancing the accuracy, inference time, and memory peak requirements. Compared to the state-of-the-art, pretrained models, MobiHisNet has fewer parameters and calculations, resulting in faster image classification. This renders it more viable for production purposes and applications on edge devices. In addition, MobiHisNet is computationally faster than VGG16, ResNet50, Xception, and InceptionV3 by twenty-seven, eight, six, and five times, respectively. This also outperforms all the baseline models with the moderate model size and FLOP counts. Experiments on breast cancer HIC (BreakHis) data sets show superior performance of MobiHisNet on edge devices in terms of higher accuracy, lesser complexity, and lesser memory requirements. Thus, it has a high potential for deployment in mobile edge devices.

Accurate breast cancer detection using automated algorithms remains a problem within the literature. Although a plethora of work has tried to address this issue, an exact solution is yet to be found. This problem is further exacerbated by the fact that most of the existing datasets are imbalanced, i.e., the number of instances of a particular class far exceeds that of the others. In this paper, we propose a framework based on the notion of transfer learning to address this issue and focus our efforts on histopathological and imbalanced image classification. We use the popular VGG-19 as the base model and complement it with several state-of-the-art techniques to improve the overall performance of the system. With the ImageNet dataset taken as the source domain, we apply the learned knowledge in the target domain consisting of histopathological images. With experimentation performed on a large-scale dataset consisting of 277,524 images, we show that the framework proposed in this paper gives superior performance than those available in the existing literature. Through numerical simulations conducted on a supercomputer, we also present guidelines for work in transfer learning and imbalanced image classification.

Machine learning (ML) has emerged as a powerful tool for pattern recognition. Traditional ML algorithms have limited ability to reveal the most sophisticated features of cancer histopathological images, but their robustness and fault tolerance can be enhanced by using fuzzy modeling to capture the uncertainty in image data. Therefore, this article proposes a novel CoMHisP framework based on a fuzzy support vector machine with within-class density information (FSVM-WD). It utilizes a novel feature extraction technique by optimizing the block size to extract image micropatterns and computing center of mass (CoM) for each pixel to extract feature vectors. The performance of the proposed framework is evaluated using a CMTHis dataset comprising histopathological images of canine mammary tumor (CMT), a prevalent neoplastic disease in female dogs, and an established model for human breast cancer. Data analysis reveals that stain normalization and magnification influence the performance of the CoMHisP framework, with the best results achieved at lower magnifications after stain normalization. The proposed framework achieves a classification accuracy of 97.25% ($\pm$1.80%) using a FSVM-WD classifier, outperforming both traditional ML and deep FE-VGGNET16-based feature descriptors. To the best of our knowledge, this is the first time a CoM-based feature descriptor has been proposed for histopathological image analysis of CMTs and its performance was evaluated using a fuzzy SVM-based classifier. The proposed method performs well with datasets of limited size and low-magnification images and, therefore, has the potential to provide rapid and accurate diagnosis in low-cost clinical settings.

The advancements in the Internet of Things (IoT) and cloud services have enabled the availability of smart e-healthcare services in a distant and distributed environment. However, this has also raised major privacy and efficiency concerns that need to be addressed. While sharing clinical data across the cloud that often consists of sensitive patient-related information, privacy is a major challenge. Adequate protection of patients’ privacy helps to increase public trust in medical research. Additionally, DL-based models are complex, and in a cloud-based approach, efficient data processing in such models is complicated. To address these challenges, we propose an efficient and secure cancer diagnostic framework for histopathological image classification by utilizing both differential privacy and secure multi-party computation. For efficient computation, instead of performing the whole operation on the cloud, we decouple the layers into two modules: one for feature extraction using the VGGNet module at the user side and the remaining layers for private prediction over the cloud. The efficacy of the framework is validated on two datasets composed of histopathological images of the canine mammary tumor and human breast cancer. The application of differential privacy preserving to the proposed model makes the model secure and capable of preserving the privacy of sensitive data from any adversary, without significantly compromising the model accuracy. Extensive experiments show that the proposed model efficiently achieves the trade-off between privacy and model performance.