Image Analysis Research Articles

Discrimination of disposal-restricted materials in waste containers by nondestructive testing and image analysis with high-energy X-ray computed tomography

ABSTRACT Nondestructive methods were investigated to effectively discriminate disposal-restricted materials, including aluminum, batteries, combustibles, lead, and mercury, inside waste containers without opening them. An industrial computed tomography (CT) system with maximum X-ray energy of 9 MeV was used to visualize inside 27-cm diameter pails and 59-cm diameter drums filled with typical waste materials such as combustibles, glass, concrete, and metals. The CT images with 0.5 mm spacing in the height direction were acquired and three-dimensional (3D) models were constracted from the serial tomographic datasets. Generally, a good linear relationship was observed between the gray values in the obtained CT images and the densities of materials. Combustibles, lead, and mercury were extracted from the CT images using simple segmentation by selecting specific gray-value ranges and binarizing the image, due to their lower or higher apparent densities compared to other materials. 3D feature-based discriminations were further applied to batteries and certain aluminum objects based on their structural characteristics. About 97% of the batteries contained in the drums were successfully discriminated regardless of deformation. Almost all of the missing batteries were under strong metal artifacts caused by large lead. Aluminum was extracted with a portion of glass and concrete, while very thin aluminum could not be detected due to its low apparent density. The discrimination methods developed in this study will be effective in revealing the contents of waste containers, particularly for harmful materials that need to be separated for proper disposal.

Computer-assisted scatter plot analysis of cell and nuclear areas distinguishes urothelial carcinoma in urine cytology specimens

Objective: Image analysis in urine cytology typically focuses on individual cells, particularly nuclear features. This study aimed to analyze non-tumor and urothelial carcinoma cases by examining scatter plots of cell or cell cluster areas and the maximum nuclear area within them. Material and Methods: The study included 192 cases: 52 negative and 140 positive. Whole slide images were generated using a virtual slide scanner, and image analysis was conducted with cytological analysis software. Scatter plots were created for cells/cell cluster areas and the largest connected nuclear areas (scatter plot for cells/cell cluster), as well as for nuclear area and perimeter (scatter plot for nucleus). Results: In the scatter plot for the nucleus, significant differences were noted between cytology-negative and cytology-positive groups (P = 0.0134). However, when divided into cytology-negative, non-invasive, and invasive groups, a significant difference was only found between negative and non-invasive groups (P = 0.0281), not between negative and invasive groups (P = 0.1266). In the scatter plot for cell/cell cluster, plotting cell cluster area (X-axis) and maximum nuclear area (Y-axis) revealed three distribution patterns: horizontal (X-axis), vertical (Y-axis), and diagonal. Cytology-negative cases mainly showed horizontal patterns, while cytology-positive cases exhibited vertical patterns. In the non-tumor group, horizontal patterns were dominant, while vertical patterns were common in non-invasive and invasive tumor groups. The pTa low-grade group mainly showed diagonal patterns, whereas the pTa high-grade, pTis, and pTis + pTa groups predominantly showed vertical patterns. The percentage of cell/cell clusters in tumor-rich areas (along with Y-axis) was significantly higher in non-invasive and invasive tumors compared to non-tumor cases (P < 0.0001), although lower in invasive tumors compared to non-invasive ones (P = 0.0299). In addition, neutrophil-rich images were significantly more common in stromal and muscle invasion groups than in non-invasion groups. Conclusion: In urine cytology, cellular overlap and cluster density were key factors for distinguishing malignant from benign cells. This image analysis algorithm was useful in identifying malignant clusters with large, connected nuclear regions. The algorithm could potentially detect both invasive and early-stage tumors, highlighting the need for further development of such tools for routine diagnosis.

Polarity-JaM: an image analysis toolbox for cell polarity, junction and morphology quantification

Cell polarity involves the asymmetric distribution of cellular components such as signalling molecules and organelles within a cell, alterations in cell morphology and cell-cell contacts. Advances in fluorescence microscopy and deep learning algorithms open up a wealth of unprecedented opportunities to characterise various aspects of cell polarity, but also create new challenges for comprehensible and interpretable image data analysis workflows to fully exploit these new opportunities. Here we present Polarity-JaM, an open source package for reproducible exploratory image analysis that provides versatile methods for single cell segmentation, feature extraction and statistical analysis. We demonstrate our analysis using fluorescence image data of endothelial cells and their collective behaviour, which has been shown to be essential for vascular development and disease. The general architecture of the software allows its application to other cell types and imaging modalities, as well as seamless integration into common image analysis workflows, see https://polarityjam.readthedocs.io. We also provide a web application for circular statistics and data visualisation, available at www.polarityjam.com, and a Napari plug-in, each with a graphical user interface to facilitate exploratory analysis. We propose a holistic image analysis workflow that is accessible to the end user in bench science, enabling comprehensive analysis of image data.

ANO7 expression in the prostate modulates mitochondrial function and lipid metabolism

BackgroundProstate cancer (PrCa) is a significant health concern, ranking as the second most common cancer in males globally. Genetic factors contribute substantially to PrCa risk, with up to 57% of the risk being attributed to genetic determinants. A major challenge in managing PrCa is the early identification of aggressive cases for targeted treatment, while avoiding unnecessary interventions in slow-progressing cases. Therefore, there is a critical need for genetic biomarkers that can distinguish between aggressive and non-aggressive PrCa cases. Previous research, including our own, has shown that germline variants in ANO7 are associated with aggressive PrCa. However, the function of ANO7 in the prostate remains unknown.MethodsWe performed RNA-sequencing (RNA-seq) on RWPE1 cells engineered to express ANO7 protein, alongside the analysis of a single-cell RNA-sequencing (scRNA-seq) dataset and RNA-seq from prostate tissues. Differential gene expression analysis and gene set enrichment analysis (GSEA) were conducted to identify key pathways. Additionally, we assessed oxidative phosphorylation (OXPHOS), glycolysis, and targeted metabolomics. Image analysis of mitochondrial morphology and lipidomics were also performed to provide further insight into the functional role of ANO7 in prostate cells.ResultsANO7 expression resulted in the downregulation of metabolic pathways, particularly genes associated with the MYC pathway and oxidative phosphorylation (OXPHOS) in both prostate tissue and ANO7-expressing cells. Measurements of OXPHOS and glycolysis in the ANO7-expressing cells revealed a metabolic shift towards glycolysis. Targeted metabolomics showed reduced levels of the amino acid aspartate, indicating disrupted mitochondrial function in the ANO7-expressing cells. Image analysis demonstrated altered mitochondrial morphology in these cells. Additionally, ANO7 downregulated genes involved in fatty acid metabolism and induced changes in lipid composition of the cells, characterized by longer acyl chain lengths and increased unsaturation, suggesting a role for ANO7 in regulating lipid metabolism in the prostate.ConclusionsThis study provides new insights into the function of ANO7 in prostate cells, highlighting its involvement in metabolic pathways, particularly OXPHOS and lipid metabolism. The findings suggest that ANO7 may act as a key regulator of cellular lipid metabolism and mitochondrial function in the prostate, shedding light on a previously unknown aspect of ANO7’s biology.

Single-molecule magnetic tweezers to unravel protein folding dynamics under force

Abstract Single-molecule magnetic tweezers have recently emerged as a powerful technique for measuring the equilibrium dynamics of individual proteins under force. In magnetic tweezers, a single protein is tethered between a glass coverslip and a superparamagnetic bead, and by applying and controlling a magnetic field, the protein is mechanically stretched while force-induced conformational changes are measured by tracking the vertical position of the bead. The soft trap created by the magnetic field provides intrinsic force-clamp conditions, which makes magnetic tweezers particularly well-suited to measure protein conformational dynamics. Traditionally employed to study DNA due to their initially low spatial and temporal resolutions, magnetic tweezers instrumentation has experienced significant progress in recent years. The development of high-speed cameras, stronger illumination sources, advanced image analysis algorithms, and dedicated chemical functionalization strategies, now allow for high-resolution and ultra-stable experiments. Together with their ability to apply and control low forces, magnetic tweezers can capture long-term equilibrium protein folding dynamics, not possible with any other technique. These capabilities have proven particularly valuable in the study of force-sensing protein systems, which often exhibit low mechanical stabilities that are challenging to measure with other techniques. In this review, we will discuss the current status of magnetic tweezers instrumentation for studying protein folding dynamics, focusing on both the instrumental aspects and methodologies to interpret nanomechanical experiments.

Multimodal Image Analysis Based PedestrianDetection Using Optimization with Classification byHybrid Machine Learning Model

Multimodal Image Analysis Based PedestrianDetection Using Optimization with Classification byHybrid Machine Learning Model

Multi-task AI models in dermatology: Overcoming critical clinical translation challenges for enhanced skin lesion diagnosis.

The surge in AI models for diagnosing skin lesions through image analysis is notable, yet their clinical implementation faces challenges. Common limitations include an over reliance on dermoscopy, lack of real-world applicability when only binary output (e.g. benign/malignant) is offered and low accuracy when faced with rare skin conditions. To address these common constraints associated with limited diagnostic output, and applicability to real-world settings. We developed an All-In-One Hierarchical-Out of Distribution-Clinical Triage (HOT) AI model for skin lesion analysis. Trained on a large dataset of ~208,000 lesion images, our HOT AI model generates three outputs: a hierarchical three-level prediction, an alert for out-of-distribution (OOD) images and a recommendation for dermoscopy to improve diagnostic prediction. Our hierarchical prediction output provides a binary level 1 prediction (benign/malignant), Level 2 prediction of eight possible categories (e.g. melanocytic and keratinocytic) and a more definitive Level 3 prediction from 44 lesion categories. The model produced high sensitivity for Level 1 prediction (88.14% CI: 87.42-88.51); however, significantly lower for Level 3 prediction (63.90%, CI: 62.27-65.61). By relying on all three prediction levels for consensus, Level 1 false-positives were reduced by 20-25%, and false-negatives were decreased by 11-13% of cases. OOD detection was benchmarked against previous landmark models and outperformed comparative models. Lastly, 44% of images were recommended for dermoscopy, and with additional image input, Level 3 sensitivity increased from 48.13% (CI:45.08-49.57) to 52.54% (CI:50.25-55.04). Our HOT-AI model attempts to address common challenges in existing models by combining three tasks in one model to increase accuracy and clinical utility. By providing a more nuanced prediction, and alert for OOD, the model output provides greater explainability of the AI decision process. Prospective clinical testing is required to measure how this additional output impacts user trust, and how the model performs in a real-world setting.

Image Analysis as tool for Predicting Colorectal Cancer Molecular Alterations: A Scoping Review.

Among the most important diagnostic indicators of colorectal cancer; however, measuring molecular alterations are invasive and expensive. This study aimed to investigate the application of image processing to predict molecular alterations in colorectal cancer. In this scoping review, we searched for relevant literature by searching the Web of Science, Scopus, and PubMed databases. The method of selecting the articles and reporting the findings was according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses; moreover, the Strengthening the Reporting of Observational Studies in Epidemiology checklist was used to assess the quality of the studies. Sixty seven out of 2,223 articles, 67 were relevant to the aim of the study, and finally 41 studies with sufficient quality were reviewed. The prediction of Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS), Neuroblastoma RAS Viral (NRAS), B-Raf proto-oncogene, serine/threonine kinase (BRAF), Tumor Protein 53 (TP53), Adenomatous Polyposis Coli, and microsatellite instability (MSI) with the help of image analysis has received more attention than other molecular characteristics. The studies used computed tomography (CT), magnetic resonance imaging (MRI), and 18F-FDG positron emission tomography (PET)/CT with radionics and quantitative analysis to predict molecular alterations in colorectal cancer, analyzing features like texture, maximum standard uptake value, and MTV using various statistical methods. In 39 studies, there was a significant relationship between the features extracted from these images and molecular alterations. Different modalities were used to measure the area under the receiver operating characteristic curve for predicting the alterations in KRAS, MSI, BRAF, and TP53, with an average of 78, 81, 80 and 71%, respectively. This scoping review underscores the potential of radiogenomics in predicting molecular alterations in colorectal cancer through non-invasive imaging modalities, like CT, MRI, and 18F-FDG PET/CT. The analysis of 41 studies showed the appropriate prediction of key alterations, such as KRAS, NRAS, BRAF, TP53, and MSI, highlighting the promise of radionics and texture features in enhancing predictive accuracy.

A Personalized Muscle–Tendon Assessment and Exercise Prescription Concept Reduces Muscle–Tendon Imbalances in Female Adolescent Athletes

BackgroundImbalances between muscle strength and tendon stiffness influence the mechanical demand on the tendon (i.e., tendon strain) and may increase tendon injury risk. The purpose of this study was to identify muscle–tendon imbalances and to promote a more balanced musculotendinous adaptation through a personalized assessment and exercise prescription concept in female adolescent handball athletes (13–16 years).MethodsAt four measurement time points during a competitive season, we used dynamometry and ultrasonography to assess knee extensor muscle strength, patellar tendon stiffness and strain. Tendon micromorphology was assessed with a peak spatial frequency (PSF) analysis of proximal tendon ultrasound images. Muscle–tendon imbalances were identified based on tendon strain during maximum voluntary contractions. A control group (n = 15) followed their usual training. In the intervention group, athletes with a deficit in tendon stiffness (strain ≥ 9%; n = 6) or no muscle–tendon imbalances (strain between 4.5% and 9%; n = 15) performed exercises (3x/week for 32 weeks) with a personalized load to reach ~ 6.2% tendon strain to predominantly promote tendon or both muscle and tendon adaptation. Individuals with a muscle strength deficit (strain ≤ 4.5%; n = 1) trained with submaximal loads to failure to promote muscle strength.ResultsIn the intervention group we found lower fluctuations of maximum tendon strain (p = 0.005) and a decrease in tendon strain over time (p = 0.010), which was more pronounced in individuals with initially high tendon strain. While there were no systematic changes in muscle strength or tendon stiffness at the group level (p > 0.05), the marked decrease in tendon strain in individuals with a deficit in tendon stiffness was caused by a predominant increase in tendon stiffness. Overall, the prevalence of muscle–tendon imbalances was reduced in the intervention group, while it temporarily increased in the control group. PSF did not differ between groups at baseline but decreased significantly in the intervention group (p = 0.013).ConclusionsThe findings suggest that the personalized concept is suitable to promote a more uniform adaptation of knee extensor muscle strength and patellar tendon stiffness and to reduce the prevalence of musculotendinous imbalances in female adolescent athletes, which may have important implications for tendon injury prevention.Trial RegistrationDRKS, DRKS00035110. Registered 20 September 2024–retrospectively registered, https://drks.de/search/de/trial/DRKS00035110.

A novel deep learning framework for retinal disease detection leveraging contextual and local features cues from retinal images.

Retinal diseases are a serious global threat to human vision, and early identification is essential for effective prevention and treatment. However, current diagnostic methods rely on manual analysis of fundus images, which heavily depends on the expertise of ophthalmologists. This manual process is time-consuming and labor-intensive and can sometimes lead to missed diagnoses. With advancements in computer vision technology, several automated models have been proposed to improve diagnostic accuracy for retinal diseases and medical imaging in general. However, these methods face challenges in accurately detecting specific diseases within images due to inherent issues associated with fundus images, including inter-class similarities, intra-class variations, limited local information, insufficient contextual understanding, and class imbalances within datasets. To address these challenges, we propose a novel deep learning framework for accurate retinal disease classification. This framework is designed to achieve high accuracy in identifying various retinal diseases while overcoming inherent challenges associated with fundus images. Generally, the framework consists of three main modules. The first module is Densely Connected Multidilated Convolution Neural Network (DCM-CNN) that extracts global contextual information by effectively integrating novel Casual Dilated Dense Convolutional Blocks (CDDCBs). The second module of the framework, namely, Local-Patch-based Convolution Neural Network (LP-CNN), utilizes Class Activation Map (CAM) (obtained from DCM-CNN) to extract local and fine-grained information. To identify the correct class and minimize the error, a synergic network is utilized that takes the feature maps of both DCM-CNN and LP-CNN and connects both maps in a fully connected fashion to identify the correct class and minimize the errors. The framework is evaluated through a comprehensive set of experiments, both quantitatively and qualitatively, using two publicly available benchmark datasets: RFMiD and ODIR-5K. Our experimental results demonstrate the effectiveness of the proposed framework and achieves higher performance on RFMiD and ODIR-5K datasets compared to reference methods.

Anatomical variations of the intracranial arteries and their association with intracranial aneurysms: Insights from digital subtraction angiographies.

The purpose of this study was to identify the anatomical variants (AVs) in the intracranial arterial circulation of patients who underwent neuro-interventional procedures (NIPs) and to describe their relationship with intracranial aneurysms (IA). We performed a cross-sectional analysis of angiographic images from patients who underwent NIP at the Interventional Neuroradiology Department of the National Institute of Neurology and Neurosurgery in México between July 1, 2020, and January 1, 2022. After reviewing images from 150 NIPs, we found 144 AVs., yielding a prevalence of 81%. Of these, 49 AVs (34%) were located in the anterior circulation (AC) and 95 (66%) in the posterior circulation (PC); 23 NIPs (16%) showed AVs in both the AC and PC. The most frequent AVs were the fetal pattern of the posterior cerebral artery (19%) and hypoplasia of the A1 segment (12%). AVs were significantly more common in patients with neurovascular disorders than those without (80% vs 49%, p = .003) and in patients with IA compared to those without (68% vs 47%, p = .048). In the studied population, AVs were predominantly located in the PC and are significantly more frequent in patients with neurovascular disorders, particularly those with IA.

BenthicNet: A global compilation of seafloor images for deep learning applications

Advances in underwater imaging enable collection of extensive seafloor image datasets necessary for monitoring important benthic ecosystems. The ability to collect seafloor imagery has outpaced our capacity to analyze it, hindering mobilization of this crucial environmental information. Machine learning approaches provide opportunities to increase the efficiency with which seafloor imagery is analyzed, yet large and consistent datasets to support development of such approaches are scarce. Here we present BenthicNet: a global compilation of seafloor imagery designed to support the training and evaluation of large-scale image recognition models. An initial set of over 11.4 million images was collected and curated to represent a diversity of seafloor environments using a representative subset of 1.3 million images. These are accompanied by 3.1 million annotations translated to the CATAMI scheme, which span 190,000 of the images. A large deep learning model was trained on this compilation and preliminary results suggest it has utility for automating large and small-scale image analysis tasks. The compilation and model are made openly available for reuse.

Comprehensive characterization of micropapillary colorectal adenocarcinoma.

Micropapillary colorectal adenocarcinoma is a morphologic subtype of colorectal cancer (CRC) with insufficiently characterized prognostic significance and biological features. We analyzed the histopathological, immunological, and prognostic features of micropapillary adenocarcinoma in two independent CRC cohorts (N = 1,876). We found that micropapillary adenocarcinomas accounted for 4.9% and 6.4% of CRCs in the two cohorts. A micropapillary growth pattern was associated with advanced stage and lymphovascular invasion (p < 0.001), but also with shorter overall survival independent of these factors and other prognostic parameters (Cohort 1: hazard ratio [HR] 1.76, 95% confidence interval [CI] 1.08-2.87; Cohort 2: HR 1.47, 95% CI 1.08-2.00). Multiplex immunohistochemistry and machine learning-assisted image analysis showed that the micropapillary growth pattern was associated with decreased CD3+ T-cell and CD14+HLA-DR+ monocytic cell densities. Molecular features of micropapillary adenocarcinoma were studied using bioinformatic analyses in The Cancer Genome Atlas (TCGA) cohort (N = 629) and validated with optical genome mapping and immunohistochemistry. These analyses revealed that micropapillary adenocarcinomas frequently present with chromosome region 8q24 copy number gain, TP53 mutation, and overexpression of UPK2, MUC16, and epithelial-mesenchymal transition involved genes, such as L1CAM. These results indicate that micropapillary colorectal adenocarcinoma is an aggressive morphologic subtype of CRC characterized by shorter overall survival, decreased antitumorigenic immune response, and unique molecular features. Our findings support the classification of micropapillary adenocarcinoma as a distinct, high-risk subtype of CRC, which should be systematically evaluated in patient care. © 2025 The Author(s). The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Leveraging Vision Transformers for Early Detection of Alzheimer's disease Through MRI Image Analysis

Alzheimer's disease (AD) poses challenges for early recognition due to subtle and overlapping symptoms, making timely care and intervention difficult. This work presents a new method for identifying Alzheimer's disease by combining magnetic resonance imaging (MRI) data with a Vision Transformer (ViT) model. In this research, we develop a method that leverages MRI scans to classify Alzheimer's disease (AD) into four distinct stages: Non-Demented, Very Mild Demented, Mild Demented, and Moderate Demented. To improve the identification of clinical and structural alterations in the brain linked to AD, sophisticated characteristics are retrieved, including voxel-intensity patterns, grey matter volume, and cortical thickness. These characteristics allow for a more accurate representation of MRI data when paired with sophisticated processing methods. Additionally, even in situations where data is scarce, transfer learning is used to maximize the Vision Transformer (ViT) model's classification accuracy, especially when it comes to differentiating between neighboring stages of dementia. The ViT model, a cutting-edge deep learning technique, shows strong performance and dependability in identifying AD phases. Our findings highlight its effectiveness, showcasing a significant improvement in diagnostic accuracy compared to existing methods. The accuracy with which the ViT model can distinguish between different phases of AD emphasizes both its suitability for handling complicated MRI data and its promise for early diagnosis and detection.This paper offers a promising method for enhancing MRI image processing for Alzheimer's disease diagnosis through sophisticated feature extraction, transfer learning, and the use of Vision Transformers (ViTs). Present authors believe this method successfully tackles the difficulties of early and precise detection, making it essential for appropriate medical image analysis.

Advancing Optical Coherence Tomography Diagnostic Capabilities: Machine Learning Approaches to Detect Autoimmune Inflammatory Diseases.

In many parts of the world including India, the prevalence of autoimmune inflammatory diseases such as neuromyelitis optica spectrum disorders (NMOSD), myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), and multiple sclerosis (MS) is rising. A diagnosis is often delayed due to insufficient diagnostic tools. Machine learning (ML) models have accurately differentiated eyes of patients with MS from those of healthy controls (HCs) using optical coherence tomography (OCT)-based retinal images. Examining OCT characteristics may allow for early differentiation of these conditions. The objective of this study was to determine feasibility of ML analyses to distinguish between patients with different autoimmune inflammatory diseases, other ocular diseases, and HCs based on OCT measurements of the peripapillary retinal nerve fiber layer (pRNFL), ganglion cell-inner plexiform layer (GCIPL), and inner nuclear layers (INLs). Eyes of people with MS (n = 99 patients), NMOSD (n = 40), MOGAD (n = 74), other ocular diseases (OTHER, n = 16), and HCs (n = 54) from the Mangalore Demyelinating Disease Registry were included. Support vector machine (SVM) classification models incorporating age, pRNFL, GCIPL, and INL were performed. Data were split into training (70%) and testing (30%) data and accounted for within-patient correlations. Cross-validation was used in training to choose the best parameters for the SVM model. Accuracy and area under receiver operating characteristic curves (AUROCs) were used to assess model performance. The SVM models distinguished between eyes of patients with each condition (i.e., MOGAD vs NMOSD, NMOSD vs HC, MS vs OTHER, etc) with strong discriminatory power demonstrated from the AUROCs for these comparisons ranging from 0.81 to 1.00. These models also performed with moderate to high accuracy, ranging from 0.66 to 0.81, with the exception of the MS vs NMOSD comparison, which had an accuracy of 0.53. ML models are useful for distinguishing between autoimmune inflammatory diseases and for distinguishing these from HCs and other ocular diseases based on OCT measures. This study lays the groundwork for future deep learning studies that use analyses of raw OCT images for identifying eyes of patients with such disorders and other etiologies of optic neuropathy.

