BMC Med Imaging. 2024 Apr 16;24(1):91. doi: 10.1186/s12880-024-01255-y.

ABSTRACT

BACKGROUND: The relationship between the biological pathways related to deep learning radiomics (DLR) and lymph node metastasis (LNM) of breast cancer is still poorly understood. This study explored the value of DLR based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in LNM of invasive breast cancer. It also analyzed the biological significance of DLR phenotype based on genomics.

METHODS: Two cohorts from the Cancer Imaging Archive project were used, one as the training cohort (TCGA-Breast, n = 88) and one as the validation cohort (Breast-MRI-NACT Pilot, n = 57). Radiomics and deep learning features were extracted from preoperative DCE-MRI. After dual selection by principal components analysis (PCA) and relief methods, radiomics and deep learning models for predicting LNM were constructed by the random forest (RF) method. A post-fusion strategy was used to construct the DLR nomograms (DLRNs) for predicting LNM. The performance of the models was evaluated using the receiver operating characteristic (ROC) curve and Delong test. In the training cohort, transcriptome data were downloaded from the UCSC Xena online database, and biological pathways related to the DLR phenotypes were identified. Finally, hub genes were identified to obtain DLR gene expression (RadDeepGene) scores.

RESULTS: DLRNs were based on area under curve (AUC) evaluation (training cohort, AUC = 0.98; validation cohort, AUC = 0.87), which were higher than single radiomics models or GoogLeNet models. The Delong test (radiomics model, P = 0.04; GoogLeNet model, P = 0.01) also validated the above results in the training cohorts, but they were not statistically significant in the validation cohort. The GoogLeNet phenotypes were related to multiple classical tumor signaling pathways, characterizing the biological significance of immune response, signal transduction, and cell death. In all, 20 genes related to GoogLeNet phenotypes were identified, and the RadDeepGene score represented a high risk of LNM (odd ratio = 164.00, P < 0.001).

CONCLUSIONS: DLRNs combining radiomics and deep learning features of DCE-MRI images improved the preoperative prediction of LNM in breast cancer, and the potential biological characteristics of DLRN were identified through genomics.

PMID:38627678 | DOI:10.1186/s12880-024-01255-y

Nat Med. 2024 Apr 16. doi: 10.1038/s41591-024-02915-w. Online ahead of print.

ABSTRACT

Cancer of unknown primary (CUP) site poses diagnostic challenges due to its elusive nature. Many cases of CUP manifest as pleural and peritoneal serous effusions. Leveraging cytological images from 57,220 cases at four tertiary hospitals, we developed a deep-learning method for tumor origin differentiation using cytological histology (TORCH) that can identify malignancy and predict tumor origin in both hydrothorax and ascites. We examined its performance on three internal (n = 12,799) and two external (n = 14,538) testing sets. In both internal and external testing sets, TORCH achieved area under the receiver operating curve values ranging from 0.953 to 0.991 for cancer diagnosis and 0.953 to 0.979 for tumor origin localization. TORCH accurately predicted primary tumor origins, with a top-1 accuracy of 82.6% and top-3 accuracy of 98.9%. Compared with results derived from pathologists, TORCH showed better prediction efficacy (1.677 versus 1.265, P < 0.001), enhancing junior pathologists' diagnostic scores significantly (1.326 versus 1.101, P < 0.001). Patients with CUP whose initial treatment protocol was concordant with TORCH-predicted origins had better overall survival than those who were administrated discordant treatment (27 versus 17 months, P = 0.006). Our study underscores the potential of TORCH as a valuable ancillary tool in clinical practice, although further validation in randomized trials is warranted.

PMID:38627559 | DOI:10.1038/s41591-024-02915-w

World J Urol. 2024 Apr 16;42(1):238. doi: 10.1007/s00345-024-04921-6.

ABSTRACT

BACKGROUND: Accurate estimation of the glomerular filtration rate (GFR) is clinically crucial for determining the status of obstruction, developing treatment strategies, and predicting prognosis in obstructive nephropathy (ON). We aimed to develop a deep learning-based system, named UroAngel, for non-invasive and convenient prediction of single-kidney function level.

METHODS: We retrospectively collected computed tomography urography (CTU) images and emission computed tomography diagnostic reports of 520 ON patients. A 3D U-Net model was used to segment the renal parenchyma, and a logistic regression multi-classification model was used to predict renal function level. We compared the predictive performance of UroAngel with the Modification of Diet in Renal Disease (MDRD), Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations, and two expert radiologists in an additional 40 ON patients to validate clinical effectiveness.

RESULTS: UroAngel based on 3D U-Net convolutional neural network could segment the renal cortex accurately, with a Dice similarity coefficient of 0.861. Using the segmented renal cortex to predict renal function stage had high performance with an accuracy of 0.918, outperforming MDRD and CKD-EPI and two radiologists.

CONCLUSIONS: We proposed an automated 3D U-Net-based analysis system for direct prediction of single-kidney function stage from CTU images. UroAngel could accurately predict single-kidney function in ON patients, providing a novel, reliable, convenient, and non-invasive method.

PMID:38627315 | DOI:10.1007/s00345-024-04921-6

Eur Radiol. 2024 Apr 16. doi: 10.1007/s00330-024-10728-1. Online ahead of print.

ABSTRACT

OBJECTIVES: To build self-supervised foundation models for multicontrast MRI of the whole brain and evaluate their efficacy in assisting diagnosis of brain tumors.

METHODS: In this retrospective study, foundation models were developed using 57,621 enhanced head MRI scans through self-supervised learning with a pretext task of cross-contrast context restoration with two different content dropout schemes. Downstream classifiers were constructed based on the pretrained foundation models and fine-tuned for brain tumor detection, discrimination, and molecular status prediction. Metrics including accuracy, sensitivity, specificity, and area under the ROC curve (AUC) were used to evaluate the performance. Convolutional neural networks trained exclusively on downstream task data were employed for comparative analysis.

RESULTS: The pretrained foundation models demonstrated their ability to extract effective representations from multicontrast whole-brain volumes. The best classifiers, endowed with pretrained weights, showed remarkable performance with accuracies of 94.9, 92.3, and 80.4%, and corresponding AUC values of 0.981, 0.972, and 0.852 on independent test datasets in brain tumor detection, discrimination, and molecular status prediction, respectively. The classifiers with pretrained weights outperformed the convolutional classifiers trained from scratch by approximately 10% in terms of accuracy and AUC across all tasks. The saliency regions in the correctly predicted cases are mainly clustered around the tumors. Classifiers derived from the two dropout schemes differed significantly only in the detection of brain tumors.

CONCLUSIONS: Foundation models obtained from self-supervised learning have demonstrated encouraging potential for scalability and interpretability in downstream brain tumor-related tasks and hold promise for extension to neurological diseases with diffusely distributed lesions.

CLINICAL RELEVANCE STATEMENT: The application of our proposed method to the prediction of key molecular status in gliomas is expected to improve treatment planning and patient outcomes. Additionally, the foundation model we developed could serve as a cornerstone for advancing AI applications in the diagnosis of brain-related diseases.

PMID:38627290 | DOI:10.1007/s00330-024-10728-1

J Imaging Inform Med. 2024 Apr 16. doi: 10.1007/s10278-024-01085-y. Online ahead of print.

ABSTRACT

Architectural distortion (AD) is one of the most common findings on mammograms, and it may represent not only cancer but also a lesion such as a radial scar that may have an associated cancer. AD accounts for 18-45% missed cancer, and the positive predictive value of AD is approximately 74.5%. Early detection of AD leads to early diagnosis and treatment of the cancer and improves the overall prognosis. However, detection of AD is a challenging task. In this work, we propose a new approach for detecting architectural distortion in mammography images by combining preprocessing methods and a novel structure fusion attention model. The proposed structure-focused weighted orientation preprocessing method is composed of the original image, the architecture enhancement map, and the weighted orientation map, highlighting suspicious AD locations. The proposed structure fusion attention model captures the information from different channels and outperforms other models in terms of false positives and top sensitivity, which refers to the maximum sensitivity that a model can achieve under the acceptance of the highest number of false positives, reaching 0.92 top sensitivity with only 0.6590 false positive per image. The findings suggest that the combination of preprocessing methods and a novel network architecture can lead to more accurate and reliable AD detection. Overall, the proposed approach offers a novel perspective on detecting ADs, and we believe that our method can be applied to clinical settings in the future, assisting radiologists in the early detection of ADs from mammography, ultimately leading to early treatment of breast cancer patients.

PMID:38627268 | DOI:10.1007/s10278-024-01085-y

Sci Total Environ. 2024 Apr 14:172466. doi: 10.1016/j.scitotenv.2024.172466. Online ahead of print.

ABSTRACT

The burgeoning issue of plasmid-mediated resistance genes (ARGs) dissemination poses a significant threat to environmental integrity. However, the prediction of ARGs prevalence is overlooked, especially for emerging ARGs that are potentially evolving gene exchange hotspot. Here, we explored to classify plasmid or chromosome sequences and detect resistance gene prevalence by using DNABERT. Initially, the DNABERT fine-tuned in plasmid and chromosome sequences followed by multilayer perceptron (MLP) classifier could achieve 0.764 AUC (Area under curve) on external datasets across 23 genera, outperforming 0.02 AUC than traditional statistic-based model. Furthermore, Escherichia, Pseudomonas single genera based model were also be trained to explore its predict performance to ARGs prevalence detection. By integrating K-mer frequency attributes, our model could boost the performance to predict the prevalence of ARGs in an external dataset in Escherichia with 0.0281-0.0615 AUC and Pseudomonas with 0.0196-0.0928 AUC. Finally, we established a random forest model aimed at forecasting the relative conjugation transfer rate of plasmids with 0.7956 AUC, drawing on data from existing literature. It identifies the plasmid's repression status, cellular density, and temperature as the most important factors influencing transfer frequency. With these two models combined, they provide useful reference for quick and low-cost integrated evaluation of resistance gene transfer, accelerating the process of computer-assisted quantitative risk assessment of ARGs transfer in environmental field.

PMID:38626826 | DOI:10.1016/j.scitotenv.2024.172466

Biomed Mater. 2024 Apr 16. doi: 10.1088/1748-605X/ad3f60. Online ahead of print.

ABSTRACT

Accurate segmentation of coronary artery tree and personalised 3D printing from medical images is essential for CAD diagnosis and treatment. The current literature on 3D printing relies solely on generic models created with different software or 3D coronary artery models manually segmented from medical images. Moreover, there are not many studies examining the bioprintability of a 3D model generated by artificial intelligence (AI) segmentation for complex and branched structures. In this study, deep learning algorithms with transfer learning have been employed for accurate segmentation of the coronary artery tree from medical images to generate printable segmentations. We propose a combination of deep learning and 3D printing, which accurately segments and prints complex vascular patterns in coronary arteries. Then, we performed the 3D printing of the AI-generated coronary artery segmentation for the fabrication of bifurcated hollow vascular structure. Our results indicate improved performance of segmentation with the aid of transfer learning with a Dice overlap score of 0.86 on a test set of 10 CTA images. Then, bifurcated regions from 3D models were printed into the Pluronic F-127 support bath using alginate+glucomannan hydrogel. We successfully fabricated the bifurcated coronary artery structures with high length and wall thickness accuracy, however, the outer diameters of the vessels and length of the bifurcation point differ from the 3D models. The extrusion of unnecessary material, primarily observed when the nozzle moves from left to the right vessel during 3D printing, can be mitigated by adjusting the nozzle speed. Moreover, the shape accuracy can also be improved by designing a multi-axis printhead that can change the printing angle in three dimensions. Thus, this study demonstrates the potential of the use of AI-segmented 3D models in the 3D printing of coronary artery structures and, when further improved, can be used for the fabrication of patient-specific vascular implants.

PMID:38626778 | DOI:10.1088/1748-605X/ad3f60

Cell Rep Methods. 2024 Apr 10:100759. doi: 10.1016/j.crmeth.2024.100759. Online ahead of print.

ABSTRACT

We designed a Nextflow DSL2-based pipeline, Spatial Transcriptomics Quantification (STQ), for simultaneous processing of 10x Genomics Visium spatial transcriptomics data and a matched hematoxylin and eosin (H&E)-stained whole-slide image (WSI), optimized for patient-derived xenograft (PDX) cancer specimens. Our pipeline enables the classification of sequenced transcripts for deconvolving the mouse and human species and mapping the transcripts to reference transcriptomes. We align the H&E WSI with the spatial layout of the Visium slide and generate imaging and quantitative morphology features for each Visium spot. The pipeline design enables multiple analysis workflows, including single or dual reference genome input and stand-alone image analysis. We show the utility of our pipeline on a dataset from Visium profiling of four melanoma PDX samples. The clustering of Visium spots and clustering of H&E imaging features reveal similar patterns arising from the two data modalities.

PMID:38626768 | DOI:10.1016/j.crmeth.2024.100759

Spectrochim Acta A Mol Biomol Spectrosc. 2024 Apr 4;315:124251. doi: 10.1016/j.saa.2024.124251. Online ahead of print.

ABSTRACT

Uyghur medicine is one of the four major ethnic medicines in China and is a component of traditional Chinese medicine. The intrinsic quality of Uyghur medicinal materials will directly affect the clinical efficacy of Uyghur medicinal preparations. However, in recent years, problems such as adulteration of Uyghur medicinal materials and foreign bodies with the same name still exist, so it is necessary to strengthen the quality control of Uyghur medicines to guarantee Uyghur medicinal efficacy. Identifying the components of Uyghur medicines can clarify the types of medicinal materials used, is a crucial step to realizing the quality control of Uyghur medicines, and is also an important step in screening the effective components of Uyghur medicines. Currently, the method of identifying the components of Uyghur medicines relies on manual detection, which has the problems of high toxicity of the unfolding agent, poor stability, high cost, low efficiency, etc. Therefore, this paper proposes a method based on Raman spectroscopy and multi-label deep learning model to construct a model Mix2Com for accurate identification of Uyghur medicine components. The experiments use computer-simulated mixtures as the dataset, introduce the Long Short-Term Memory Model (LSTM) and Attention mechanism to encode the Raman spectral data, use multiple parallel networks for decoding, and ultimately realize the macro parallel prediction of medicine components. The results show that the model is trained to achieve 90.76% accuracy, 99.41% precision, 95.42% recall value and 97.37% F1 score. Compared to the traditional XGBoost model, the method proposed in the experiment improves the accuracy by 49% and the recall value by 18%; compared with the DeepRaman model, the accuracy is improved by 9% and the recall value is improved by 14%. The method proposed in this paper provides a new solution for the accurate identification of Uyghur medicinal components. It helps to improve the quality standard of Uyghur medicinal materials, advance the research on screening of effective chemical components of Uyghur medicines and their action mechanisms, and then promote the modernization and development of Uyghur medicine.

PMID:38626675 | DOI:10.1016/j.saa.2024.124251

Comput Med Imaging Graph. 2024 Apr 12;115:102380. doi: 10.1016/j.compmedimag.2024.102380. Online ahead of print.

ABSTRACT

The absence of bone wall located in the jugular bulb and sigmoid sinus of the temporal bone is one of the important reasons for pulsatile tinnitus. Automatic and accurate detection of these abnormal singes in CT slices has important theoretical significance and clinical value. Due to the shortage of abnormal samples, imbalanced samples, small inter-class differences, and low interpretability, existing deep-learning methods are greatly challenged. In this paper, we proposed a sub-features orthogonal decoupling model, which can effectively disentangle the representation features into class-specific sub-features and class-independent sub-features in a latent space. The former contains the discriminative information, while, the latter preserves information for image reconstruction. In addition, the proposed method can generate image samples using category conversion by combining the different class-specific sub-features and the class-independent sub-features, achieving corresponding mapping between deep features and images of specific classes. The proposed model improves the interpretability of the deep model and provides image synthesis methods for downstream tasks. The effectiveness of the method was verified in the detection of bone wall absence in the temporal bone jugular bulb and sigmoid sinus.

PMID:38626631 | DOI:10.1016/j.compmedimag.2024.102380

Psychiatry Res. 2024 Apr 6;336:115896. doi: 10.1016/j.psychres.2024.115896. Online ahead of print.

ABSTRACT

Evaluating the rehabilitation status of individuals with serious mental illnesses (SMI) necessitates a comprehensive analysis of multimodal data, including unstructured text records and structured diagnostic data. However, progress in the effective assessment of rehabilitation status remains limited. Our study develops a deep learning model integrating Bidirectional Encoder Representations from Transformers (BERT) and TabNet through a late fusion strategy to enhance rehabilitation prediction, including referral risk, dangerous behaviors, self-awareness, and medication adherence, in patients with SMI. BERT processes unstructured textual data, such as doctor's notes, whereas TabNet manages structured diagnostic information. The model's interpretability function serves to assist healthcare professionals in understanding the model's predictive decisions, improving patient care. Our model exhibited excellent predictive performance for all four tasks, with an accuracy exceeding 0.78 and an area under the curve of 0.70. In addition, a series of tests proved the model's robustness, fairness, and interpretability. This study combines multimodal and multitask learning strategies into a model and applies it to rehabilitation assessment tasks, offering a promising new tool that can be seamlessly integrated with the clinical workflow to support the provision of optimized patient care.

PMID:38626625 | DOI:10.1016/j.psychres.2024.115896

Comput Biol Med. 2024 Apr 9;174:108449. doi: 10.1016/j.compbiomed.2024.108449. Online ahead of print.

ABSTRACT

Spatial transcriptomics (ST), containing gene expression with fine-grained (i.e., different windows) spatial location within tissue samples, has become vital in developing innovative treatments. Traditional ST technology, however, rely on costly specialized commercial equipment. Addressing this, our article aims to creates a cost-effective, virtual ST approach using standard tissue images for gene expression prediction, eliminating the need for expensive equipment. Conventional approaches in this field often overlook the long-distance spatial dependencies between different sample windows or need prior gene expression data. To overcome these limitations, we propose the Edge-Relational Window-Attentional Network (ErwaNet), enhancing gene prediction by capturing both local interactions and global structural information from tissue images, without prior gene expression data. ErwaNet innovatively constructs heterogeneous graphs to model local window interactions and incorporates an attention mechanism for global information analysis. This dual framework not only provides a cost-effective solution for gene expression predictions but also obviates the necessity of prior knowledge gene expression information, a significant advantage in the field of cancer research where it enables a more efficient and accessible analytical paradigm. ErwaNet stands out as a prior-free and easy-to-implement Graph Convolution Network (GCN) method for predicting gene expression from tissue images. Evaluation of the two public breast cancer datasets shows that ErwaNet, without additional information, outperforms the state-of-the-art (SOTA) methods. Code is available at https://github.com/biyecc/ErwaNet.

PMID:38626512 | DOI:10.1016/j.compbiomed.2024.108449

Comput Biol Med. 2024 Apr 9;174:108461. doi: 10.1016/j.compbiomed.2024.108461. Online ahead of print.

ABSTRACT

BACKGROUND: Positron emission tomography (PET) is extensively employed for diagnosing and staging various tumors, including liver cancer, lung cancer, and lymphoma. Accurate subtype classification of tumors plays a crucial role in formulating effective treatment plans for patients. Notably, lymphoma comprises subtypes like diffuse large B-cell lymphoma and Hodgkin's lymphoma, while lung cancer encompasses adenocarcinoma, small cell carcinoma, and squamous cell carcinoma. Similarly, liver cancer consists of subtypes such as cholangiocarcinoma and hepatocellular carcinoma. Consequently, the subtype classification of tumors based on PET images holds immense clinical significance. However, in clinical practice, the number of cases available for each subtype is often limited and imbalanced. Therefore, the primary challenge lies in achieving precise subtype classification using a small dataset.

METHOD: This paper presents a novel approach for tumor subtype classification in small datasets using RA-DL (Radiomics-DeepLearning) attention. To address the limited sample size, Support Vector Machines (SVM) is employed as the classifier for tumor subtypes instead of deep learning methods. Emphasizing the importance of texture information in tumor subtype recognition, radiomics features are extracted from the tumor regions during the feature extraction stage. These features are compressed using an autoencoder to reduce redundancy. In addition to radiomics features, deep features are also extracted from the tumors to leverage the feature extraction capabilities of deep learning. In contrast to existing methods, our proposed approach utilizes the RA-DL-Attention mechanism to guide the deep network in extracting complementary deep features that enhance the expressive capacity of the final features while minimizing redundancy. To address the challenges of limited and imbalanced data, our method avoids using classification labels during deep feature extraction and instead incorporates 2D Region of Interest (ROI) segmentation and image reconstruction as auxiliary tasks. Subsequently, all lesion features of a single patient are aggregated into a feature vector using a multi-instance aggregation layer.

RESULT: Validation experiments were conducted on three PET datasets, specifically the liver cancer dataset, lung cancer dataset, and lymphoma dataset. In the context of lung cancer, our proposed method achieved impressive performance with Area Under Curve (AUC) values of 0.82, 0.84, and 0.83 for the three-classification task. For the binary classification task of lymphoma, our method demonstrated notable results with AUC values of 0.95 and 0.75. Moreover, in the binary classification task of liver tumor, our method exhibited promising performance with AUC values of 0.84 and 0.86.

CONCLUSION: The experimental results clearly indicate that our proposed method outperforms alternative approaches significantly. Through the extraction of complementary radiomics features and deep features, our method achieves a substantial improvement in tumor subtype classification performance using small PET datasets.

PMID:38626509 | DOI:10.1016/j.compbiomed.2024.108461

Parasit Vectors. 2024 Apr 16;17(1):188. doi: 10.1186/s13071-024-06215-7.

ABSTRACT

BACKGROUND: Malaria is a serious public health concern worldwide. Early and accurate diagnosis is essential for controlling the disease's spread and avoiding severe health complications. Manual examination of blood smear samples by skilled technicians is a time-consuming aspect of the conventional malaria diagnosis toolbox. Malaria persists in many parts of the world, emphasising the urgent need for sophisticated and automated diagnostic instruments to expedite the identification of infected cells, thereby facilitating timely treatment and reducing the risk of disease transmission. This study aims to introduce a more lightweight and quicker model-but with improved accuracy-for diagnosing malaria using a YOLOv4 (You Only Look Once v. 4) deep learning object detector.

METHODS: The YOLOv4 model is modified using direct layer pruning and backbone replacement. The primary objective of layer pruning is the removal and individual analysis of residual blocks within the C3, C4 and C5 (C3-C5) Res-block bodies of the backbone architecture's C3-C5 Res-block bodies. The CSP-DarkNet53 backbone is simultaneously replaced for enhanced feature extraction with a shallower ResNet50 network. The performance metrics of the models are compared and analysed.

RESULTS: The modified models outperform the original YOLOv4 model. The YOLOv4-RC3_4 model with residual blocks pruned from the C3 and C4 Res-block body achieves the highest mean accuracy precision (mAP) of 90.70%. This mAP is > 9% higher than that of the original model, saving approximately 22% of the billion floating point operations (B-FLOPS) and 23 MB in size. The findings indicate that the YOLOv4-RC3_4 model also performs better, with an increase of 9.27% in detecting the infected cells upon pruning the redundant layers from the C3 Res-block bodies of the CSP-DarkeNet53 backbone.

CONCLUSIONS: The results of this study highlight the use of the YOLOv4 model for detecting infected red blood cells. Pruning the residual blocks from the Res-block bodies helps to determine which Res-block bodies contribute the most and least, respectively, to the model's performance. Our method has the potential to revolutionise malaria diagnosis and pave the way for novel deep learning-based bioinformatics solutions. Developing an effective and automated process for diagnosing malaria will considerably contribute to global efforts to combat this debilitating disease. We have shown that removing undesirable residual blocks can reduce the size of the model and its computational complexity without compromising its precision.

PMID:38627870 | DOI:10.1186/s13071-024-06215-7

J Cheminform. 2024 Apr 16;16(1):45. doi: 10.1186/s13321-024-00835-y.

ABSTRACT

In this paper we present a method that allows leveraging 3D electron density information to train a deep neural network pipeline to segment regions of high, medium and low electronegativity and classify substances as health hazardous or non-hazardous. We show that this can be used for use-cases such as cosmetics and food products. For this purpose, we first generate 3D electron density cubes using semiempirical molecular calculations for a custom European Chemicals Agency (ECHA) subset consisting of substances labelled as hazardous and non-hazardous for cosmetic usage. Together with their 3-class electronegativity maps we train a modified 3D-UNet with electron density cubes to segment reactive sites in molecules and classify substances with an accuracy of 78.1%. We perform the same process on a custom food dataset (CompFood) consisting of hazardous and non-hazardous substances compiled from European Food Safety Authority (EFSA) OpenFoodTox, Food and Drug Administration (FDA) Generally Recognized as Safe (GRAS) and FooDB datasets to achieve a classification accuracy of 64.1%. Our results show that 3D electron densities and particularly masked electron densities, calculated by taking a product of original electron densities and regions of high and low electronegativity can be used to classify molecules for different use-cases and thus serve not only to guide safe-by-design product development but also aid in regulatory decisions. SCIENTIFIC CONTRIBUTION: We aim to contribute to the diverse 3D molecular representations used for training machine learning algorithms by showing that a deep learning network can be trained on 3D electron density representation of molecules. This approach has previously not been used to train machine learning models and it allows utilization of the true spatial domain of the molecule for prediction of properties such as their suitability for usage in cosmetics and food products and in future, to other molecular properties. The data and code used for training is accessible at https://github.com/s-singh-ivv/eDen-Substances .

PMID:38627862 | DOI:10.1186/s13321-024-00835-y

Ultrasonics. 2024 Apr 9;140:107314. doi: 10.1016/j.ultras.2024.107314. Online ahead of print.

ABSTRACT

The increasing number of passengers and services using railways and the corresponding increase in rail use has caused the acceleration of rail wear and surface defects which makes rail defect identification an important issue for rail maintenance and monitoring to ensure safe and efficient operation. Traditional visual inspection methods for identifying rail defects are time-consuming, less accurate, and associated with human errors. Deep learning has been used to improve railway maintenance and monitoring tasks. This study aims to develop a structured model for detecting railway artifacts and defects by comparing different deep-learning models using ultrasonic image data. This research showed whether it is practical to identify rail indications using image classification and object detection techniques from ultrasonic data and which model performs better among the above-mentioned methods. The methodology includes data processing, labeling, and using different conventional neural networks to develop the model for both image classification and object detection. The results of CNNs for image classification, and YOLOv5 for object detection show 98%, and 99% accuracy respectively. These models can identify rail artifacts efficiently and accurately in real-life scenarios, which can improve automated railway infrastructure monitoring and maintenance.

PMID:38626489 | DOI:10.1016/j.ultras.2024.107314

J Proteome Res. 2024 Apr 16. doi: 10.1021/acs.jproteome.3c00678. Online ahead of print.

ABSTRACT

With the rapid expansion of sequencing of genomes, the functional annotation of proteins becomes a bottleneck in understanding proteomes. The Chromosome-centric Human Proteome Project (C-HPP) aims to identify all proteins encoded by the human genome and find functional annotations for them. However, until now there are still 1137 identified human proteins without functional annotation, called uPE1 proteins. Sequence alignment was insufficient to predict their functions, and the crystal structures of most proteins were unavailable. In this study, we demonstrated a new functional annotation strategy, AlphaFun, based on structural alignment using deep-learning-predicted protein structures. Using this strategy, we functionally annotated 99% of the human proteome, including the uPE1 proteins and missing proteins, which have not been identified yet. The accuracy of the functional annotations was validated using the known-function proteins. The uPE1 proteins shared similar functions to the known-function PE1 proteins and tend to express only in very limited tissues. They are evolutionally young genes and thus should conduct functions only in specific tissues and conditions, limiting their occurrence in commonly studied biological models. Such functional annotations provide hints for functional investigations on the uPE1 proteins. This proteome-wide-scale functional annotation strategy is also applicable to any other species.

PMID:38626392 | DOI:10.1021/acs.jproteome.3c00678

J Am Med Inform Assoc. 2024 Apr 16:ocae080. doi: 10.1093/jamia/ocae080. Online ahead of print.

ABSTRACT

OBJECTIVE: Machine learning (ML) is increasingly employed to diagnose medical conditions, with algorithms trained to assign a single label using a black-box approach. We created an ML approach using deep learning that generates outcomes that are transparent and in line with clinical, diagnostic rules. We demonstrate our approach for autism spectrum disorders (ASD), a neurodevelopmental condition with increasing prevalence.

METHODS: We use unstructured data from the Centers for Disease Control and Prevention (CDC) surveillance records labeled by a CDC-trained clinician with ASD A1-3 and B1-4 criterion labels per sentence and with ASD cases labels per record using Diagnostic and Statistical Manual of Mental Disorders (DSM5) rules. One rule-based and three deep ML algorithms and six ensembles were compared and evaluated using a test set with 6773 sentences (N = 35 cases) set aside in advance. Criterion and case labeling were evaluated for each ML algorithm and ensemble. Case labeling outcomes were compared also with seven traditional tests.

RESULTS: Performance for criterion labeling was highest for the hybrid BiLSTM ML model. The best case labeling was achieved by an ensemble of two BiLSTM ML models using a majority vote. It achieved 100% precision (or PPV), 83% recall (or sensitivity), 100% specificity, 91% accuracy, and 0.91 F-measure. A comparison with existing diagnostic tests shows that our best ensemble was more accurate overall.

CONCLUSIONS: Transparent ML is achievable even with small datasets. By focusing on intermediate steps, deep ML can provide transparent decisions. By leveraging data redundancies, ML errors at the intermediate level have a low impact on final outcomes.

PMID:38626184 | DOI:10.1093/jamia/ocae080

PLoS One. 2024 Apr 16;19(4):e0302275. doi: 10.1371/journal.pone.0302275. eCollection 2024.

ABSTRACT

Although deep-learning methods can achieve human-level performance in boundary detection, their improvements mostly rely on larger models and specific datasets, leading to significant computational power consumption. As a fundamental low-level vision task, a single model with fewer parameters to achieve cross-dataset boundary detection merits further investigation. In this study, a lightweight universal boundary detection method was developed based on convolution and a transformer. The network is called a "transformer with difference convolutional network" (TDCN), which implies the introduction of a difference convolutional network rather than a pure transformer. The TDCN structure consists of three parts: convolution, transformer, and head function. First, a convolution network fused with edge operators is used to extract multiscale difference features. These pixel difference features are then fed to the hierarchical transformer as tokens. Considering the intrinsic characteristics of the boundary detection task, a new boundary-aware self-attention structure was designed in the transformer to provide inductive bias. By incorporating the proposed attention loss function, it introduces the direction of the boundary as strongly supervised information to improve the detection ability of the model. Finally, several head functions with multiscale feature inputs were trained using a bidirectional additive strategy. In the experiments, the proposed method achieved competitive performance on multiple public datasets with fewer model parameters. A single model was obtained to realize universal prediction even for different datasets without retraining, demonstrating the effectiveness of the method. The code is available at https://github.com/neulmc/TDCN.

PMID:38626177 | DOI:10.1371/journal.pone.0302275

Network. 2024 Apr 16:1-33. doi: 10.1080/0954898X.2024.2343341. Online ahead of print.

ABSTRACT

Aiming at early detection and accurate prediction of cardiovascular disease (CVD) to reduce mortality rates, this study focuses on the development of an intelligent predictive system to identify individuals at risk of CVD. The primary objective of the proposed system is to combine deep learning models with advanced data mining techniques to facilitate informed decision-making and precise CVD prediction. This approach involves several essential steps, including the preprocessing of acquired data, optimized feature selection, and disease classification, all aimed at enhancing the effectiveness of the system. The chosen optimal features are fed as input to the disease classification models and into some Machine Learning (ML) algorithms for improved performance in CVD classification. The experiment was simulated in the Python platform and the evaluation metrics such as accuracy, sensitivity, and F1_score were employed to assess the models' performances. The ML models (Extra Trees (ET), Random Forest (RF), AdaBoost, and XG-Boost) classifiers achieved high accuracies of 94.35%, 97.87%, 96.44%, and 99.00%, respectively, on the test set, while the proposed CardioVitalNet (CVN) achieved 87.45% accuracy. These results offer valuable insights into the process of selecting models for medical data analysis, ultimately enhancing the ability to make more accurate diagnoses and predictions.

PMID:38626055 | DOI:10.1080/0954898X.2024.2343341