Sci Rep. 2024 Jul 31;14(1):17633. doi: 10.1038/s41598-024-68489-2.

ABSTRACT

Several studies published so far used highly selective image datasets from unclear sources to train computer vision models and that may lead to overestimated results, while those studies conducted in real-life remain scarce. To avoid image selection bias, we stacked convolutional and recurrent neural networks (CNN-RNN) to analyze complete optical coherence tomography (OCT) cubes in a row and predict diabetic macular edema (DME), in a real-world diabetic retinopathy screening program. A retrospective cohort study was carried out. Throughout 4-years, 5314 OCT cubes from 4408 subjects who attended to the diabetic retinopathy (DR) screening program were included. We arranged twenty-two (22) pre-trained CNNs in parallel with a bidirectional RNN layer stacked at the bottom, allowing the model to make a prediction for the whole OCT cube. The staff of retina experts built a ground truth of DME later used to train a set of these CNN-RNN models with different configurations. For each trained CNN-RNN model, we performed threshold tuning to find the optimal cut-off point for binary classification of DME. Finally, the best models were selected according to sensitivity, specificity, and area under the receiver operating characteristics curve (AUROC) with their 95% confidence intervals (95%CI). An ensemble of the best models was also explored. 5188 cubes were non-DME and 126 were DME. Three models achieved an AUROC of 0.94. Among these, sensitivity, and specificity (95%CI) ranged from 84.1-90.5 and 89.7-93.3, respectively, at threshold 1, from 89.7-92.1 and 80-83.1 at threshold 2, and from 80.2-81 and 93.8-97, at threshold 3. The ensemble model improved these results, and lower specificity was observed among subjects with sight-threatening DR. Analysis by age, gender, or grade of DME did not vary the performance of the models. CNN-RNN models showed high diagnostic accuracy for detecting DME in a real-world setting. This engine allowed us to detect extra-foveal DMEs commonly overlooked in other studies, and showed potential for application as the first filter of non-referable patients in an outpatient center within a population-based DR screening program, otherwise ended up in specialized care.

PMID:39085461 | DOI:10.1038/s41598-024-68489-2

Sci Rep. 2024 Jul 31;14(1):17635. doi: 10.1038/s41598-024-68705-z.

ABSTRACT

Our objective was to develop and evaluate the clinical feasibility of deep-learning-based synthetic contrast-enhanced computed tomography (DL-SynCCT) in patients designated for nonenhanced CT (NECT). We proposed a weakly supervised learning with the utilization of virtual non-contrast CT (VNC) for the development of DL-SynCCT. Training and internal validations were performed with 2202 pairs of retrospectively collected contrast-enhanced CT (CECT) images with the corresponding VNC images acquired from dual-energy CT. Clinical validation was performed using an external validation set including 398 patients designated for true nonenhanced CT (NECT), from multiple vendors at three institutes. Detection of lesions was performed by three radiologists with only NECT in the first session and an additionally provided DL-SynCCT in the second session. The mean peak signal-to-noise ratio (PSNR) and structural similarity index map (SSIM) of the DL-SynCCT compared to CECT were 43.25 ± 0.41 and 0.92 ± 0.01, respectively. With DL-SynCCT, the pooled sensitivity for lesion detection (72.0% to 76.4%, P < 0.001) and level of diagnostic confidence (3.0 to 3.6, P < 0.001) significantly increased. In conclusion, DL-SynCCT generated by weakly supervised learning showed significant benefit in terms of sensitivity in detecting abnormal findings when added to NECT in patients designated for nonenhanced CT scans.

PMID:39085456 | DOI:10.1038/s41598-024-68705-z

Phys Med Biol. 2024 Jul 31. doi: 10.1088/1361-6560/ad69fb. Online ahead of print.

ABSTRACT

Normalization in positron emission tomography (PET) corrects for non-uniformity of sensitivity across all system lines of response (LOR). Self-normalization is a framework that aims to estimate normalization components from the emission data without a separate scan of a normalization phantom. In this work, we propose for the first time an image-based end-to-end self-normalization framework using conditional generative adversarial networks (cGAN). We evaluated different approaches by exploring each of the following three methodologies. First, we used images that were either unnormalized or corrected for geometric factors, which encompass all time-invariant factors, as input data types. Second, we set the input tensor shape as either a single axial slice (2-D) or three contiguous axial slices (2.5-D). Third, we chose either Pix2Pix or polarized self-attention (PSA) Pix2Pix, which we developed for this work, as a deep learning network. The targets for all approaches were the axial slices of images normalized using the direct normalization method. We performed Monte Carlo simulations of ten voxelized phantoms with the SimSET simulation tool and produced 26,000 pairs of axial image slices for training and testing. The results showed that 2.5-D PSA Pix2Pix trained with geometric-factors-corrected input images achieved the best performance among all the methods we tested. All approaches improved general image quality figures of merit peak signal to noise ratio (PSNR) and structural similarity index (SSIM) from ~15% to ~55%, and 2.5-D PSA Pix2Pix showed the highest PSNR (28.074) and SSIM (0.921). Lesion detectability, measured with region of interest (ROI) PSNR, SSIM, normalized contrast recovery coefficient (NCRC), and contrast-to-noise ratio (CNR), was generally improved for all approaches, and 2.5-D PSA Pix2Pix trained with geometric-factors-corrected input images achieved the highest ROI PSNR (28.920) and SSIM (0.973).

PMID:39084640 | DOI:10.1088/1361-6560/ad69fb

Am J Obstet Gynecol. 2024 Jul 29:S0002-9378(24)00790-7. doi: 10.1016/j.ajog.2024.07.031. Online ahead of print.

NO ABSTRACT

PMID:39084497 | DOI:10.1016/j.ajog.2024.07.031

Brain Res. 2024 Jul 29:149133. doi: 10.1016/j.brainres.2024.149133. Online ahead of print.

ABSTRACT

The progress in neuroimaging and electrophysiological techniques has shown substantial promise in improving the clinical assessment of disorders of consciousness (DOC). Through the examination of both stimulus-induced and spontaneous brain activity, numerous comprehensive investigations have explored variations in brain activity patterns among patients with DOC, yielding valuable insights for clinical diagnosis and prognostic purposes. Nonetheless, reaching a consensus on precise neuroimaging biomarkers for patients with DOC remains a challenge. Therefore, in this review, we begin by summarizing the empirical evidence related to neuroimaging biomarkers for DOC using various paradigms, including active, passive, and resting-state approaches, by employing task-based fMRI, resting-state fMRI (rs-fMRI), electroencephalography (EEG), and positron emission tomography (PET) techniques. Subsequently, we conducted a review of studies examining the neural correlates of consciousness in patients with DOC, with the findings holding potential value for the clinical application of DOC. Notably, previous research indicates that neuroimaging techniques have the potential to unveil covert awareness that conventional behavioral assessments might overlook. Furthermore, when integrated with various task paradigms or analytical approaches, this combination has the potential to significantly enhance the accuracy of both diagnosis and prognosis in DOC patients. Nonetheless, the stability of these neural biomarkers still needs additional validation, and future directions may entail integrating diagnostic and prognostic methods with big data and deep learning approaches.

PMID:39084451 | DOI:10.1016/j.brainres.2024.149133

Int J Biol Macromol. 2024 Jul 29:134289. doi: 10.1016/j.ijbiomac.2024.134289. Online ahead of print.