Diagnosis of Sacroiliitis Through Semi-Supervised Segmentation and Radiomics Feature Analysis of MRI Images.

Sacroiliitis is a hallmark of ankylosing spondylitis (AS), and early detection plays an important role in managing the condition effectively. MRI is commonly used for diagnosing sacroiliitis, traditional methods often depend on subjective interpretation or limited automation which can introduce variability in diagnoses. The integration of semi-supervised segmentation and radiomics features may reduce reliance on expert interpretation and the need for large annotated datasets, potentially enhancing diagnostic workflows. To develop a diagnostic model for sacroiliitis and bone marrow edema (BME) using semi-supervised segmentation and radiomics analysis of MRI images. Retrospective cohort study. A total of 257 patients (161 males, 93 females; age 11-74 years), including 155 sacroiliitis and 175 BME patients. A total of 514 sacroiliac joint (SIJ) MRI images are analyzed, with 359 used for training and 155 for testing. 3.0 T, spin echo T1-weighted imaging (T1WI) and short-tau inversion recovery (STIR). SIJ segmentation is automated using the semi-supervised segmentation-based Unimatch framework. Manual delineation of SIJ regions of interest (ROIs) on T1WI images by an experienced radiologist (W.M., 10-year experience) served as the reference standard for segmentation performance evaluation. Radiomics features from T1WI and STIR are used to train machine learning models, including support vector machine (SVM), logistic regression (LR), and light gradient boosting machine (LightGBM), for sacroiliitis and BME detection. Performance is assessed using area under the curve (AUC), sensitivity, specificity, and accuracy. The Dice coefficient is used to assess the performance of the semi-supervised segmentation model on SIJ segmentation. Performance is evaluated using receiver operating characteristic (ROC) curves and decision curve analysis (DCA). The Unimatch model achieves an average Dice coefficient of 0.859 for SIJ segmentation. AUCs for sacroiliitis detection are 0.84 (LR), 0.86 (SVM), and 0.78 (LightGBM), while for BME detection, AUCs are 0.73 (LR), 0.76 (SVM), and 0.70 (LightGBM). This study demonstrates that semi-supervised segmentation combined with radiomics features and machine learning models provides a promising approach for diagnosis of sacroiliitis and BME. This study aimed to improve the diagnosis of sacroiliitis and bone marrow edema in patients with ankylosing spondylitis. The researchers used a method that automatically segments MRI images and analyzes features from those images. By applying machine learning, they created models to help detect sacroiliitis and bone marrow edema more accurately. The results show that this approach can effectively assist in identifying these conditions, with the best accuracy for sacroiliitis and bone marrow edema reaching 81.2% and 74.2%, respectively. This method could help doctors make better decisions, offering a promising tool for improving diagnosis in clinical settings. 3 TECHNICAL EFFICACY: Stage 2.

Instance-level semantic segmentation of nuclei based on multimodal structure encoding

BackgroundAccurate segmentation and classification of cell nuclei are crucial for histopathological image analysis. However, existing deep neural network-based methods often struggle to capture complex morphological features and global spatial distributions of cell nuclei due to their reliance on local receptive fields.MethodsThis study proposes a graph neural structure encoding framework based on a vision-language model. The framework incorporates: (1) A multi-scale feature fusion and knowledge distillation module utilizing the Contrastive Language-Image Pre-training (CLIP) model’s image encoder; (2) A method to transform morphological features of cells into textual descriptions for semantic representation; and (3) A graph neural network approach to learn spatial relationships and contextual information between cell nuclei.ResultsExperimental results demonstrate that the proposed method significantly improves the accuracy of cell nucleus segmentation and classification compared to existing approaches. The framework effectively captures complex nuclear structures and global distribution features, leading to enhanced performance in histopathological image analysis.ConclusionsBy deeply mining the morphological features of cell nuclei and their spatial topological relationships, our graph neural structure encoding framework achieves high-precision nuclear segmentation and classification. This approach shows significant potential for enhancing histopathological image analysis, potentially leading to more accurate diagnoses and improved understanding of cellular structures in pathological tissues.

Digital Humanities for Increasing Disaster Resilience in Art Nouveau and Modernist Buildings

The paper will focus on the topic of adapting digital humanities methods from architectural history to technical history, considering mapping and image analysis for increasing disaster resilience in Art Nouveau and Modernist buildings in different geographical areas—including lessons from Europe to the USA. The project proposes the transformation of the collection of photographs of early 20th-century architecture gathered by the applicant over about 30 years of travel into a database by answering the research question on how threats from the hazards of earthquakes, floods, and fires can be answered by taking into account the local culture in the European countries covered, for buildings from a period when the architecture styles were already global at that time. For this purpose, digital humanities methods of image annotation (including architectural volumetric analysis) and mapping are employed. From the knowledge gathered and the resulting database, a prototyping ontology and taxonomy is derived. This outcome can be further developed into a set of evaluation criteria, considering the decisions that can be taken to prioritize the retrofit interventions depending on the geographic positions of the buildings.

LeaData a novel reference data of leather images for automatic species identification

In the leather industry, the mammalian skins of buffalo, cow, goat, and sheep are the permissible materials for leather-making. They serve the trade of quality leather products; hence, the knowledge of animal species in leather is inevitable. The traditional identification techniques are prone to ambiguous predictions due to insufficient reference studies. Indeed, leather image analysis with big data can pave the way for automatic and objective analysis with accurate prediction. This study focuses on creating novel and unique leather image data, LeaData. The objective is to automatically determine species from grain surface analysis. Hence, it employs a simple, cheaper, handheld digital microscope for leather image acquisition. The magnifying parameter 47 captures the species-unique grain patterns distributed over the leather surface. In total, the LeaData encloses 38,172 images of four species from 137 leather samples. This big data spans leather images with theoretically ideal and practically non-ideal grain patterns. It also includes images of grain patterns varying over different body parts. Thus, the novel LeaData is an adequately larger pool of leather images with diverse behavior. The motive is to establish a smart leather species identification technique that can be easily accessible by leather specialists, customs officials, and leather product manufacturers. Hence, this paper solely creates the bigger LeaData and presents its different versions to the digital image processing and computer vision research community. This digitized source of permissible leather species helps enable digitization in leather technology for species identification. In turn, in maintaining biodiversity preservation and consumer protection.

Deep learning–driven bacterial cytological profiling to determine antimicrobial mechanisms in Mycobacterium tuberculosis

Tuberculosis (TB), caused by Mycobacterium tuberculosis, remains a significant global health threat, affecting an estimated 10.6 million people in 2022. The emergence of multidrug resistant and extensively drug resistant strains necessitates the development of novel and effective drugs. Accelerating the determination of mechanisms of action (MOAs) for these drugs is crucial for advancing TB treatment. This study introduces MycoBCP, a unique adaptation of bacterial cytological profiling (BCP) tailored to M. tuberculosis, utilizing the application of convolutional neural networks (CNNs) within BCP to overcome challenges posed by traditional image analysis techniques. Using MycoBCP, we analyzed the morphological effects of various antimicrobial compounds on M. tuberculosis, capturing broad patterns rather than relying on precise cell segmentation. This approach circumvented issues such as cell clumping and uneven staining, which are prevalent in M. tuberculosis. In a blind test, MycoBCP accurately identified the MOA for 96% of the compounds, with a single misclassification of rifabutin, which was incorrectly categorized as affecting translation rather than transcription. The similar morphologies resulting from transcription and translation inhibition indicate a need for further refinement to distinguish them more effectively. Application of MycoBCP to a series of antitubercular agents successfully identified known MOAs and revealed unique effects, demonstrating its utility in early drug discovery and development. Our findings underscore the potential of CNN-based BCP to enhance the accuracy and efficiency of MOA determination, particularly for challenging pathogens like M. tuberculosis. MycoBCP represents a significant advancement in TB drug development, offering a robust and adaptable method for high-throughput screening of antimicrobial compounds.