It matters what you see: Graphic media images of war and terror may amplify distress.

Media exposure to graphic images of violence has proliferated in contemporary society, particularly with the advent of social media. Extensive exposure to media coverage immediately after the 9/11 attacks and the Boston Marathon bombings (BMB) was associated with more early traumatic stress symptoms; in fact, several hours of BMB-related daily media exposure was a stronger correlate of distress than being directly exposed to the bombings themselves. Researchers have replicated these findings across different traumatic events, extending this work to document that exposure to graphic images is independently and significantly associated with stress symptoms and poorer functioning. The media exposure-distress association also appears to be cyclical over time, with increased exposure predicting greater distress and greater distress predicting more media exposure following subsequent tragedies. The war in Israel and Gaza, which began on October 7, 2023, provides a current, real-time context to further explore these issues as journalists often share graphic images of death and destruction, making media-based graphic images once again ubiquitous and potentially challenging public well-being. For individuals sharing an identity with the victims or otherwise feeling emotionally connected to the Middle East, it may be difficult to avoid viewing these images. Through a review of research on the association between exposure to graphic images and public health, we discuss differing views on the societal implications of viewing such images and advocate for media literacy campaigns to educate the public to identify mis/disinformation and understand the risks of viewing and sharing graphic images with others.

Natural scenes reveal diverse representations of 2D and 3D body pose in the human brain.

Human pose, defined as the spatial relationships between body parts, carries instrumental information supporting the understanding of motion and action of a person. A substantial body of previous work has identified cortical areas responsive to images of bodies and different body parts. However, the neural basis underlying the visual perception of body part relationships has received less attention. To broaden our understanding of body perception, we analyzed high-resolution fMRI responses to a wide range of poses from over 4,000 complex natural scenes. Using ground-truth annotations and an application of three-dimensional (3D) pose reconstruction algorithms, we compared similarity patterns of cortical activity with similarity patterns built from human pose models with different levels of depth availability and viewpoint dependency. Targeting the challenge of explaining variance in complex natural image responses with interpretable models, we achieved statistically significant correlations between pose models and cortical activity patterns (though performance levels are substantially lower than the noise ceiling). We found that the 3D view-independent pose model, compared with two-dimensional models, better captures the activation from distinct cortical areas, including the right posterior superior temporal sulcus (pSTS). These areas, together with other pose-selective regions in the LOTC, form a broader, distributed cortical network with greater view-tolerance in more anterior patches. We interpret these findings in light of the computational complexity of natural body images, the wide range of visual tasks supported by pose structures, and possible shared principles for view-invariant processing between articulated objects and ordinary, rigid objects.

A computational account of transsaccadic attentional allocation based on visual gain fields.

Coordination of goal-directed behavior depends on the brain's ability to recover the locations of relevant objects in the world. In humans, the visual system encodes the spatial organization of sensory inputs, but neurons in early visual areas map objects according to their retinal positions, rather than where they are in the world. How the brain computes world-referenced spatial information across eye movements has been widely researched and debated. Here, we tested whether shifts of covert attention are sufficiently precise in space and time to track an object's real-world location across eye movements. We found that observers' attentional selectivity is remarkably precise and is barely perturbed by the execution of saccades. Inspired by recent neurophysiological discoveries, we developed an observer model that rapidly estimates the real-world locations of objects and allocates attention within this reference frame. The model recapitulates the human data and provides a parsimonious explanation for previously reported phenomena in which observers allocate attention to task-irrelevant locations across eye movements. Our findings reveal that visual attention operates in real-world coordinates, which can be computed rapidly at the earliest stages of cortical processing.

Deep learning in spatially resolved transcriptfomics: a comprehensive technical view

Spatially resolved transcriptomics (SRT) is a pioneering method for simultaneously studying morphological contexts and gene expression at single-cell precision. Data emerging from SRT are multifaceted, presenting researchers with intricate gene expression matrices, precise spatial details and comprehensive histology visuals. Such rich and intricate datasets, unfortunately, render many conventional methods like traditional machine learning and statistical models ineffective. The unique challenges posed by the specialized nature of SRT data have led the scientific community to explore more sophisticated analytical avenues. Recent trends indicate an increasing reliance on deep learning algorithms, especially in areas such as spatial clustering, identification of spatially variable genes and data alignment tasks. In this manuscript, we provide a rigorous critique of these advanced deep learning methodologies, probing into their merits, limitations and avenues for further refinement. Our in-depth analysis underscores that while the recent innovations in deep learning tailored for SRT have been promising, there remains a substantial potential for enhancement. A crucial area that demands attention is the development of models that can incorporate intricate biological nuances, such as phylogeny-aware processing or in-depth analysis of minuscule histology image segments. Furthermore, addressing challenges like the elimination of batch effects, perfecting data normalization techniques and countering the overdispersion and zero inflation patterns seen in gene expression is pivotal. To support the broader scientific community in their SRT endeavors, we have meticulously assembled a comprehensive directory of readily accessible SRT databases, hoping to serve as a foundation for future research initiatives.

Improving artificial intelligence pipeline for liver malignancy diagnosis using ultrasound images and video frames.

Recent developments of deep learning methods have demonstrated their feasibility in liver malignancy diagnosis using ultrasound (US) images. However, most of these methods require manual selection and annotation of US images by radiologists, which limit their practical application. On the other hand, US videos provide more comprehensive morphological information about liver masses and their relationships with surrounding structures than US images, potentially leading to a more accurate diagnosis. Here, we developed a fully automated artificial intelligence (AI) pipeline to imitate the workflow of radiologists for detecting liver masses and diagnosing liver malignancy. In this pipeline, we designed an automated mass-guided strategy that used segmentation information to direct diagnostic models to focus on liver masses, thus increasing diagnostic accuracy. The diagnostic models based on US videos utilized bi-directional convolutional long short-term memory modules with an attention-boosted module to learn and fuse spatiotemporal information from consecutive video frames. Using a large-scale dataset of 50 063 US images and video frames from 11 468 patients, we developed and tested the AI pipeline and investigated its applications. A dataset of annotated US images is available at https://doi.org/10.5281/zenodo.7272660.

THItoGene: a deep learning method for predicting spatial transcriptomics from histological images.

Spatial transcriptomics unveils the complex dynamics of cell regulation and transcriptomes, but it is typically cost-prohibitive. Predicting spatial gene expression from histological images via artificial intelligence offers a more affordable option, yet existing methods fall short in extracting deep-level information from pathological images. In this paper, we present THItoGene, a hybrid neural network that utilizes dynamic convolutional and capsule networks to adaptively sense potential molecular signals in histological images for exploring the relationship between high-resolution pathology image phenotypes and regulation of gene expression. A comprehensive benchmark evaluation using datasets from human breast cancer and cutaneous squamous cell carcinoma has demonstrated the superior performance of THItoGene in spatial gene expression prediction. Moreover, THItoGene has demonstrated its capacity to decipher both the spatial context and enrichment signals within specific tissue regions. THItoGene can be freely accessed at https://github.com/yrjia1015/THItoGene.

CRCS: An automatic image processing pipeline for hormone level analysis of Cushing's disease.

Due to the abnormal secretion of adreno-cortico-tropic-hormone (ACTH) by tumors, Cushing's disease leads to hypercortisonemia, a precursor to a series of metabolic disorders and serious complications. Cushing's disease has high recurrence rate, short recurrence time and undiscovered recurrence reason after surgical resection. Qualitative or quantitative automatic image analysis of histology images can potentially in providing insights into Cushing's disease, but still no software has been available to the best of our knowledge. In this study, we propose a quantitative image analysis-based pipeline CRCS, which aims to explore the relationship between the expression level of ACTH in normal cell tissues adjacent to tumor cells and the postoperative prognosis of patients. CRCS mainly consists of image-level clustering, cluster-level multi-modal image registration, patch-level image classification and pixel-level image segmentation on the whole slide imaging (WSI). On both image registration and classification tasks, our method CRCS achieves state-of-the-art performance compared to recently published methods on our collected benchmark dataset. In addition, CRCS achieves an accuracy of 0.83 for postoperative prognosis of 12 cases. CRCS demonstrates great potential for instrumenting automatic diagnosis and treatment for Cushing's disease.

MRUNet-3D: A multi-stride residual 3D UNet for lung nodule segmentation.

Obtaining an accurate segmentation of the pulmonary nodules in computed tomography (CT) images is challenging. This is due to: (1) the heterogeneous nature of the lung nodules; (2) comparable visual characteristics between the nodules and their surroundings. A robust multi-scale feature extraction mechanism that can effectively obtain multi-scale representations at a granular level can improve segmentation accuracy. As the most commonly used network in lung nodule segmentation, UNet, its variants, and other image segmentation methods lack this robust feature extraction mechanism. In this study, we propose a multi-stride residual 3D UNet (MRUNet-3D) to improve the segmentation accuracy of lung nodules in CT images. It incorporates a multi-slide Res2Net block (MSR), which replaces the simple sequence of convolution layers in each encoder stage to effectively extract multi-scale features at a granular level from different receptive fields and resolutions while conserving the strengths of 3D UNet. The proposed method has been extensively evaluated on the publicly available LUNA16 dataset. Experimental results show that it achieves competitive segmentation performance with an average dice similarity coefficient of 83.47 % and an average surface distance of 0.35 mm on the dataset. More notably, our method has proven to be robust to the heterogeneity of lung nodules. It has also proven to perform better at segmenting small lung nodules. Ablation studies have shown that the proposed MSR and RFIA modules are fundamental to improving the performance of the proposed model.

Larger images are better remembered during naturalistic encoding.

We are constantly exposed to multiple visual scenes, and while freely viewing them without an intentional effort to memorize or encode them, only some are remembered. It has been suggested that image memory is influenced by multiple factors, such as depth of processing, familiarity, and visual category. However, this is typically investigated when people are instructed to perform a task (e.g., remember or make some judgment about the images), which may modulate processing at multiple levels and thus, may not generalize to naturalistic visual behavior. Visual memory is assumed to rely on high-level visual perception that shows a level of size invariance and therefore is not assumed to be highly dependent on image size. Here, we reasoned that during naturalistic vision, free of task-related modulations, bigger images stimulate more visual system processing resources (from retina to cortex) and would, therefore, be better remembered. In an extensive set of seven experiments, naïve participants (n = 182) were asked to freely view presented images (sized 3° to 24°) without any instructed encoding task. Afterward, they were given a surprise recognition test (midsized images, 50% already seen). Larger images were remembered better than smaller ones across all experiments (∼20% higher accuracy or ∼1.5 times better). Memory was proportional to image size, faces were better remembered, and outdoors the least. Results were robust even when controlling for image set, presentation order, screen resolution, image scaling at test, or the amount of information. While multiple factors affect image memory, our results suggest that low- to high-level processes may all contribute to image memory.

Quantitative, in situ visualization of intracellular insulin vesicles in pancreatic beta cells.

Characterizing relationships between Zn2+, insulin, and insulin vesicles is of vital importance to the study of pancreatic beta cells. However, the precise content of Zn2+ and the specific location of insulin inside insulin vesicles are not clear, which hinders a thorough understanding of the insulin secretion process and diseases caused by blood sugar dysregulation. Here, we demonstrated the colocalization of Zn2+ and insulin in both single extracellular insulin vesicles and pancreatic beta cells by using an X-ray scanning coherent diffraction imaging (ptychography) technique. We also analyzed the elemental Zn2+ and Ca2+ contents of insulin vesicles using electron microscopy and energy dispersive spectroscopy (EDS) mapping. We found that the presence of Zn2+ is an important characteristic that can be used to distinguish insulin vesicles from other types of vesicles in pancreatic beta cells and that the content of Zn2+ is proportional to the size of insulin vesicles. By using dual-energy contrast X-ray microscopy and scanning transmission X-ray microscopy (STXM) image stacks, we observed that insulin accumulates in the off-center position of extracellular insulin vesicles. Furthermore, the spatial distribution of insulin vesicles and their colocalization with other organelles inside pancreatic beta cells were demonstrated using three-dimensional (3D) imaging by combining X-ray ptychography and an equally sloped tomography (EST) algorithm. This study describes a powerful method to univocally describe the location and quantitative analysis of intracellular insulin, which will be of great significance to the study of diabetes and other blood sugar diseases.

Preferred Orientation and Phase Distribution of Quasi-2D Perovskite for Bifunctional Light-Emitting Photodetectors.

In traditional optical wireless communication (OWC) systems, the simultaneous use of multiple sets of light-emitting diodes (LEDs) and photodetectors (PDs) increases the system complexity and instability. Here we report bifunctional light-emitting photodetectors (LEPDs) fabricated with quasi-2D perovskite (F-PEA)2Cs4Pb5I11Br5 as light-emitting/detecting layers for efficient, miniaturized, and intelligent bidirectional OWC. By simply changing the solvent composition of the precursor solution and using antisolvent engineering, we manipulated the crystal orientation and phase distribution of (F-PEA)2Cs4Pb5I11Br5, realizing high irradiance (4.36 µW cm-2) and a -3 dB refresh rate (0.21 MHz) of electroluminescence in LED mode as well as low noise (below 1 pA Hz-1/2) and high responsivity (0.1 A W-1) in PD mode. The rapid and accurate OWC process was demonstrated through interaction of LEPDs. We also demonstrated the high-fidelity compression and digitization of high-resolution (256 × 256 pixels) color images using the four-step phase shift method to realize intelligent encrypted image OWC.

Deep learning -- promises for 3D nuclear imaging: a guide for biologists.

For the past century, the nucleus has been the focus of extensive investigations in cell biology. However, many questions remain about how its shape and size are regulated during development, in different tissues, or during disease and aging. To track these changes, microscopy has long been the tool of choice. Image analysis has revolutionized this field of research by providing computational tools that can be used to translate qualitative images into quantitative parameters. Many tools have been designed to delimit objects in 2D and, eventually, in 3D in order to define their shapes, their number or their position in nuclear space. Today, the field is driven by deep-learning methods, most of which take advantage of convolutional neural networks. These techniques are remarkably adapted to biomedical images when trained using large datasets and powerful computer graphics cards. To promote these innovative and promising methods to cell biologists, this Review summarizes the main concepts and terminologies of deep learning. Special emphasis is placed on the availability of these methods. We highlight why the quality and characteristics of training image datasets are important and where to find them, as well as how to create, store and share image datasets. Finally, we describe deep-learning methods well-suited for 3D analysis of nuclei and classify them according to their level of usability for biologists. Out of more than 150 published methods, we identify fewer than 12 that biologists can use, and we explain why this is the case. Based on this experience, we propose best practices to share deep-learning methods with biologists.

Three-dimensional object geometry of mitochondria-associated signal: 3-D analysis pipeline for two-photon image stacks of cerebrovascular endothelial mitochondria.

Increasing evidence indicates the role of mitochondrial and vascular dysfunction in aging and aging-associated pathologies; however, the exact mechanisms and chronological processes remain enigmatic. High-energy demand organs, such as the brain, depend on the health of their mitochondria and vasculature for the maintenance of normal functions, therefore representing vulnerable targets for aging. This methodology article describes an analysis pipeline for three-dimensional (3-D) mitochondria-associated signal geometry of two-photon image stacks of brain vasculature. The analysis methods allow the quantification of mitochondria-associated signals obtained in real time in their physiological environment. In addition, signal geometry results will allow the extrapolation of fission and fusion events under normal conditions, during aging, or in the presence of different pathological conditions, therefore contributing to our understanding of the role mitochondria play in a variety of aging-associated diseases with vascular etiology.NEW & NOTEWORTHY Analysis pipeline for 3-D mitochondria-associated signal geometry of two-photon image stacks of brain vasculature.

Revolutionizing breast cancer Ki-67 diagnosis: ultrasound radiomics and fully connected neural networks (FCNN) combination method.

PURPOSE: This study aims to assess the diagnostic value of ultrasound habitat sub-region radiomics feature parameters using a fully connected neural networks (FCNN) combination method L2,1-norm in relation to breast cancer Ki-67 status. METHODS: Ultrasound images from 528 cases of female breast cancer at the Affiliated Hospital of Xiangnan University and 232 cases of female breast cancer at the Affiliated Rehabilitation Hospital of Xiangnan University were selected for this study. We utilized deep learning methods to automatically outline the gross tumor volume and perform habitat clustering. Subsequently, habitat sub-regions were extracted to identify radiomics features and underwent feature engineering using the L1,2-norm. A prediction model for the Ki-67 status of breast cancer patients was then developed using a FCNN. The model's performance was evaluated using accuracy, area under the curve (AUC), specificity (Spe), positive predictive value (PPV), negative predictive value (NPV), Recall, and F1. In addition, calibration curves and clinical decision curves were plotted for the test set to visually assess the predictive accuracy and clinical benefit of the models. RESULT: Based on the feature engineering using the L1,2-norm, a total of 9 core features were identified. The predictive model, constructed by the FCNN model based on these 9 features, achieved the following scores: ACC 0.856, AUC 0.915, Spe 0.843, PPV 0.920, NPV 0.747, Recall 0.974, and F1 0.890. Furthermore, calibration curves and clinical decision curves of the validation set demonstrated a high level of confidence in the model's performance and its clinical benefit. CONCLUSION: Habitat clustering of ultrasound images of breast cancer is effectively supported by the combined implementation of the L1,2-norm and FCNN algorithms, allowing for the accurate classification of the Ki-67 status in breast cancer patients.

Bio-net dataset: AI-based diagnostic solutions using peripheral blood smear images.

Peripheral blood smear examination is one of the basic steps in the evaluation of different blood cells. It is a confirmatory step after an automated complete blood count analysis. Manual microscopy is time-consuming and requires professional laboratory expertise. Therefore, the turn-around time for peripheral smear in a health care center is approximately 3-4 hours. To avoid the traditional method of manual counting under the microscope a computerized automation of peripheral blood smear examination has been adopted, which is a challenging task in medical diagnostics. In recent times, deep learning techniques have overcome the challenges associated with human microscopic evaluation of peripheral smears and this has led to reduced cost and precise diagnosis. However, their application can be significantly improved by the availability of annotated datasets. This study presents a large customized annotated blood cell dataset (named the Bio-Net dataset from healthy individuals) and blood cell detection and counting in the peripheral blood smear images. A mini-version of the dataset for specialized WBC-based image processing tasks is also equipped to classify the healthy and mature WBCs in their respective classes. An object detection algorithm called You Only Look Once (YOLO) with a refashion disposition has been trained on the novel dataset to automatically detect and classify blood cells into RBCs, WBCs, and platelets and compare the results with other publicly available datasets to highlight the versatility. In short the introduction of the Bio-Net dataset and AI-powered detection and counting offers a significant potential for advancement in biomedical research for analyzing and understanding biological data.

Multi-modality artificial intelligence in digital pathology.

In common medical procedures, the time-consuming and expensive nature of obtaining test results plagues doctors and patients. Digital pathology research allows using computational technologies to manage data, presenting an opportunity to improve the efficiency of diagnosis and treatment. Artificial intelligence (AI) has a great advantage in the data analytics phase. Extensive research has shown that AI algorithms can produce more up-to-date and standardized conclusions for whole slide images. In conjunction with the development of high-throughput sequencing technologies, algorithms can integrate and analyze data from multiple modalities to explore the correspondence between morphological features and gene expression. This review investigates using the most popular image data, hematoxylin-eosin stained tissue slide images, to find a strategic solution for the imbalance of healthcare resources. The article focuses on the role that the development of deep learning technology has in assisting doctors' work and discusses the opportunities and challenges of AI.

SIFLoc: a self-supervised pre-training method for enhancing the recognition of protein subcellular localization in immunofluorescence microscopic images.

With the rapid growth of high-resolution microscopy imaging data, revealing the subcellular map of human proteins has become a central task in the spatial proteome. The cell atlas of the Human Protein Atlas (HPA) provides precious resources for recognizing subcellular localization patterns at the cell level, and the large-scale annotated data enable learning via advanced deep neural networks. However, the existing predictors still suffer from the imbalanced class distribution and the lack of labeled data for minor classes. Thus, it is necessary to develop new methods for coping with these issues. We leverage the self-supervised learning protocol to address these problems. Especially, we propose a pre-training scheme to enhance the conventional supervised learning framework called SIFLoc. The pre-training is featured by a hybrid data augmentation method and a modified contrastive loss function, aiming to learn good feature representations from microscopic images. The experiments are performed on a large-scale immunofluorescence microscopic image dataset collected from the HPA database. Using the same deep neural networks as the classifier, the model pre-trained via SIFLoc not only outperforms the model without pre-training by a large margin but also shows advantages over the state-of-the-art self-supervised learning methods. Especially, SIFLoc improves the prediction accuracy for minor organelles significantly.

Conditional generative adversarial network driven radiomic prediction of mutation status based on magnetic resonance imaging of breast cancer.

BACKGROUND: Breast Cancer (BC) is a highly heterogeneous and complex disease. Personalized treatment options require the integration of multi-omic data and consideration of phenotypic variability. Radiogenomics aims to merge medical images with genomic measurements but encounter challenges due to unpaired data consisting of imaging, genomic, or clinical outcome data. In this study, we propose the utilization of a well-trained conditional generative adversarial network (cGAN) to address the unpaired data issue in radiogenomic analysis of BC. The generated images will then be used to predict the mutations status of key driver genes and BC subtypes. METHODS: We integrated the paired MRI and multi-omic (mRNA gene expression, DNA methylation, and copy number variation) profiles of 61 BC patients from The Cancer Imaging Archive (TCIA) and The Cancer Genome Atlas (TCGA). To facilitate this integration, we employed a Bayesian Tensor Factorization approach to factorize the multi-omic data into 17 latent features. Subsequently, a cGAN model was trained based on the matched side-view patient MRIs and their corresponding latent features to predict MRIs for BC patients who lack MRIs. Model performance was evaluated by calculating the distance between real and generated images using the Fréchet Inception Distance (FID) metric. BC subtype and mutation status of driver genes were obtained from the cBioPortal platform, where 3 genes were selected based on the number of mutated patients. A convolutional neural network (CNN) was constructed and trained using the generated MRIs for mutation status prediction. Receiver operating characteristic area under curve (ROC-AUC) and precision-recall area under curve (PR-AUC) were used to evaluate the performance of the CNN models for mutation status prediction. Precision, recall and F1 score were used to evaluate the performance of the CNN model in subtype classification. RESULTS: The FID of the images from the well-trained cGAN model based on the test set is 1.31. The CNN for TP53, PIK3CA, and CDH1 mutation prediction yielded ROC-AUC values 0.9508, 0.7515, and 0.8136 and PR-AUC are 0.9009, 0.7184, and 0.5007, respectively for the three genes. Multi-class subtype prediction achieved precision, recall and F1 scores of 0.8444, 0.8435 and 0.8336 respectively. The source code and related data implemented the algorithms can be found in the project GitHub at https://github.com/mattthuang/BC_RadiogenomicGAN . CONCLUSION: Our study establishes cGAN as a viable tool for generating synthetic BC MRIs for mutation status prediction and subtype classification to better characterize the heterogeneity of BC in patients. The synthetic images also have the potential to significantly augment existing MRI data and circumvent issues surrounding data sharing and patient privacy for future BC machine learning studies.

Predictive models for personalized precision medical intervention in spontaneous regression stages of cervical precancerous lesions.

BACKGROUND: During the prolonged period from Human Papillomavirus (HPV) infection to cervical cancer development, Low-Grade Squamous Intraepithelial Lesion (LSIL) stage provides a critical opportunity for cervical cancer prevention, giving the high potential for reversal in this stage. However, there is few research and a lack of clear guidelines on appropriate intervention strategies at this stage, underscoring the need for real-time prognostic predictions and personalized treatments to promote lesion reversal. METHODS: We have established a prospective cohort. Since 2018, we have been collecting clinical data and pathological images of HPV-infected patients, followed by tracking the progression of their cervical lesions. In constructing our predictive models, we applied logistic regression and six machine learning models, evaluating each model's predictive performance using metrics such as the Area Under the Curve (AUC). We also employed the SHAP method for interpretative analysis of the prediction results. