Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning

Application of artificial intelligence-based detection of furcation involvement in mandibular first molar using cone beam tomography images- a preliminary study

BACKGROUND

Radiographs play a key role in diagnosis of periodontal diseases. Deep learning models have been explored for image analysis in periodontal diseases. However, there is lacuna of research in the deep learning model-based detection of furcation involvements [FI]. The objective of this study was to determine the accuracy of deep learning model in the detection of FI in axial CBCT images.

METHODOLOGY

We obtained initial dataset 285 axial CBCT images among which 143 were normal (without FI) and 142 were abnormal (with FI). Data augmentation technique was used to create 600(300 normal and 300 abnormal) images by using 200 images from the training dataset. Remaining 85(43 normal and 42 abnormal) images were kept for testing of model. ResNet101V2 with transfer learning was used employed for the analysis of images.

RESULTS

Training accuracy of model is 98%, valid accuracy is 97% and test accuracy is 91%. The precision and F1 score were 0.98 and 0.98 respectively. The Area under curve (AUC) was reported at 0.98. The test loss was reported at 0.2170.

CONCLUSION

The deep learning model (ResNet101V2) can accurately detect the FI in axial CBCT images. However, since our study was preliminary in nature and carried out with relatively smaller dataset, a study with larger dataset will further confirm the accuracy of deep learning models.

The application of the combination between artificial intelligence and endoscopy in gastrointestinal tumors

Gastrointestinal (GI) tumors have always been a major type of malignant tumor and a leading cause of tumor‐related deaths worldwide. The main principles of modern medicine for GI tumors are early prevention, early diagnosis, and early treatment, with early diagnosis being the most effective measure. Endoscopy, due to its ability to visualize lesions, has been one of the primary modalities for screening, diagnosing, and treating GI tumors. However, a qualified endoscopist often requires long training and extensive experience, which to some extent limits the wider use of endoscopy. With advances in data science, artificial intelligence (AI) has brought a new development direction for the endoscopy of GI tumors. AI can quickly process large quantities of data and images and improve diagnostic accuracy with some training, greatly reducing the workload of endoscopists and assisting them in early diagnosis. Therefore, this review focuses on the combined application of endoscopy and AI in GI tumors in recent years, describing the latest research progress on the main types of tumors and their performance in clinical trials, the application of multimodal AI in endoscopy, the development of endoscopy, and the potential applications of AI within it, with the aim of providing a reference for subsequent research.

Foundation models in gastrointestinal endoscopic AI: Impact of architecture, pre-training approach and data efficiency

Pre-training deep learning models with large data sets of natural images, such as ImageNet, has become the standard for endoscopic image analysis. This approach is generally superior to training from scratch, due to the scarcity of high-quality medical imagery and labels. However, it is still unknown whether the learned features on natural imagery provide an optimal starting point for the downstream medical endoscopic imaging tasks. Intuitively, pre-training with imagery closer to the target domain could lead to better-suited feature representations. This study evaluates whether leveraging in-domain pre-training in gastrointestinal endoscopic image analysis has potential benefits compared to pre-training on natural images. To this end, we present a dataset comprising of 5,014,174 gastrointestinal endoscopic images from eight different medical centers (GastroNet-5M), and exploit self-supervised learning with SimCLRv2, MoCov2 and DINO to learn relevant features for in-domain downstream tasks. The learned features are compared to features learned on natural images derived with multiple methods, and variable amounts of data and/or labels (e.g. Billion-scale semi-weakly supervised learning and supervised learning on ImageNet-21k). The effects of the evaluation is performed on five downstream data sets, particularly designed for a variety of gastrointestinal tasks, for example, GIANA for angiodyplsia detection and Kvasir-SEG for polyp segmentation. The findings indicate that self-supervised domain-specific pre-training, specifically using the DINO framework, results into better performing models compared to any supervised pre-training on natural images. On the ResNet50 and Vision-Transformer-small architectures, utilizing self-supervised in-domain pre-training with DINO leads to an average performance boost of 1.63% and 4.62%, respectively, on the downstream datasets. This improvement is measured against the best performance achieved through pre-training on natural images within any of the evaluated frameworks. Moreover, the in-domain pre-trained models also exhibit increased robustness against distortion perturbations (noise, contrast, blur, etc.), where the in-domain pre-trained ResNet50 and Vision-Transformer-small with DINO achieved on average 1.28% and 3.55% higher on the performance metrics, compared to the best performance found for pre-trained models on natural images. Overall, this study highlights the importance of in-domain pre-training for improving the generic nature, scalability and performance of deep learning for medical image analysis. The GastroNet-5M pre-trained weights are made publicly available in our repository: huggingface.co/tgwboers/GastroNet-5M_Pretrained_Weights.

Ethical, legal, and regulatory landscape of artificial intelligence in Australian healthcare and ethical integration in radiography: A narrative review

This narrative review explores the ethical, legal, and regulatory landscape of AI integration in Australian healthcare, focusing on radiography. It examines the current legislative framework, assesses the trust and reliability of AI tools, and proposes future directions for ethical AI integration in radiography. AI systems significantly enhance diagnostic radiography by improving diagnostic accuracy and efficiency in stroke detection, brain imaging, and chest reporting. However, AI raises substantial ethical concerns due to its 'black-box' nature and potential biases in training data. The Therapeutic Goods Administration's reforms in Australia, though comprehensive, fall short of fully addressing issues related to the trustworthiness and legal liabilities of AI tools. Adopting a comprehensive research strategy that includes doctrinal, comparative, and public policy analyses will facilitate an understanding of international practices, particularly from countries with similar legal systems, and help guide Australia in refining its regulatory framework. For an ethical future in radiography, a robust, multi-disciplinary approach is required to prioritize patient safety, data privacy, and equitable AI use. A framework that balances technological innovation with ethical and legal integrity is essential for advancing healthcare while preserving trust and transparency. Healthcare professionals, policymakers, and AI developers must collaborate to establish a resilient, equitable, and transparent healthcare system. Future research should focus on multi-disciplinary methodologies, combining doctrinal, comparative, and public policy research to provide comprehensive insights. This approach will guide Australia in creating a more inclusive and ethically sound legal framework for AI in healthcare, ensuring its ethical and beneficial integration into radiography.

Exploring the efficacy of various CNN architectures in diagnosing oral cancer from squamous cell carcinoma

Oral cancer can result from mutations in cells located in the lips or mouth. Diagnosing oral cavity squamous cell carcinoma (OCSCC) is particularly challenging, often occurring at advanced stages. To address this, computer-aided diagnosis methods are increasingly being used. In this work, a deep learning-based approach utilizing models such as VGG16, ResNet50, LeNet-5, MobileNetV2, and Inception V3 is presented. NEOR and OCSCC datasets were used for feature extraction, with virtual slide images divided into tiles and classified as normal or squamous cell cancer. Performance metrics like accuracy, F1-score, AUC, precision, and recall were analyzed to determine the prerequisites for optimal CNN performance. The proposed CNN approaches were effective for classifying OCSCC and oral dysplasia, with the highest accuracy of 95.41 % achieved using MobileNetV2.

Key findings

Deep learning models, particularly MobileNetV2, achieved high classification accuracy (95.41 %) for OCSCC.CNN-based methods show promise for early-stage OCSCC and oral dysplasia diagnosis. Performance parameters like precision, recall, and F1-score help optimize CNN model selection for this task.

Principle of TEM alignment using convolutional neural networks: Case study on condenser aperture alignment

The possibility of automatically aligning the transmission electron microscope (TEM) is explored using an approach based on artificial intelligence (AI). After presenting the general concept, we test the method on the first step of the alignment process which involves centering the condenser aperture. We propose using a convolutional neural network (CNN) that learns to predict the x and y-shifts needed to realign the aperture in one step. The learning data sets were acquired automatically on the microscope by using a simplified digital twin. Different models were tested and analysed to choose the optimal design. We have developed a human-level estimator and intend to use it safely on all apertures. A similar process could be used for most steps of the alignment process with minimal changes, allowing microscopists to reduce the time and training required to perform this task. The method is also compatible with continuous correction of alignment drift during lengthy experiments or to ensure uniformity of illumination conditions during data acquisition.

Medical Transformer: Universal Encoder for 3-D Brain MRI Analysis

Transfer learning has attracted considerable attention in medical image analysis because of the limited number of annotated 3-D medical datasets available for training data-driven deep learning models in the real world. We propose Medical Transformer, a novel transfer learning framework that effectively models 3-D volumetric images as a sequence of 2-D image slices. To improve the high-level representation in 3-D-form empowering spatial relations, we use a multiview approach that leverages information from three planes of the 3-D volume, while providing parameter-efficient training. For building a source model generally applicable to various tasks, we pretrain the model using self-supervised learning (SSL) for masked encoding vector prediction as a proxy task, using a large-scale normal, healthy brain magnetic resonance imaging (MRI) dataset. Our pretrained model is evaluated on three downstream tasks: 1) brain disease diagnosis; 2) brain age prediction; and 3) brain tumor segmentation, which are widely studied in brain MRI research. Experimental results demonstrate that our Medical Transformer outperforms the state-of-the-art (SOTA) transfer learning methods, efficiently reducing the number of parameters by up to approximately 92% for classification and regression tasks and 97% for segmentation task, and it also achieves good performance in scenarios where only partial training samples are used.

Computational pathology: A survey review and the way forward

Computational Pathology (CPath) is an interdisciplinary science that augments developments of computational approaches to analyze and model medical histopathology images. The main objective for CPath is to develop infrastructure and workflows of digital diagnostics as an assistive CAD system for clinical pathology, facilitating transformational changes in the diagnosis and treatment of cancer that are mainly address by CPath tools. With evergrowing developments in deep learning and computer vision algorithms, and the ease of the data flow from digital pathology, currently CPath is witnessing a paradigm shift. Despite the sheer volume of engineering and scientific works being introduced for cancer image analysis, there is still a considerable gap of adopting and integrating these algorithms in clinical practice. This raises a significant question regarding the direction and trends that are undertaken in CPath. In this article we provide a comprehensive review of more than 800 papers to address the challenges faced in problem design all-the-way to the application and implementation viewpoints. We have catalogued each paper into a model-card by examining the key works and challenges faced to layout the current landscape in CPath. We hope this helps the community to locate relevant works and facilitate understanding of the field's future directions. In a nutshell, we oversee the CPath developments in cycle of stages which are required to be cohesively linked together to address the challenges associated with such multidisciplinary science. We overview this cycle from different perspectives of data-centric, model-centric, and application-centric problems. We finally sketch remaining challenges and provide directions for future technical developments and clinical integration of CPath. For updated information on this survey review paper and accessing to the original model cards repository, please refer to GitHub. Updated version of this draft can also be found from arXiv.

Liquid Biopsy-Based Detection and Response Prediction for Depression

Proactively predicting antidepressant treatment response before medication failures is crucial, as it reduces unsuccessful attempts and facilitates the development of personalized therapeutic strategies, ultimately enhancing treatment efficacy. The current decision-making process, which heavily depends on subjective indicators, underscores the need for an objective, indicator-based approach. This study developed a method for detecting depression and predicting treatment response through deep learning-based spectroscopic analysis of extracellular vesicles (EVs) from plasma. EVs were isolated from the plasma of both nondepressed and depressed groups, followed by Raman signal acquisition, which was used for AI algorithm development. The algorithm successfully distinguished depression patients from healthy individuals and those with panic disorder, achieving an AUC accuracy of 0.95. This demonstrates the model's capability to selectively diagnose depression within a nondepressed group, including those with other mental health disorders. Furthermore, the algorithm identified depression-diagnosed patients likely to respond to antidepressants, classifying responders and nonresponders with an AUC accuracy of 0.91. To establish a diagnostic foundation, the algorithm applied explainable AI (XAI), enabling personalized medicine for companion diagnostics and highlighting its potential for the development of liquid biopsy-based mental disorder diagnosis.

Automatic vessel segmentation and reformation of non-contrast coronary magnetic resonance angiography using transfer learning-based three-dimensional U-net with attention mechanism

BACKGROUND

Coronary magnetic resonance angiography (CMRA) presents distinct advantages, but its reliance on manual image post-processing is labor-intensive and requires specialized knowledge. This study aims to design and test an efficient artificial intelligence (AI) model capable of automating coronary artery segmentation and reformation from CMRA images for coronary artery disease (CAD) diagnosis.

METHODS

By leveraging transfer learning from a pre-existing coronary computed tomography angiography model, a three-dimensional attention-aware U-Net was established, trained, and validated on a dataset of 104 subjects' CMRA. Furthermore, an independent clinical evaluation was conducted on an additional cohort of 70 patients. The AI model's performance in segmenting coronary arteries was assessed using the Dice similarity coefficient (DSC) and recall. The comparison between the AI model and manual processing by experienced radiologists on vessel reformation was based on reformatted image quality (rIQ) scoring, post-processing time, and the number of necessary user interactions. The diagnostic performance of AI-segmented CMRA for significant stenosis (≥50% diameter reduction) was evaluated using conventional coronary angiography (CAG) as a reference in sub-set data.

RESULTS

The DSC of the AI model achieved on the training and validation sets were 0.952 and 0.944, with recalls of 0.936 and 0.923, respectively. In the clinical evaluation, the model outperformed manual processes by reducing vessel post-processing time, from 632.6±17.0 s to 77.4±8.9 s, and the number of user interactions from 221±59 to 8±2. The AI post-processed images maintained high rIQ scores comparable to those processed manually (2.7±0.8 vs 2.7±0.6; P = 0.4806). In subjects with CAG, the prevalence of CAD was 71%. The sensitivity, specificity, and accuracy at patient-based analysis were 94%, 71%, and 88%, respectively, by AI post-processed whole-heart CMRA.

CONCLUSION

The AI auto-segmentation system can effectively facilitate CMRA vessel reformation and reduce the time consumption for radiologists. It has the potential to become a standard component of daily workflows, optimizing the clinical application of CMRA in the future.

Evaluating Brain Tumor Detection with Deep Learning Convolutional Neural Networks Across Multiple MRI Modalities

Central Nervous System (CNS) tumors represent a significant public health concern due to their high morbidity and mortality rates. Magnetic Resonance Imaging (MRI) has emerged as a critical non-invasive modality for the detection, diagnosis, and management of brain tumors, offering high-resolution visualization of anatomical structures. Recent advancements in deep learning, particularly convolutional neural networks (CNNs), have shown potential in augmenting MRI-based diagnostic accuracy for brain tumor detection. In this study, we evaluate the diagnostic performance of six fundamental MRI sequences in detecting tumor-involved brain slices using four distinct CNN architectures enhanced with transfer learning techniques. Our dataset comprises 1646 MRI slices from the examinations of 62 patients, encompassing both tumor-bearing and normal findings. With our approach, we achieved a classification accuracy of 98.6%, underscoring the high potential of CNN-based models in this context. Additionally, we assessed the performance of each MRI sequence across the different CNN models, identifying optimal combinations of MRI modalities and neural networks to meet radiologists' screening requirements effectively. This study offers critical insights into the integration of deep learning with MRI for brain tumor detection, with implications for improving diagnostic workflows in clinical settings.

Evaluation of Brain Age as a Specific Marker of Brain Health

Brain age is a powerful marker of general brain health. Furthermore, brain age models are trained on large datasets, thus giving them a potential advantage in predicting specific outcomes - much like the success of finetuning large language models for specific applications. However, it is also well-accepted in machine learning that models trained to directly predict specific outcomes (i.e., direct models) often perform better than those trained on surrogate outcomes. Therefore, despite their much larger training data, it is unclear whether brain age models outperform direct models in predicting specific brain health outcomes. Here, we compare large-scale brain age models and direct models for predicting specific health outcomes in the context of Alzheimer's Disease (AD) dementia. Using anatomical T1 scans from three continents (N = 1,848), we find that direct models outperform brain age models without finetuning. Finetuned brain age models yielded similar performance as direct models, but importantly, did not outperform direct models although the brain age models were pretrained on 1000 times more data than the direct models: N = 53,542 vs N = 50. Overall, our results do not discount brain age as a useful marker of general brain health. However, in this era of large-scale brain age models, our results suggest that small-scale, targeted approaches for extracting specific brain health markers still hold significant value.

Image biomarkers and explainable AI: handcrafted features versus deep learned features

Feature extraction and selection from medical data are the basis of radiomics and image biomarker discovery for various architectures, including convolutional neural networks (CNNs). We herein describe the typical radiomics steps and the components of a CNN for both deep feature extraction and end-to-end approaches. We discuss the curse of dimensionality, along with dimensionality reduction techniques. Despite the outstanding performance of deep learning (DL) approaches, the use of handcrafted features instead of deep learned features needs to be considered for each specific study. Dataset size is a key factor: large-scale datasets with low sample diversity could lead to overfitting; limited sample sizes can provide unstable models. The dataset must be representative of all the "facets" of the clinical phenomenon/disease investigated. The access to high-performance computational resources from graphics processing units is another key factor, especially for the training phase of deep architectures. The advantages of multi-institutional federated/collaborative learning are described. When large language models are used, high stability is needed to avoid catastrophic forgetting in complex domain-specific tasks. We highlight that non-DL approaches provide model explainability superior to that provided by DL approaches. To implement explainability, the need for explainable AI arises, also through post hoc mechanisms. RELEVANCE STATEMENT: This work aims to provide the key concepts for processing the imaging features to extract reliable and robust image biomarkers. KEY POINTS: The key concepts for processing the imaging features to extract reliable and robust image biomarkers are provided. The main differences between radiomics and representation learning approaches are highlighted. The advantages and disadvantages of handcrafted versus learned features are given without losing sight of the clinical purpose of artificial intelligence models.