ABSTRACT

Human Serum Albumin (HSA), the most abundant protein in human body fluids, plays a crucial role in the transportation, absorption, metabolism, distribution, and excretion of drugs, significantly influencing their therapeutic efficacy. Despite the importance of HSA as a drug target, the available data on its interactions with external agents, such as drug-like molecules and antibodies, are limited, posing challenges for molecular modeling investigations and the development of empirical scoring functions or machine learning predictors for this target. Furthermore, the reported entries in existing databases often contain major inconsistencies due to varied experiments and conditions, raising concerns about data quality. To address these issues, a pioneering database, HSADab, was established through an extensive review of >30,000 scientific publications published between 1987 and 2023. The database encompasses over 5000 affinity data points at multiple temperatures and >130 crystal structures, including both ligand-bound and apo forms. The current HSADab resource (www.hsadab.cn) serves as a reliable foundation for validating molecular simulation protocols, such as traditional virtual screening workflows using docking, end-point, and al-chemical free energy techniques. Additionally, it provides a valuable data source for the implementation of machine learning predictors, including plasma protein binding models and plasma protein-based drug design models.

PMID:39084442 | DOI:10.1016/j.ijbiomac.2024.134289

Int J Biol Macromol. 2024 Jul 29:134293. doi: 10.1016/j.ijbiomac.2024.134293. Online ahead of print.

ABSTRACT

Proteolysis-targeting chimeras (PROTACs), as heterobifunctional molecules, have garnered significant attention for their ability to target previously undruggable proteins. Due to the challenges in obtaining crystal structures of PROTAC molecules in the ternary complex, a plethora of computational tools have been developed to aid in PROTAC design. These computational tools can be broadly classified into artificial intelligence (AI)-based or non-AI-based methods. This review aims to provide a comprehensive overview of the latest computational methods for the PROTAC design process, covering both AI and non-AI approaches, from protein selection to ternary complex modeling and prediction. Key considerations for in silico PROTAC design are discussed, along with additional considerations for deploying AI-based models. These considerations are intended to guide subsequent model development in the PROTAC design process. Finally, future directions and recommendations are provided.

PMID:39084437 | DOI:10.1016/j.ijbiomac.2024.134293

Sci Total Environ. 2024 Jul 29:175133. doi: 10.1016/j.scitotenv.2024.175133. Online ahead of print.

ABSTRACT

Predicting soil heavy metal (SHM) content is crucial for understanding SHM pollution levels in urban residential areas and guide efforts to reduce pollution. However, current research indicates low SHM prediction accuracy in urban areas. Therefore, we employed a deep learning method (fully connected deep neural network) alongside four other methods (muti-layer perceptron, radial basis function neural network, multiple stepwise linear regression, and Kriging interpolation) to predict SHM content in the urban residential areas of Beijing and demonstrated the strength of deep learning in improving prediction accuracy. We found the contents of the evaluated heavy metals (Cd, Cu, Pb, and Zn) exhibited significant correlations with numerous other soil physicochemical properties and environmental factors. The prediction accuracy for Cu, Pb, and Zn contents was relatively high across different methods. Notably, deep learning showed considerable strength in predicting the contents of the four heavy metals, with the R2 for the test set of the model ranging from 0.75 to 0.91. Compared to other methods, deep learning achieved markedly higher prediction accuracy according to different accuracy evaluation indicators (e.g., deep learning showed increases in the cumulative R2 of the four heavy metals ranging from 53.16 % to 187.36 % compared to other methods). Our study indicates that deep learning can significantly improve SHM content prediction accuracy in urban areas and is highly applicable in urban residential areas with complex environmental influences.

PMID:39084356 | DOI:10.1016/j.scitotenv.2024.175133

Ocul Surf. 2024 Jul 29:S1542-0124(24)00079-X. doi: 10.1016/j.jtos.2024.07.010. Online ahead of print.

ABSTRACT

PURPOSE: To develop an artificial intelligence (AI) model to diagnose Acanthamoeba keratitis (AK) based on in vivo confocal microscopy (IVCM) images extracted from the Heidelberg Retinal Tomograph 3 (HRT 3).

METHODS: This retrospective cohort study utilized IVCM images from patients who had received a culture-confirmed diagnosis of AK between 2013 and 2021 at Massachusetts Eye and Ear. Two cornea specialists independently labeled the images as AK or nonspecific finding (NSF) in a blind manner. Deep learning tasks were then conducted through Python and TensorFlow. Distinguishing between AK and NSF was designed as the task and completed through a devised convolutional neural network.

RESULTS: A dataset of 3,312 confocal images from 17 patients with a culture-confirmed diagnosis of AK was used in this study. The inter-rater agreement for identifying the presence or absence of AK in IVCM images was 84%, corresponding to a total of 2,782 images on which both observers agreed and were included in the model. 1,242 and 1,265 images of AK and NSF, respectively, were utilized in the training and validation sets, and 173 and 102 images of AK and NSF, respectively, were utilized in the evaluation set. Our model had an accuracy, sensitivity, and specificity of 76% each, and a precision of 78%.

CONCLUSIONS: We developed an HRT-based IVCM AI model for AK diagnosis utilizing culture-confirmed cases of AK. We achieved good accuracy in diagnosing AK and our model holds significant promise in the clinical application of AI in improving early AK diagnosis.

PMID:39084255 | DOI:10.1016/j.jtos.2024.07.010

Ultrasonics. 2024 Jul 26;143:107410. doi: 10.1016/j.ultras.2024.107410. Online ahead of print.

ABSTRACT

Ultrasound Localization Microscopy (ULM) surpasses the constraints imposed by acoustic diffraction, achieving sub-wavelength resolution visualization of microvasculature through the precise localization of minute microbubbles (MBs). Nonetheless, the analysis of densely populated regions with overlapping MB point spread responses introduces significant localization errors, limiting the use of technique to low-concentration conditions. This raises a trade-off issue between localization efficiency and MB density. In this work, we present a new deep learning framework that combines Transformer and U-Net architectures, termed ULM-TransUNet. As a non-linear model, it is able to learn the complex data patterns of overlapping MBs in dense conditions for accurate localization. To evaluate the performance of ULM-TransUNet, a series of numerical simulations and in vivo experiments are carried out. Numerical simulation results indicate that ULM-TransUNet achieves high-quality ULM imaging, with improvements of 21.93 % in detection rate, 17.36 % in detection precision, and 20.53 % in detection sensitivity, compared to previous state-of-the-art deep learning (DL) method (e.g., ULM-UNet). For the in vivo experiments, ULM-TransUNet achieves the highest spatial resolution (9.4 μm) and rapid inference speed (26.04 ms/frame). Furthermore, it consistently detects more small vessels and resolves closely spaced vessels more effectively. The outcomes of this work imply that ULM-TransUNet can potentially enhance the microvascular imaging performance on high-density MB conditions.

PMID:39084108 | DOI:10.1016/j.ultras.2024.107410

Nurse Educ Today. 2024 Jul 26;141:106321. doi: 10.1016/j.nedt.2024.106321. Online ahead of print.

ABSTRACT

OBJECTIVES: To explore research hotspots and theme trends in artificial intelligence in nurse education using bibliometric analysis.

DESIGN: Bibliometric analysis.

DATA SOURCES: Literature from the Web of Science Core Collection from the time of construction to October 31, 2023 was searched.

REVIEW METHODS: Analyses of countries, authors, institutions, journals, and keywords were conducted using Bibliometrix (based on R language), CiteSpace, the online analysis platform (bibliometric), Vosviewer, and Pajek.

RESULTS: A total of 135 articles with a straight upward trend over the last three years were retrieved. By fitting the curve R2 = 0.6022 (R2 > 0.4), we predicted that the number of annual articles is projected to grow in the coming years. The United States (n = 38), the National University of Singapore (n = 16), Professor Jun Ota (n = 8), and Nurse Education Today (n = 14) are the countries, institutions, authors, and journals that contributed to the most publications, respectively. Collaborative network analysis revealed that 32 institutional and 64 author collaborative teams were established. We identified ten high-frequency keywords and nine clusters. We categorized the research hotspots of artificial intelligence in nurse education into three areas: (1) Artificial intelligence-enhanced simulation robots, (2) machine learning and data mining, and (3) large language models based on natural language processing and deep learning. By analyzing the temporal and spatial evolution of keywords and burst detection, we found that future research trends may include (1) expanding and deepening the application of AI technology, (2) assessment of behavioral intent and educational outcomes, and (3) moral and ethical considerations.

CONCLUSIONS: Future research should be conducted on technology applications, behavioral intent, ethical policy, international cooperation, interdisciplinary cooperation, and sustainability to promote the continued development and innovation of AI in nurse education.

PMID:39084073 | DOI:10.1016/j.nedt.2024.106321

J Environ Manage. 2024 Jul 30;367:121934. doi: 10.1016/j.jenvman.2024.121934. Online ahead of print.

ABSTRACT

Ecological restoration is imperative for controlling desertification. Potential natural vegetation (PNV), the theoretical vegetation succession state, can guides near-natural restoration. Although a rising transition from traditional statistical methods to advanced machine learning and deep learning is observed in PNV simulation, a comprehensive comparison of their performance is still unexplored. Therefore, we overview the performance of PNV mapping in terms of 12 commonly used methods with varying spatial scales and sample sizes. Our findings indicate that the methodology should be carefully selected due to the variation in performance of different model types, with Area Under the Curve (AUC) values ranging from 0.65 to 0.95 for models with sample sizes up to 80% of the total sample size. Specifically, semi-supervised learning performs best with small sample sizes (i.e., 10 to 200), while Random Forest, XGBoost, and artificial neural networks perform better with large sample sizes (i.e., over 500). Further, the performance of all models tends to improve significantly as the sample size increases and the grain size of the crystals becomes smaller. Take the downstream Tarim River Basin, a hyper-arid region undergoing ecological restoration, as a case study. We showed that its potential restored areas were overestimated by 2-3 fold as the spatial scale became coarser, revealing the caution needed while planning restoration projects at coarse resolution. These findings enhance the application of PNV in the design of restoration programs to prevent desertification.

PMID:39083935 | DOI:10.1016/j.jenvman.2024.121934

J Med Internet Res. 2024 Jul 31;26:e58764. doi: 10.2196/58764.

ABSTRACT

Evidence-based medicine (EBM) emerged from McMaster University in the 1980-1990s, which emphasizes the integration of the best research evidence with clinical expertise and patient values. The Health Information Research Unit (HiRU) was created at McMaster University in 1985 to support EBM. Early on, digital health informatics took the form of teaching clinicians how to search MEDLINE with modems and phone lines. Searching and retrieval of published articles were transformed as electronic platforms provided greater access to clinically relevant studies, systematic reviews, and clinical practice guidelines, with PubMed playing a pivotal role. In the early 2000s, the HiRU introduced Clinical Queries-validated search filters derived from the curated, gold-standard, human-appraised Hedges dataset-to enhance the precision of searches, allowing clinicians to hone their queries based on study design, population, and outcomes. Currently, almost 1 million articles are added to PubMed annually. To filter through this volume of heterogenous publications for clinically important articles, the HiRU team and other researchers have been applying classical machine learning, deep learning, and, increasingly, large language models (LLMs). These approaches are built upon the foundation of gold-standard annotated datasets and humans in the loop for active machine learning. In this viewpoint, we explore the evolution of health informatics in supporting evidence search and retrieval processes over the past 25+ years within the HiRU, including the evolving roles of LLMs and responsible artificial intelligence, as we continue to facilitate the dissemination of knowledge, enabling clinicians to integrate the best available evidence into their clinical practice.

PMID:39083765 | DOI:10.2196/58764

ACS Appl Mater Interfaces. 2024 Jul 31. doi: 10.1021/acsami.4c05709. Online ahead of print.

ABSTRACT

Multifunctional metasurfaces have exhibited extensive potential in various fields, owing to their unparalleled capacity for controlling electromagnetic wave characteristics. The precise resolution is achieved through numerical simulation in conventional metasurface design methodologies. Nevertheless, the simulations using these approaches are inherently computationally costly. This paper proposes the Physical Insight Self-Correcting Convolutional Network (PISC-Net), which enables rapid prediction of infrared radiation spectra of metasurfaces with remarkable generalization capacity. In contrast to preceding prediction networks, we have enhanced the cognitive ability of the network to recognize physical mechanisms by designing parameter-communication modules and integrating a priori knowledge grounded in the parameter association mechanism. Additionally, we proposed an effective strategy for constructing data sets that facilitate precise tuning of absorption bands in the entire spectral range (3-14 μm) and serves to reduce the costs associated with data set development. Transfer learning is employed to obtain precise predictions for large-period metasurfaces from limited data sets. This approach demonstrates that a network trained exclusively on simulation data could predict experimental outcomes accurately, as proved by the comparative analysis between simulation, experimental testing, and prediction results. The average mean square error is less than 4%.

PMID:39083755 | DOI:10.1021/acsami.4c05709

J Clin Oncol. 2024 Jul 31:JCO2302641. doi: 10.1200/JCO.23.02641. Online ahead of print.

ABSTRACT

PURPOSE: Cancers with homologous recombination deficiency (HRD) can benefit from platinum salts and poly(ADP-ribose) polymerase inhibitors. Standard diagnostic tests for detecting HRD require molecular profiling, which is not universally available.

METHODS: We trained DeepHRD, a deep learning platform for predicting HRD from hematoxylin and eosin (H&E)-stained histopathological slides, using primary breast (n = 1,008) and ovarian (n = 459) cancers from The Cancer Genome Atlas (TCGA). DeepHRD was compared with four standard HRD molecular tests using breast (n = 349) and ovarian (n = 141) cancers from multiple independent data sets, including platinum-treated clinical cohorts with RECIST progression-free survival (PFS), complete response (CR), and overall survival (OS) endpoints.

RESULTS: DeepHRD predicted HRD from held-out H&E-stained breast cancer slides in TCGA with an AUC of 0.81 (95% CI, 0.77 to 0.85). This performance was confirmed in two independent primary breast cancer cohorts (AUC, 0.76 [95% CI, 0.71 to 0.82]). In an external platinum-treated metastatic breast cancer cohort, samples predicted as HRD had higher complete CR (AUC, 0.76 [95% CI, 0.54 to 0.93]) with 3.7-fold increase in median PFS (14.4 v 3.9 months; P = .0019) and hazard ratio (HR) of 0.45 (P = .0047). There were no significant differences in nonplatinum treatment outcome by predicted HRD status in three breast cancer cohorts, including CR (AUC, 0.39) and PFS (HR, 0.98, P = .95) in taxane-treated metastatic breast cancer. Through transfer learning to high-grade serous ovarian cancer, DeepHRD-predicted HRD samples had better OS after first-line (HR, 0.46; P = .030) and neoadjuvant (HR, 0.49; P = .015) platinum therapy in two cohorts.

CONCLUSION: DeepHRD can predict HRD in breast and ovarian cancers directly from routine H&E slides across multiple external cohorts, slide scanners, and tissue fixation variables. When compared with molecular testing, DeepHRD classified 1.8- to 3.1-fold more patients with HRD, which exhibited better OS in high-grade serous ovarian cancer and platinum-specific PFS in metastatic breast cancer.

PMID:39083703 | DOI:10.1200/JCO.23.02641

PLoS One. 2024 Jul 31;19(7):e0308181. doi: 10.1371/journal.pone.0308181. eCollection 2024.

ABSTRACT

PURPOSE: Organ-at-risk segmentation is essential in adaptive radiotherapy (ART). Learning-based automatic segmentation can reduce committed labor and accelerate the ART process. In this study, an auto-segmentation model was developed by employing individual patient datasets and a deep-learning-based augmentation method for tailoring radiation therapy according to the changes in the target and organ of interest in patients with prostate cancer.

METHODS: Two computed tomography (CT) datasets with well-defined labels, including contoured prostate, bladder, and rectum, were obtained from 18 patients. The labels of the CT images captured during radiation therapy (CT2nd) were predicted using CT images scanned before radiation therapy (CT1st). From the deformable vector fields (DVFs) created by using the VoxelMorph method, 10 DVFs were extracted when each of the modified CT and CT2nd images were deformed and registered to the fixed CT1st image. Augmented images were acquired by utilizing 110 extracted DVFs and spatially transforming the CT1st images and labels. An nnU-net autosegmentation network was trained by using the augmented images, and the CT2nd label was predicted. A patient-specific model was created for 18 patients, and the performances of the individual models were evaluated. The results were evaluated by employing the Dice similarity coefficient (DSC), average Hausdorff distance, and mean surface distance. The accuracy of the proposed model was compared with those of models trained with large datasets.

RESULTS: Patient-specific models were developed successfully. For the proposed method, the DSC values of the actual and predicted labels for the bladder, prostate, and rectum were 0.94 ± 0.03, 0.84 ± 0.07, and 0.83 ± 0.04, respectively.

CONCLUSION: We demonstrated the feasibility of automatic segmentation by employing individual patient datasets and image augmentation techniques. The proposed method has potential for clinical application in automatic prostate segmentation for ART.

PMID:39083552 | DOI:10.1371/journal.pone.0308181

PLoS One. 2024 Jul 31;19(7):e0305135. doi: 10.1371/journal.pone.0305135. eCollection 2024.

ABSTRACT

Smoke exposure is associated with bladder cancer (BC). However, little is known about whether the histologic changes of BC can predict the status of smoke exposure. Given this knowledge gap, the current study investigated the potential association between histology images and smoke exposure status. A total of 483 whole-slide histology images of 285 unique cases of BC were available from multiple centers for BC diagnosis. A deep learning model was developed to predict the smoke exposure status and externally validated on BC cases. The development set consisted of 66 cases from two centers. The external validation consisted of 94 cases from remaining centers for patients who either never smoked cigarettes or were active smokers at the time of diagnosis. The threshold for binary categorization was fixed to the median confidence score (65) of the development set. On external validation, AUC was used to assess the randomness of predicted smoke status; we utilized latent feature presentation to determine common histologic patterns for smoke exposure status and mixed effect logistic regression models determined the parameter independence from BC grade, gender, time to diagnosis, and age at diagnosis. We used 2,000-times bootstrap resampling to estimate the 95% Confidence Interval (CI) on the external validation set. The results showed an AUC of 0.67 (95% CI: 0.58-0.76), indicating non-randomness of model classification, with a specificity of 51.2% and sensitivity of 82.2%. Multivariate analyses revealed that our model provided an independent predictor for smoke exposure status derived from histology images, with an odds ratio of 1.710 (95% CI: 1.148-2.54). Common histologic patterns of BC were found in active or never smokers. In conclusion, deep learning reveals histopathologic features of BC that are predictive of smoke exposure and, therefore, may provide valuable information regarding smoke exposure status.

PMID:39083547 | DOI:10.1371/journal.pone.0305135

PLoS One. 2024 Jul 31;19(7):e0306549. doi: 10.1371/journal.pone.0306549. eCollection 2024.