ResNet-Transformer deep learning model-aided detection of dens evaginatus

BACKGROUND

Dens evaginatus is a dental morphological developmental anomaly. Failing to detect it may lead to tubercles fracture and pulpal/periapical disease. Consequently, early detection and intervention of dens evaginatus are significant to preserve vital pulp.

AIM

This study aimed to develop a deep learning model to assist dentists in early diagnosing dens evaginatus, thereby supporting early intervention and mitigating the risk of severe consequences.

DESIGN

In this study, a deep learning model was developed utilizing panoramic radiograph images sourced from 1410 patients aged 3-16 years, with high-quality annotations to enable the automatic detection of dens evaginatus. Model performance and model's efficacy in aiding dentists were evaluated.

RESULTS

The findings indicated that the current deep learning model demonstrated commendable sensitivity (0.8600) and specificity (0.9200), outperforming dentists in detecting dens evaginatus with an F1-score of 0.8866 compared to their average F1-score of 0.8780, indicating that the model could detect dens evaginatus with greater precision. Furthermore, with its support, young dentists heightened their focus on dens evaginatus in tooth germs and achieved improved diagnostic accuracy.

CONCLUSION

Based on these results, the integration of deep learning for dens evaginatus detection holds significance and can augment dentists' proficiency in identifying such anomaly.

Dung beetle optimization algorithm-based hybrid deep learning model for ultra-short-term PV power prediction

A hybrid model combining self-attention temporal convolutional networks (SATCN) with bidirectional long short-term memory (BiLSTM) networks was developed to improve the accuracy of ultra-short-term photovoltaic (PV) power prediction. The self-attention mechanism and SATCN were used to extract temporal and correlation features, which were then linked to BiLSTM networks. The model's hyperparameters were optimized using the dung beetle optimization algorithm. The model was tested on a year-long dataset of PV power and outperformed convolutional neural networks, BiLSTM networks, temporal convolutional networks, and other hybrid models. It reduced the root-mean-square error (RMSE) by 33.1% compared to the other models. The model achieved a mean absolute error (MAE) of 0.175, a weighted mean absolute percentage error (wMAPE) of 4.821, and a coefficient of determination (R2) of 0.997. These results highlight the model's superior accuracy and its potential applications in solar energy development.

The detection of distomolar teeth on panoramic radiographs using different artificial intelligence models

PURPOSES

One notable anomaly, presence of distomolars, arises beyond the typical sequence of the human dental system. In this study, convolutional neural networks (CNNs) based machine learning methods were employed to classify distomolar tooth existence using panoramic radiography (PR).

METHODS

PRs dataset, composed of 117 subjects with distomolar teeth and 146 subjects without distomolar teeth, was constructed. These images were assessed using AlexNet, DarkNet, DenseNet, EfficientNet, GoogLeNet, ResNet, MobileNet, NasNet-Mobile, VGG, and XceptionNet frameworks for distomolar teeth existence. Considering the moderate number dataset samples, transfer learning was also utilized to improve the performance of these CNN based networks along with 5-fold cross-validation. The final classification was obtained through the fusion of the classifiers results.

RESULTS

Performance of the experimental studies was assessed utilizing accuracy (Acc), sensitivity (sen), specificity (spe) and precision (pre) metrics. Best accuracy value of 96.2 % was obtained for the fusion of DarkNet, DenseNet, and ResNet, three best individual performing architectures, in distomolar teeth classification problem.

CONCLUSION AND PRACTICAL IMPLICATIONS

In summary, this study has demonstrated the significant potential of CNNs in accurately detecting distomolar teeth in dental radiographs, a critical task for clinical diagnosis and treatment planning. The fusion of CNN architectures, particularly ResNet, Darknet, and DenseNet, has shown exceptional performance, pointing towards the future of artificial intelligence (AI) driven dental diagnostics. Our findings showed that these systems can help clinicians during radiologic examinations.

Development of a Cost-Efficient and Glaucoma-Specialized OD/OC Segmentation Model for Varying Clinical Scenarios

Most existing optic disc (OD) and cup (OC) segmentation models are biased to the dominant size and easy class (normal class), resulting in suboptimal performances on glaucoma-confirmed samples. Thus, these models are not optimal choices for assisting in tracking glaucoma progression and prognosis. Moreover, fully supervised models employing annotated glaucoma samples can achieve superior performances, although restricted by the high cost of collecting and annotating the glaucoma samples. Therefore, in this paper, we are dedicated to developing a glaucoma-specialized model by exploiting low-cost annotated normal fundus images, simultaneously adapting various common scenarios in clinical practice. We employ a contrastive learning and domain adaptation-based model by exploiting shared knowledge from normal samples. To capture glaucoma-related features, we utilize a Gram matrix to encode style information and the domain adaptation strategy to encode domain information, followed by narrowing the style and domain gaps between normal and glaucoma samples by contrastive and adversarial learning, respectively. To validate the efficacy of our proposed model, we conducted experiments utilizing two public datasets to mimic various common scenarios. The results demonstrate the superior performance of our proposed model across multi-scenarios, showcasing its proficiency in both the segmentation- and glaucoma-related metrics. In summary, our study illustrates a concerted effort to target confirmed glaucoma samples, mitigating the inherent bias issue in most existing models. Moreover, we propose an annotation-efficient strategy that exploits low-cost, normal-labeled fundus samples, mitigating the economic- and labor-related burdens by employing a fully supervised strategy. Simultaneously, our approach demonstrates its adaptability across various scenarios, highlighting its potential utility in both assisting in the monitoring of glaucoma progression and assessing glaucoma prognosis.

A novel benign and malignant classification model for lung nodules based on multi-scale interleaved fusion integrated network

One of the precursors of lung cancer is the presence of lung nodules, and accurate identification of their benign or malignant nature is important for the long-term survival of patients. With the development of artificial intelligence, deep learning has become the main method for lung nodule classification. However, successful deep learning models usually require large number of parameters and carefully annotated data. In the field of medical images, the availability of such data is usually limited, which makes deep networks often perform poorly on new test data. In addition, the model based on the linear stacked single branch structure hinders the extraction of multi-scale features and reduces the classification performance. In this paper, to address this problem, we propose a lightweight interleaved fusion integration network with multi-scale feature learning modules, called MIFNet. The MIFNet consists of a series of MIF blocks that efficiently combine multiple convolutional layers containing 1 × 1 and 3 × 3 convolutional kernels with shortcut links to extract multiscale features at different levels and preserving them throughout the block. The model has only 0.7 M parameters and requires low computational cost and memory space compared to many ImageNet pretrained CNN architectures. The proposed MIFNet conducted exhaustive experiments on the reconstructed LUNA16 dataset, achieving impressive results with 94.82% accuracy, 97.34% F1 value, 96.74% precision, 97.10% sensitivity, and 84.75% specificity. The results show that our proposed deep integrated network achieves higher performance than pre-trained deep networks and state-of-the-art methods. This provides an objective and efficient auxiliary method for accurately classifying the type of lung nodule in medical images.

DeepBP: Ensemble deep learning strategy for bioactive peptide prediction

BACKGROUND

Bioactive peptides are important bioactive molecules composed of short-chain amino acids that play various crucial roles in the body, such as regulating physiological processes and promoting immune responses and antibacterial effects. Due to their significance, bioactive peptides have broad application potential in drug development, food science, and biotechnology. Among them, understanding their biological mechanisms will contribute to new ideas for drug discovery and disease treatment.

RESULTS

This study employs generative adversarial capsule networks (CapsuleGAN), gated recurrent units (GRU), and convolutional neural networks (CNN) as base classifiers to achieve ensemble learning through voting methods, which not only obtains high-precision prediction results on the angiotensin-converting enzyme (ACE) inhibitory peptides dataset and the anticancer peptides (ACP) dataset but also demonstrates effective model performance. For this method, we first utilized the protein language model-evolutionary scale modeling (ESM-2)-to extract relevant features for the ACE inhibitory peptides and ACP datasets. Following feature extraction, we trained three deep learning models-CapsuleGAN, GRU, and CNN-while continuously adjusting the model parameters throughout the training process. Finally, during the voting stage, different weights were assigned to the models based on their prediction accuracy, allowing full utilization of the model's performance. Experimental results show that on the ACE inhibitory peptide dataset, the balanced accuracy is 0.926, the Matthews correlation coefficient (MCC) is 0.831, and the area under the curve is 0.966; on the ACP dataset, the accuracy (ACC) is 0.779, and the MCC is 0.558. The experimental results on both datasets are superior to existing methods, demonstrating the effectiveness of the experimental approach.

CONCLUSION

In this study, CapsuleGAN, GRU, and CNN were successfully employed as base classifiers to implement ensemble learning, which not only achieved good results in the prediction of two datasets but also surpassed existing methods. The ability to predict peptides with strong ACE inhibitory activity and ACPs more accurately and quickly is significant, and this work provides valuable insights for predicting other functional peptides. The source code and dataset for this experiment are publicly available at https://github.com/Zhou-Jianren/bioactive-peptides .

Compare three deep learning-based artificial intelligence models for classification of calcified lumbar disc herniation: a multicenter diagnostic study

Objective

To develop and validate an artificial intelligence diagnostic model for identifying calcified lumbar disc herniation based on lateral lumbar magnetic resonance imaging(MRI).

Methods

During the period from January 2019 to March 2024, patients meeting the inclusion criteria were collected. All patients had undergone both lumbar spine MRI and computed tomography(CT) examinations, with regions of interest (ROI) clearly marked on the lumbar sagittal MRI images. The participants were then divided into separate sets for training, testing, and external validation. Ultimately, we developed a deep learning model using the ResNet-34 algorithm model and evaluated its diagnostic efficacy.

Results

A total of 1,224 eligible patients were included in this study, consisting of 610 males and 614 females, with an average age of 53.34 ± 10.61 years. Notably, the test datasets displayed an impressive classification accuracy rate of 91.67%, whereas the external validation datasets achieved a classification accuracy rate of 88.76%. Among the test datasets, the ResNet34 model outperformed other models, yielding the highest area under the curve (AUC) of 0.96 (95% CI: 0.93, 0.99). Additionally, the ResNet34 model also exhibited superior performance in the external validation datasets, exhibiting an AUC of 0.88 (95% CI: 0.80, 0.93).

Conclusion

In this study, we established a deep learning model with excellent performance in identifying calcified intervertebral discs, thereby offering a valuable and efficient diagnostic tool for clinical surgeons.

Deep learning-based automatic image classification of oral cancer cells acquiring chemoresistance in vitro

Cell shape reflects the spatial configuration resulting from the equilibrium of cellular and environmental signals and is considered a highly relevant indicator of its function and biological properties. For cancer cells, various physiological and environmental challenges, including chemotherapy, cause a cell state transition, which is accompanied by a continuous morphological alteration that is often extremely difficult to recognize even by direct microscopic inspection. To determine whether deep learning-based image analysis enables the detection of cell shape reflecting a crucial cell state alteration, we used the oral cancer cell line resistant to chemotherapy but having cell morphology nearly indiscernible from its non-resistant parental cells. We then implemented the automatic approach via deep learning methods based on EfficienNet-B3 models, along with over- and down-sampling techniques to determine whether image analysis of the Convolutional Neural Network (CNN) can accomplish three-class classification of non-cancer cells vs. cancer cells with and without chemoresistance. We also examine the capability of CNN-based image analysis to approximate the composition of chemoresistant cancer cells within a population. We show that the classification model achieves at least 98.33% accuracy by the CNN model trained with over- and down-sampling techniques. For heterogeneous populations, the best model can approximate the true proportions of non-chemoresistant and chemoresistant cancer cells with Root Mean Square Error (RMSE) reduced to 0.16 by Ensemble Learning (EL). In conclusion, our study demonstrates the potential of CNN models to identify altered cell shapes that are visually challenging to recognize, thus supporting future applications with this automatic approach to image analysis.

Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images

PURPOSE

A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data for deep learning. In that case, we can consider a case where a classifier trained at a certain medical institution that has a sufficient number of data is reused at other institutions. However, when using data from multiple institutions, it is necessary to unify the feature distribution because the feature of the data differs due to differences in data acquisition conditions.

METHODS

To unify the feature distribution, the data from Institution T are transformed to have the closer distribution to that from Institution S by applying a domain transformation using semi-supervised CycleGAN. The proposed method enhances CycleGAN considering the feature distribution of classes for making appropriate domain transformation for classification. In addition, to address the problem of imbalanced data with different numbers of data for each cancer type, several methods dealing with imbalanced data are applied to semi-supervised CycleGAN.

RESULTS

The experimental results showed that the classification performance was enhanced when the dataset from Institution S was used as training data and the testing dataset from Institution T was classified after applying domain transformation. In addition, focal loss contributed to improving the mean F1 score the best as a method that addresses the class imbalance.

CONCLUSION

The proposed method achieved the domain transformation of thyroid tissue images between two domains, where it retained the important features related to the classes across domains and showed the best F1 score with significant differences compared with other methods. In addition, the proposed method was further enhanced by addressing the class imbalance of the dataset.

Artificial intelligence-based motion tracking in cancer radiotherapy: A review

Radiotherapy aims to deliver a prescribed dose to the tumor while sparing neighboring organs at risk (OARs). Increasingly complex treatment techniques such as volumetric modulated arc therapy (VMAT), stereotactic radiosurgery (SRS), stereotactic body radiotherapy (SBRT), and proton therapy have been developed to deliver doses more precisely to the target. While such technologies have improved dose delivery, the implementation of intra-fraction motion management to verify tumor position at the time of treatment has become increasingly relevant. Artificial intelligence (AI) has recently demonstrated great potential for real-time tracking of tumors during treatment. However, AI-based motion management faces several challenges, including bias in training data, poor transparency, difficult data collection, complex workflows and quality assurance, and limited sample sizes. This review presents the AI algorithms used for chest, abdomen, and pelvic tumor motion management/tracking for radiotherapy and provides a literature summary on the topic. We will also discuss the limitations of these AI-based studies and propose potential improvements.

Deep learning Radiomics Based on Two-Dimensional Ultrasound for Predicting the Efficacy of Neoadjuvant Chemotherapy in Breast Cancer

We investigate the predictive value of a comprehensive model based on preoperative ultrasound radiomics, deep learning, and clinical features for pathological complete response (pCR) after neoadjuvant chemotherapy (NAC) for the breast cancer. We enrolled 155 patients with pathologically confirmed breast cancer who underwent NAC. The patients were randomly divided into the training set and the validation set in the ratio of 7:3. The deep learning and radiomics features of pre-treatment ultrasound images were extracted, and the random forest recursive elimination algorithm and the least absolute shrinkage and selection operator were used for feature screening and DL-Score and Rad-Score construction. According to multifactorial logistic regression, independent clinical predictors, DL-Score, and Rad-Score were selected to construct the comprehensive prediction model DLRC. The performance of the model was evaluated in terms of its predictive effect, and clinical practicability. Compared to the clinical, radiomics (Rad-Score), and deep learning (DL-Score) models, the DLRC accurately predicted the pCR status, with an area under the curve (AUC) of 0.937 (95%CI: 0.895-0.970) in the training set and 0.914 (95%CI: 0.838-0.973) in the validation set. Moreover, decision curve analysis confirmed that the DLRC had the highest clinical value among all models. The comprehensive model DLRC based on ultrasound radiomics, deep learning, and clinical features can effectively and accurately predict the pCR status of breast cancer after NAC, which is conducive to assisting clinical personalized diagnosis and treatment plan.

Fast Multiphoton Microscopic Imaging Joint Image Super-Resolution for Automated Gleason Grading of Prostate Cancers

Gleason grading system is dependable for quantifying prostate cancer. This paper introduces a fast multiphoton microscopic imaging method via deep learning for automatic Gleason grading. Due to the contradiction between multiphoton microscopy (MPM) imaging speed and quality, a deep learning architecture (SwinIR) is used for image super-resolution to address this issue. The quality of low-resolution image is improved, which increased the acquisition speed from 7.55 s per frame to 0.24 s per frame. A classification network (Swin Transformer) was introduced for automated Gleason grading. The classification accuracy and Macro-F1 achieved by training on high-resolution images are respectively 90.9% and 90.9%. For training on super-resolution images, the classification accuracy and Macro-F1 are respectively 89.9% and 89.9%. It shows that super-resolution image can provide a comparable performance to high-resolution image. Our results suggested that MPM joint image super-resolution and automatic classification methods hold the potential to be a real-time clinical diagnostic tool for prostate cancer diagnosis.

AI Detection of Glottic Neoplasm Using Voice Signals, Demographics, and Structured Medical Records

OBJECTIVE

This study investigated whether artificial intelligence (AI) models combining voice signals, demographics, and structured medical records can detect glottic neoplasm from benign voice disorders.

METHODS

We used a primary dataset containing 2-3 s of vowel "ah", demographics, and 26 items of structured medical records (e.g., symptoms, comorbidity, smoking and alcohol consumption, vocal demand) from 60 patients with pathology-proved glottic neoplasm (i.e., squamous cell carcinoma, carcinoma in situ, and dysplasia) and 1940 patients with benign voice disorders. The validation dataset comprised data from 23 patients with glottic neoplasm and 1331 patients with benign disorders. The AI model combined convolutional neural networks, gated recurrent units, and attention layers. We used 10-fold cross-validation (training-validation-testing: 8-1-1) and preserved the percentage between neoplasm and benign disorders in each fold.

RESULTS

Results from the AI model using voice signals reached an area under the ROC curve (AUC) value of 0.631, and additional demographics increased this to 0.807. The highest AUC of 0.878 was achieved when combining voice, demographics, and medical records (sensitivity: 0.783, specificity: 0.816, accuracy: 0.815). External validation yielded an AUC value of 0.785 (voice plus demographics; sensitivity: 0.739, specificity: 0.745, accuracy: 0.745). Subanalysis showed that AI had higher sensitivity but lower specificity than human assessment (p < 0.01). The accuracy of AI detection with additional medical records was comparable with human assessment (82% vs. 83%, p = 0.78).

CONCLUSIONS

Voice signal alone was insufficient for AI differentiation between glottic neoplasm and benign voice disorders, but additional demographics and medical records notably improved AI performance and approximated the prediction accuracy of humans.

LEVEL OF EVIDENCE

NA Laryngoscope, 134:4585-4592, 2024.

Older Adult Fall Risk Prediction with Deep Learning and Timed Up and Go (TUG) Test Data

Falls are a major health hazard for older adults; therefore, in the context of an aging population, predicting the risk of a patient suffering falls in the near future is of great impact for health care systems. Currently, the standard prospective fall risk assessment instrument relies on a set of clinical and functional mobility assessment tools, one of them being the Timed Up and Go (TUG) test. Recently, wearable inertial measurement units (IMUs) have been proposed to capture motion data that would allow for the building of estimates of fall risk. The hypothesis of this study is that the data gathered from IMU readings while the patient is performing the TUG test can be used to build a predictive model that would provide an estimate of the probability of suffering a fall in the near future, i.e., assessing prospective fall risk. This study applies deep learning convolutional neural networks (CNN) and recurrent neural networks (RNN) to build such predictive models based on features extracted from IMU data acquired during TUG test realizations. Data were obtained from a cohort of 106 older adults wearing wireless IMU sensors with sampling frequencies of 100 Hz while performing the TUG test. The dependent variable is a binary variable that is true if the patient suffered a fall in the six-month follow-up period. This variable was used as the output variable for the supervised training and validations of the deep learning architectures and competing machine learning approaches. A hold-out validation process using 75 subjects for training and 31 subjects for testing was repeated one hundred times to obtain robust estimations of model performances At each repetition, 5-fold cross-validation was carried out to select the best model over the training subset. Best results were achieved by a bidirectional long short-term memory (BLSTM), obtaining an accuracy of 0.83 and AUC of 0.73 with good sensitivity and specificity values.

Deep Learning-Based Detection of Impacted Teeth on Panoramic Radiographs

Objective

The aim is to detect impacted teeth in panoramic radiology by refining the pretrained MedSAM model.

Study design

Impacted teeth are dental issues that can cause complications and are diagnosed via radiographs. We modified SAM model for individual tooth segmentation using 1016 X-ray images. The dataset was split into training, validation, and testing sets, with a ratio of 16:3:1. We enhanced the SAM model to automatically detect impacted teeth by focusing on the tooth's center for more accurate results.

Results

With 200 epochs, batch size equals to 1, and a learning rate of 0.001, random images trained the model. Results on the test set showcased performance up to an accuracy of 86.73%, F1-score of 0.5350, and IoU of 0.3652 on SAM-related models.

Conclusion

This study fine-tunes MedSAM for impacted tooth segmentation in X-ray images, aiding dental diagnoses. Further improvements on model accuracy and selection are essential for enhancing dental practitioners' diagnostic capabilities.

Exploring the Impact of Model Complexity on Laryngeal Cancer Detection

Background: Laryngeal cancer accounts for a third of all head and neck malignancies, necessitating timely detection for effective treatment and enhanced patient outcomes. Machine learning shows promise in medical diagnostics, but the impact of model complexity on diagnostic efficacy in laryngeal cancer detection can be ambiguous. Methods: In this study, we examine the relationship between model sophistication and diagnostic efficacy by evaluating three approaches: Logistic Regression, a small neural network with 4 layers of neurons and a more complex convolutional neural network with 50 layers and examine their efficacy on laryngeal cancer detection on computed tomography images. Results: Logistic regression achieved 82.5% accuracy. The 4-Layer NN reached 87.2% accuracy, while ResNet-50, a deep learning architecture, achieved the highest accuracy at 92.6%. Its deep learning capabilities excelled in discerning fine-grained CT image features. Conclusion: Our study highlights the choices involved in selecting a laryngeal cancer detection model. Logistic regression is interpretable but may struggle with complex patterns. The 4-Layer NN balances complexity and accuracy. ResNet-50 excels in image classification but demands resources. This research advances understanding affect machine learning model complexity could have on learning features of laryngeal tumor features in contrast CT images for purposes of disease prediction.

Advancing Sensing Resolution of Impedance Hand Gesture Recognition Devices

Gestures are composed of motion information (e.g. movements of fingers) and force information (e.g. the force exerted on fingers when interacting with other objects). Current hand gesture recognition solutions such as cameras and strain sensors primarily focus on correlating hand gestures with motion information and force information is seldom addressed. Here we propose a bio-impedance wearable that can recognize hand gestures utilizing both motion information and force information. Compared with previous impedance-based gesture recognition devices that can only recognize a few multi-degrees-of-freedom gestures, the proposed device can recognize 6 single-degree-of-freedom gestures and 20 multiple-degrees-of-freedom gestures, including 8 gestures in 2 force levels. The device uses textile electrodes, is benchmarked over a selected frequency spectrum, and uses a new drive pattern. Experimental results show that 179 kHz achieves the highest signal-to-noise ratio (SNR) and reveals the most distinct features. By analyzing the 49,920 samples from 6 participants, the device is demonstrated to have an average recognition accuracy of 98.96%. As a comparison, the medical electrodes achieved an accuracy of 98.05%.

Deep Learning for Automatic Knee Osteoarthritis Severity Grading and Classification

Introduction

Knee osteoarthritis (OA) is a prevalent condition that significantly impacts the quality of life, often leading to the need for knee replacement surgery. Accurate and timely identification of knee degeneration is crucial for effective treatment and management. Traditional methods of diagnosing OA rely heavily on radiological assessments, which can be time-consuming and subjective. This study aims to address these challenges by developing a deep learning-based method to predict the likelihood of knee replacement and the Kellgren-Lawrence (KL) grade of knee OA from X-ray images.

Methodology

We employed the Osteoarthritis Initiative (OAI) dataset and utilized a transfer learning approach with the Inception V3 architecture to enhance the accuracy of OA detection. Our approach involved training 14 different models-Xception, VGG16, VGG19, ResNet50, ResNet101, ResNet152, ResNet50V2, ResNet101V2, ResNet152V2, Inception V3, Inception, ResNetV2, DenseNet121, DenseNet169, DenseNet201-and comparing their performance.

Results

The study incorporated pixel ratio computation and picture pre-processing, alongside a decision tree model for prediction. Our experiments revealed that the Inception V3 model achieved the highest training accuracy of 91% and testing accuracy of 67%, with notable performance in both training and validation phases. This model effectively identified the presence and severity of OA, correlating with the Kellgren-Lawrence scale and facilitating the assessment of knee replacement needs.

Conclusion

By integrating advanced deep learning techniques with radiological diagnostics, our methodology supports radiologists in making more accurate and prompt decisions regarding knee degeneration. The Inception V3 model stands out as the optimal choice for knee X-ray analysis, contributing to more efficient and timely healthcare delivery for patients with knee osteoarthritis.

Identifying nucleotide-binding leucine-rich repeat receptor and pathogen effector pairing using transfer-learning and bilinear attention network

MOTIVATION

Nucleotide-binding leucine-rich repeat (NLR) family is a class of immune receptors capable of detecting and defending against pathogen invasion. They have been widely used in crop breeding. Notably, the correspondence between NLRs and effectors (CNE) determines the applicability and effectiveness of NLRs. Unfortunately, CNE data is very scarce. In fact, we've found a substantial 91 291 NLRs confirmed via wet experiments and bioinformatics methods but only 387 CNEs are recognized, which greatly restricts the potential application of NLRs.

RESULTS

We propose a deep learning algorithm called ProNEP to identify NLR-effector pairs in a high-throughput manner. Specifically, we conceptualized the CNE prediction task as a protein-protein interaction (PPI) prediction task. Then, ProNEP predicts the interaction between NLRs and effectors by combining the transfer learning with a bilinear attention network. ProNEP achieves superior performance against state-of-the-art models designed for PPI predictions. Based on ProNEP, we conduct extensive identification of potential CNEs for 91 291 NLRs. With the rapid accumulation of genomic data, we expect that this tool will be widely used to predict CNEs in new species, advancing biology, immunology, and breeding.

AVAILABILITY AND IMPLEMENTATION

The ProNEP is available at http://nerrd.cn/#/prediction. The project code is available at https://github.com/QiaoYJYJ/ProNEP.

An AI-Based Low-Risk Lung Health Image Visualization Framework Using LR-ULDCT

In this article, we propose an AI-based low-risk visualization framework for lung health monitoring using low-resolution ultra-low-dose CT (LR-ULDCT). We present a novel deep cascade processing workflow to achieve diagnostic visualization on LR-ULDCT (<0.3 mSv) at par high-resolution CT (HRCT) of 100 mSV radiation technology. To this end, we build a low-risk and affordable deep cascade network comprising three sequential deep processes: restoration, super-resolution (SR), and segmentation. Given degraded LR-ULDCT, the first novel network unsupervisedly learns restoration function from augmenting patch-based dictionaries and residuals. The restored version is then super-resolved (SR) for target (sensor) resolution. Here, we combine perceptual and adversarial losses in novel GAN to establish the closeness between probability distributions of generated SR-ULDCT and restored LR-ULDCT. Thus SR-ULDCT is presented to the segmentation network that first separates the chest portion from SR-ULDCT followed by lobe-wise colorization. Finally, we extract five lobes to account for the presence of ground glass opacity (GGO) in the lung. Hence, our AI-based system provides low-risk visualization of input degraded LR-ULDCT to various stages, i.e., restored LR-ULDCT, restored SR-ULDCT, and segmented SR-ULDCT, and achieves diagnostic power of HRCT. We perform case studies by experimenting on real datasets of COVID-19, pneumonia, and pulmonary edema/congestion while comparing our results with state-of-the-art. Ablation experiments are conducted for better visualizing different operating pipelines. Finally, we present a verification report by fourteen (14) experienced radiologists and pulmonologists.

Effective Connectivity Estimation by a Hybrid Neural Network, Empirical Wavelet Transform, and Bayesian Optimization

Accurately measuring nonlinear effective connectivity is a crucial step in investigating brain functions. Brain signals like EEG is nonstationary. Many effective connectivity methods have been proposed but they have drawbacks in their models such as a weakness in proposing a way for hyperparameter and time lag selection as well as dealing with non-stationarity of the time series. This paper proposes an effective connectivity model based on a hybrid neural network model which uses Empirical Wavelet Transform (EWT) and a long short-term memory network (LSTM). The best hyperparameters and time lag are selected using Bayesian Optimization (BO). Due to the importance of generalizability in neural networks and calculating GC, an algorithm was proposed to choose the best generalizable weights. The model was evaluated using simulated and real EEG data consisting of attention deficit hyperactivity disorder (ADHD) and healthy subjects. The proposed model's performance on simulated data was evaluated by comparing it with other neural networks, including LSTM, CNN-LSTM, GRU, RNN, and MLP, using a Blocked cross-validation approach. GC of the simulated data was compared with GRU, linear Granger causality (LGC), Kernel Granger Causality (KGC), Partial Directed Coherence (PDC), and Directed Transfer Function (DTF). Our results demonstrated that the proposed model was superior to the mentioned models. Another advantage of our model is robustness against noise. The results showed that the proposed model can identify the connections in noisy conditions. The comparison of the effective connectivity of ADHD and the healthy group showed that the results are in accordance with previous studies.

Θ-Net: A Deep Neural Network Architecture for the Resolution Enhancement of Phase-Modulated Optical Micrographs In Silico

Optical microscopy is widely regarded to be an indispensable tool in healthcare and manufacturing quality control processes, although its inability to resolve structures separated by a lateral distance under ~200 nm has culminated in the emergence of a new field named fluorescence nanoscopy, while this too is prone to several caveats (namely phototoxicity, interference caused by exogenous probes and cost). In this regard, we present a triplet string of concatenated O-Net ('bead') architectures (termed 'Θ-Net' in the present study) as a cost-efficient and non-invasive approach to enhancing the resolution of non-fluorescent phase-modulated optical microscopical images in silico. The quality of the afore-mentioned enhanced resolution (ER) images was compared with that obtained via other popular frameworks (such as ANNA-PALM, BSRGAN and 3D RCAN), with the Θ-Net-generated ER images depicting an increased level of detail (unlike previous DNNs). In addition, the use of cross-domain (transfer) learning to enhance the capabilities of models trained on differential interference contrast (DIC) datasets [where phasic variations are not as prominently manifested as amplitude/intensity differences in the individual pixels unlike phase-contrast microscopy (PCM)] has resulted in the Θ-Net-generated images closely approximating that of the expected (ground truth) images for both the DIC and PCM datasets. This thus demonstrates the viability of our current Θ-Net architecture in attaining highly resolved images under poor signal-to-noise ratios while eliminating the need for a priori PSF and OTF information, thereby potentially impacting several engineering fronts (particularly biomedical imaging and sensing, precision engineering and optical metrology).

Multi-label remote sensing classification with self-supervised gated multi-modal transformers

Introduction

With the great success of Transformers in the field of machine learning, it is also gradually attracting widespread interest in the field of remote sensing (RS). However, the research in the field of remote sensing has been hampered by the lack of large labeled data sets and the inconsistency of data modes caused by the diversity of RS platforms. With the rise of self-supervised learning (SSL) algorithms in recent years, RS researchers began to pay attention to the application of "pre-training and fine-tuning" paradigm in RS. However, there are few researches on multi-modal data fusion in remote sensing field. Most of them choose to use only one of the modal data or simply splice multiple modal data roughly.

Method

In order to study a more efficient multi-modal data fusion scheme, we propose a multi-modal fusion mechanism based on gated unit control (MGSViT). In this paper, we pretrain the ViT model based on BigEarthNet dataset by combining two commonly used SSL algorithms, and propose an intra-modal and inter-modal gated fusion unit for feature learning by combining multispectral (MS) and synthetic aperture radar (SAR). Our method can effectively combine different modal data to extract key feature information.

Results and discussion

After fine-tuning and comparison experiments, we outperform the most advanced algorithms in all downstream classification tasks. The validity of our proposed method is verified.

Integrated Transfer Learning and Multitask Learning Strategies to Construct Graph Neural Network Models for Predicting Bioaccumulation Parameters of Chemicals

Accurate prediction of parameters related to the environmental exposure of chemicals is crucial for the sound management of chemicals. However, the lack of large data sets for training models may result in poor prediction accuracy and robustness. Herein, integrated transfer learning (TL) and multitask learning (MTL) was proposed for constructing a graph neural network (GNN) model (abbreviated as TL-MTL-GNN model) using n-octanol/water partition coefficients as a source domain. The TL-MTL-GNN model was trained to predict three bioaccumulation parameters based on enlarged data sets that cover 2496 compounds with at least one bioaccumulation parameter. Results show that the TL-MTL-GNN model outperformed single-task GNN models with and without the TL, as well as conventional machine learning models trained with molecular descriptors or fingerprints. Applicability domains were characterized by a state-of-the-art structure-activity landscape-based (abbreviated as ADSAL) methodology. The TL-MTL-GNN model coupled with the optimal ADSAL was employed to predict bioaccumulation parameters for around 60,000 chemicals, with more than 13,000 compounds identified as bioaccumulative chemicals. The high predictive accuracy and robustness of the TL-MTL-GNN model demonstrate the feasibility of integrating the TL and MTL strategy in modeling small-sized data sets. The strategy holds significant potential for addressing small data challenges in modeling environmental chemicals.

Standards for quantitative assessments by coronary computed tomography angiography (CCTA): An expert consensus document of the society of cardiovascular computed tomography (SCCT)

In current clinical practice, qualitative or semi-quantitative measures are primarily used to report coronary artery disease on cardiac CT. With advancements in cardiac CT technology and automated post-processing tools, quantitative measures of coronary disease severity have become more broadly available. Quantitative coronary CT angiography has great potential value for clinical management of patients, but also for research. This document aims to provide definitions and standards for the performance and reporting of quantitative measures of coronary artery disease by cardiac CT.

Artificial Intelligence in Forensic Sciences: A Systematic Review of Past and Current Applications and Future Perspectives

The aim of this study is to review the available knowledge concerning the use of artificial Intelligence (AI) in general in different areas of Forensic Sciences from human identification to postmortem interval estimation and the estimation of different causes of death. This paper aims to emphasize the different uses of AI, especially in Forensic Medicine, and elucidate its technical part. This will be achieved through an explanation of different technologies that have been so far employed and through new ideas that may contribute as a first step to the adoption of new practices and to the development of new technologies. A systematic literature search was performed in accordance with the Preferred Reported Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines in the PubMed Database and Cochrane Central Library. Neither time nor regional constrictions were adopted, and all the included papers were written in English. Terms used were MACHINE AND LEARNING AND FORENSIC AND PATHOLOGY and ARTIFICIAL AND INTELIGENCE AND FORENSIC AND PATHOLOGY. Quality control was performed using the Joanna Briggs Institute critical appraisal tools. A search of 224 articles was performed. Seven more articles were extracted from the references of the initial selection. After excluding all non-relevant articles, the remaining 45 articles were thoroughly reviewed through the whole text. A final number of 33 papers were identified as relevant to the subject, in accordance with the criteria previously established. It must be clear that AI is not meant to replace forensic experts but to assist them in their everyday work life.

TransC-ac4C: Identification of N4-Acetylcytidine (ac4C) Sites in mRNA Using Deep Learning

N4-acetylcytidine (ac4C) is a post-transcriptional modification in mRNA that is critical in mRNA translation in terms of stability and regulation. In the past few years, numerous approaches employing convolutional neural networks (CNN) and Transformer have been proposed for the identification of ac4C sites, with each variety of approaches processing distinct characteristics. CNN-based methods excel at extracting local features and positional information, whereas Transformer-based ones stands out in establishing long-range dependencies and generating global representations. Given the importance of both local and global features in mRNA ac4C sites identification, we propose a novel method termed TransC-ac4C which combines CNN and Transformer together for enhancing the feature extraction capability and improving the identification accuracy. Five different feature encoding strategies (One-hot, NCP, ND, EIIP, and K-mer) are employed to generate the mRNA sequence representations, in which way the sequence attributes and physical and chemical properties of the sequences can be embedded. To strengthen the relevance of features, we construct a novel feature fusion method. Firstly, the CNN is employed to process five single features, stitch them together and feed them to the Transformer layer. Then, our approach employs CNN to extract local features and Transformer subsequently to establish global long-range dependencies among extracted features. We use 5-fold cross-validation to evaluate the model, and the evaluation indicators are significantly improved. The prediction accuracy of the two datasets is as high as 81.42% and 80.69%, respectively. It demonstrates the stronger competitiveness and generalization performance of our model.

Using Domain Adaptation and Inductive Transfer Learning to Improve Patient Outcome Prediction in the Intensive Care Unit: Retrospective Observational Study

BACKGROUND

Accurate patient outcome prediction in the intensive care unit (ICU) can potentially lead to more effective and efficient patient care. Deep learning models are capable of learning from data to accurately predict patient outcomes, but they typically require large amounts of data and computational resources. Transfer learning (TL) can help in scenarios where data and computational resources are scarce by leveraging pretrained models. While TL has been widely used in medical imaging and natural language processing, it has been rare in electronic health record (EHR) analysis. Furthermore, domain adaptation (DA) has been the most common TL method in general, whereas inductive transfer learning (ITL) has been rare. To the best of our knowledge, DA and ITL have never been studied in-depth in the context of EHR-based ICU patient outcome prediction.

OBJECTIVE

This study investigated DA, as well as rarely researched ITL, in EHR-based ICU patient outcome prediction under simulated, varying levels of data scarcity.

METHODS

Two patient cohorts were used in this study: (1) eCritical, a multicenter ICU data from 55,689 unique admission records from 48,672 unique patients admitted to 15 medical-surgical ICUs in Alberta, Canada, between March 2013 and December 2019, and (2) Medical Information Mart for Intensive Care III, a single-center, publicly available ICU data set from Boston, Massachusetts, acquired between 2001 and 2012 containing 61,532 admission records from 46,476 patients. We compared DA and ITL models with baseline models (without TL) of fully connected neural networks, logistic regression, and lasso regression in the prediction of 30-day mortality, acute kidney injury, ICU length of stay, and hospital length of stay. Random subsets of training data, ranging from 1% to 75%, as well as the full data set, were used to compare the performances of DA and ITL with the baseline models at various levels of data scarcity.

RESULTS

Overall, the ITL models outperformed the baseline models in 55 of 56 comparisons (all P values <.001). The DA models outperformed the baseline models in 45 of 56 comparisons (all P values <.001). ITL resulted in better performance than DA in terms of the number of times and the margin with which it outperformed the baseline models. In 11 of 16 cases (8 of 8 for ITL and 3 of 8 for DA), TL models outperformed baseline models when trained using 1% data subset.

CONCLUSIONS

TL-based ICU patient outcome prediction models are useful in data-scarce scenarios. The results of this study can be used to estimate ICU outcome prediction performance at different levels of data scarcity, with and without TL. The publicly available pretrained models from this study can serve as building blocks in further research for the development and validation of models in other ICU cohorts and outcomes.

Application of improved Unet network in the recognition and segmentation of lung CT images in patients with pneumoconiosis

BACKGROUND

Pneumoconiosis has a significant impact on the quality of patient survival. This study aims to evaluate the performance and application value of improved Unet network technology in the recognition and segmentation of lesion areas of lung CT images in patients with pneumoconiosis.

METHODS

A total of 1212 lung CT images of patients with pneumoconiosis were retrospectively included. The improved Unet network was used to identify and segment the CT image regions of the patients' lungs, and the image data of the granular regions of the lungs were processed by the watershed and region growing algorithms. After random sorting, 848 data were selected into the training set and 364 data into the validation set. The experimental dataset underwent data augmentation and were used for model training and validation to evaluate segmentation performance. The segmentation results were compared with FCN-8s, Unet network (Base), Unet (Squeeze-and-Excitation, SE + Rectified Linear Unit, ReLU), and Unet + + networks.

RESULTS

In the segmentation of lung CT granular region with the improved Unet network, the four evaluation indexes of Dice similarity coefficient, positive prediction value (PPV), sensitivity coefficient (SC) and mean intersection over union (MIoU) reached 0.848, 0.884, 0.895 and 0.885, respectively, increasing by 7.6%, 13.3%, 3.9% and 6.4%, respectively, compared with those of Unet network (Base), and increasing by 187.5%, 249.4%, 131.9% and 51.0%, respectively, compared with those of FCN-8s, and increasing by 14.0%, 31.2%, 4.7% and 9.7%, respectively, compared with those of Unet network (SE + ReLU), while the segmentation performance was also not inferior to that of the Unet + + network.

CONCLUSIONS

The improved Unet network proposed shows good performance in the recognition and segmentation of abnormal regions in lung CT images in patients with pneumoconiosis, showing potential application value for assisting clinical decision-making.

Efficient state of charge estimation of lithium-ion batteries in electric vehicles using evolutionary intelligence-assisted GLA-CNN-Bi-LSTM deep learning model

The battery's performance heavily influences the safety, dependability, and operational efficiency of electric vehicles (EVs). This paper introduces an innovative hybrid deep learning architecture that dramatically enhances the estimation of the state of charge (SoC) of lithium-ion (Li-ion) batteries, crucial for efficient EV operation. Our model uniquely integrates a convolutional neural network (CNN) with bidirectional long short-term memory (Bi-LSTM), optimized through evolutionary intelligence, enabling an advanced level of precision in SoC estimation. A novel aspect of this work is the application of the Group Learning Algorithm (GLA) to tune the hyperparameters of the CNN-Bi-LSTM network meticulously. This approach not only refines the model's accuracy but also significantly enhances its efficiency by optimizing each parameter to best capture and integrate both spatial and temporal information from the battery data. This is in stark contrast to conventional models that typically focus on either spatial or temporal data, but not both effectively. The model's robustness is further demonstrated through its training across six diverse datasets that represent a range of EV discharge profiles, including the Highway Fuel Economy Test (HWFET), the US06 test, the Beijing Dynamic Stress Test (BJDST), the dynamic stress test (DST), the federal urban driving schedule (FUDS), and the urban development driving schedule (UDDS). These tests are crucial for ensuring that the model can perform under various real-world conditions. Experimentally, our hybrid model not only surpasses the performance of existing LSTM and CNN frameworks in tracking SoC estimation but also achieves an impressively quick convergence to true SoC values, maintaining an average root mean square error (RMSE) of less than 1 %. Furthermore, the experimental outcomes suggest that this new deep learning methodology outstrips conventional approaches in both convergence speed and estimation accuracy, thus promising to significantly enhance battery life and overall EV efficiency.

Utilization of machine learning methods for prediction of acute and late rectal toxicity due to curative prostate radiotherapy

OBJECTIVE

Rectal toxicity is one of the primary dose-limiting side effects of prostate cancer radiotherapy, and consequential impairment on quality of life in these patients with long survival is an important problem. In this study, we aimed to evaluate the possibility of predicting rectal toxicity with artificial intelligence model which was including certain dosimetric parameters.

MATERIALS AND METHODS

One hundred and thirty-seven patients with a diagnosis of prostate cancer who received curative radiotherapy for prostate +/- pelvic lymphatics were included in the study. The association of the clinical data and dosimetric data between early and late rectal toxicity reported during follow-up was evaluated. The sample size was increased to 274 patients by synthetic data generation method. To determine suitable models, 15 models were studied with machine learning algorithms using Python 2.3, Pycaret library. Random forest classifier was used with to detect active variables.

RESULTS

The area under the curve and accuracy were found to be 0.89-0.97 and 95%-99%, respectively, with machine learning algorithms. The sensitivity values for acute and toxicity were found to be 0.95 and 0.99, respectively.

CONCLUSION

Early or late rectal toxicity can be predicted with a high probability via dosimetric and physical data and machine learning algorithms of patients who underwent prostate +/- pelvic radiotherapy. The fact that rectal toxicity can be predicted before treatment, which may result in limiting the dose and duration of treatment, makes us think that artificial intelligence can enter our daily practice in a short time in this sense.

A deep learning radiomics model based on CT images for predicting the biological activity of hepatic cystic echinococcosis

Introduction: Hepatic cystic echinococcosis (HCE) is a widely seen parasitic infection. Biological activity is crucial for treatment planning. This work aims to explore the potential applications of a deep learning radiomics (DLR) model, based on CT images, in predicting the biological activity grading of hepatic cystic echinococcosis. Methods: A retrospective analysis of 160 patients with hepatic echinococcosis was performed (127 and 33 in training and validation sets). Volume of interests (VOIs) were drawn, and radiomics features and deep neural network features were extracted. Feature selection was performed on the training set, and radiomics score (Rad Score) and deep learning score (Deep Score) were calculated. Seven diagnostics models (based on logistic regression algorithm) for the biological activity grading were constructed using the selected radiomics features and two deep model features respectively. All models were evaluated using the receiver operating characteristic curve, and the area under the curve (AUC) was calculated. A nomogram was constructed using the combined model, and its calibration, discriminatory ability, and clinical utility were assessed. Results: 12, 6 and 10 optimal radiomics features, deep learning features were selected from two deep learning network (DLN) features, respectively. For biological activity grading of hepatic cystic echinococcosis, the combined model demonstrated strong diagnostic performance, with an AUC value of 0.888 (95% CI: 0.837-0.936) in the training set and 0.876 (0.761-0.964) in the validation set. The clinical decision analysis curve indicated promising results, while the calibration curve revealed that the nomogram's prediction result was highly compatible with the actual result. Conclusion: The DLR model can be used for predicting the biological activity grading of hepatic echinococcosis.

Artificial intelligence for ultrasound microflow imaging in breast cancer diagnosis

PURPOSE

To develop and evaluate artificial intelligence (AI) algorithms for ultrasound (US) microflow imaging (MFI) in breast cancer diagnosis.

MATERIALS AND METHODS

We retrospectively collected a dataset consisting of 516 breast lesions (364 benign and 152 malignant) in 471 women who underwent B-mode US and MFI. The internal dataset was split into training (n = 410) and test datasets (n = 106) for developing AI algorithms from deep convolutional neural networks from MFI. AI algorithms were trained to provide malignancy risk (0-100%). The developed AI algorithms were further validated with an independent external dataset of 264 lesions (229 benign and 35 malignant). The diagnostic performance of B-mode US, AI algorithms, or their combinations was evaluated by calculating the area under the receiver operating characteristic curve (AUROC).

RESULTS

The AUROC of the developed three AI algorithms (0.955-0.966) was higher than that of B-mode US (0.842, P < 0.0001). The AUROC of the AI algorithms on the external validation dataset (0.892-0.920) was similar to that of the test dataset. Among the AI algorithms, no significant difference was found in all performance metrics combined with or without B-mode US. Combined B-mode US and AI algorithms had a higher AUROC (0.963-0.972) than that of B-mode US (P < 0.0001). Combining B-mode US and AI algorithms significantly decreased the false-positive rate of BI-RADS category 4A lesions from 87% to 13% (P < 0.0001).

CONCLUSION

AI-based MFI diagnosed breast cancers with better performance than B-mode US, eliminating 74% of false-positive diagnoses in BI-RADS category 4A lesions.

Advancing maxillofacial prosthodontics by using pre-trained convolutional neural networks: Image-based classification of the maxilla

PURPOSE

The study aimed to compare the performance of four pre-trained convolutional neural networks in recognizing seven distinct prosthodontic scenarios involving the maxilla, as a preliminary step in developing an artificial intelligence (AI)-powered prosthesis design system.

MATERIALS AND METHODS

Seven distinct classes, including cleft palate, dentulous maxillectomy, edentulous maxillectomy, reconstructed maxillectomy, completely dentulous, partially edentulous, and completely edentulous, were considered for recognition. Utilizing transfer learning and fine-tuned hyperparameters, four AI models (VGG16, Inception-ResNet-V2, DenseNet-201, and Xception) were employed. The dataset, consisting of 3541 preprocessed intraoral occlusal images, was divided into training, validation, and test sets. Model performance metrics encompassed accuracy, precision, recall, F1 score, area under the receiver operating characteristic curve (AUC), and confusion matrix.

RESULTS

VGG16, Inception-ResNet-V2, DenseNet-201, and Xception demonstrated comparable performance, with maximum test accuracies of 0.92, 0.90, 0.94, and 0.95, respectively. Xception and DenseNet-201 slightly outperformed the other models, particularly compared with InceptionResNet-V2. Precision, recall, and F1 scores exceeded 90% for most classes in Xception and DenseNet-201 and the average AUC values for all models ranged between 0.98 and 1.00.

CONCLUSIONS

While DenseNet-201 and Xception demonstrated superior performance, all models consistently achieved diagnostic accuracy exceeding 90%, highlighting their potential in dental image analysis. This AI application could help work assignments based on difficulty levels and enable the development of an automated diagnosis system at patient admission. It also facilitates prosthesis designing by integrating necessary prosthesis morphology, oral function, and treatment difficulty. Furthermore, it tackles dataset size challenges in model optimization, providing valuable insights for future research.

DCNNLFS: A Dilated Convolutional Neural Network With Late Fusion Strategy for Intelligent Classification of Gastric Histopathology Images

Gastric cancer has a high incidence rate, significantly threatening patients' health. Gastric histopathology images can reliably diagnose related diseases. Still, the data volume of histopathology images is too large, making misdiagnosis or missed diagnosis easy. The classification model based on deep learning has made some progress on gastric histopathology images. However, traditional convolutional neural networks (CNNs) generally use pooling operations, which will reduce the spatial resolution of the image, resulting in poor prediction results. The image feature in previous CNN has a poor perception of details. Therefore, we design a dilated CNN with a late fusion strategy (DCNNLFS) for gastric histopathology image classification. The DCNNLFS model utilizes dilated convolutions, enabling it to expand the receptive field. The dilated convolutions can learn the different contextual information by adjusting the dilation rate. The DCNNLFS model uses a late fusion strategy to enhance the classification ability of DCNNLFS. We run related experiments on a gastric histopathology image dataset to verify the excellence of the DCNNLFS model, where the three metrics Precision, Accuracy, and F1-Score are 0.938, 0.935, and 0.959.

Consensus modeling: Safer transfer learning for small health systems

Predictive modeling is becoming an essential tool for clinical decision support, but health systems with smaller sample sizes may construct suboptimal or overly specific models. Models become over-specific when beside true physiological effects, they also incorporate potentially volatile site-specific artifacts. These artifacts can change suddenly and can render the model unsafe. To obtain safer models, health systems with inadequate sample sizes may adopt one of the following options. First, they can use a generic model, such as one purchased from a vendor, but often such a model is not sufficiently specific to the patient population and is thus suboptimal. Second, they can participate in a research network. Paradoxically though, sites with smaller datasets contribute correspondingly less to the joint model, again rendering the final model suboptimal. Lastly, they can use transfer learning, starting from a model trained on a large data set and updating this model to the local population. This strategy can also result in a model that is over-specific. In this paper we present the consensus modeling paradigm, which uses the help of a large site (source) to reach a consensus model at the small site (target). We evaluate the approach on predicting postoperative complications at two health systems with 9,044 and 38,045 patients (rare outcomes at about 1% positive rate), and conduct a simulation study to understand the performance of consensus modeling relative to the other three approaches as a function of the available training sample size at the target site. We found that consensus modeling exhibited the least over-specificity at either the source or target site and achieved the highest combined predictive performance.