ABSTRACT

Endometrial cancer screening is crucial for clinical treatment. Currently, cytopathologists analyze cytopathology images is considered a popular screening method, but manual diagnosis is time-consuming and laborious. Deep learning can provide objective guidance efficiency. But endometrial cytopathology images often come from different medical centers with different staining styles. It decreases the generalization ability of deep learning models in cytopathology images analysis, leading to poor performance. This study presents a robust automated screening framework for endometrial cancer that can be applied to cytopathology images with different staining styles, and provide an objective diagnostic reference for cytopathologists, thus contributing to clinical treatment. We collected and built the XJTU-EC dataset, the first cytopathology dataset that includes segmentation and classification labels. And we propose an efficient two-stage framework for adapting different staining style images, and screening endometrial cancer at the cellular level. Specifically, in the first stage, a novel CM-UNet is utilized to segment cell clumps, with a channel attention (CA) module and a multi-level semantic supervision (MSS) module. It can ignore staining variance and focus on extracting semantic information for segmentation. In the second stage, we propose a robust and effective classification algorithm based on contrastive learning, ECRNet. By momentum-based updating and adding labeled memory banks, it can reduce most of the false negative results. On the XJTU-EC dataset, CM-UNet achieves an excellent segmentation performance, and ECRNet obtains an accuracy of 98.50%, a precision of 99.32% and a sensitivity of 97.67% on the test set, which outperforms other competitive classical models. Our method robustly predicts endometrial cancer on cytopathologic images with different staining styles, which will further advance research in endometrial cancer screening and provide early diagnosis for patients. The code will be available on GitHub.

PMID:39083516 | DOI:10.1371/journal.pone.0306549

Microsc Microanal. 2024 Jul 31:ozae068. doi: 10.1093/mam/ozae068. Online ahead of print.

ABSTRACT

Automated particle analysis (APA) provides a vast amount of compositional data via energy-dispersive X-ray spectroscopy along with size and shape data via scanning electron microscopy for individual particles in a sample. In many instances, APA data are leveraged to support identification of the source of a sample based on the detection of particles of a specific composition. Often, the particles that provide context make up a minuscule portion of the sample. Additionally, the interpretation of complex samples can be difficult due to the diversity of compositions both in the mixture and within a particle. In this work, we demonstrate a method to compute and cluster similarity graphs that describe inter-particle relationships within a sample using a multi-modal few-shot learning neural network. As a proof-of-concept, we show that samples known to have been exposed to gunshot residue can be distinguished from samples occasionally mistaken for gunshot residue. Our workflow builds upon standard APA techniques and data processing methods to unveil additional information in a readily interpretable and quantitatively comparable format.

PMID:39083424 | DOI:10.1093/mam/ozae068

Mol Biol Cell. 2024 Jul 31:mbcE23110446. doi: 10.1091/mbc.E23-11-0446. Online ahead of print.

ABSTRACT

The key bacterial cell division protein FtsZ can adopt multiple conformations and prevailing models suggest that transitions from the closed to open state are necessary for filament formation and stability. Using all-atom molecular dynamics simulations, we analyzed state transitions of Staphylococcus aureus FtsZ as a monomer, dimer, and hexamer. We found that monomers can adopt intermediate states but preferentially adopt a closed state that is robust to forced re-opening. Dimer subunits transitioned between open and closed states, and dimers with both subunits in the closed state remained highly stable, suggesting that open-state conformations are not necessary for filament formation. Mg2+ strongly stabilized the conformation of GTP-bound subunits and the dimer filament interface. Our hexamer simulations indicate that the plus end subunit preferentially closes and that other subunits can transition between states without affecting inter-subunit stability. We found that rather than being correlated with subunit opening, inter-subunit stability was strongly correlated with catalytic site interactions. By leveraging deep-learning models, we identified key intra-subunit interactions governing state transitions. Our findings suggest a greater range of possible monomer and filament states than previously considered, and offer new insights into the nuanced interplay between subunit states and the critical role of nucleotide hydrolysis and Mg2+ in FtsZ filament dynamics.

PMID:39083352 | DOI:10.1091/mbc.E23-11-0446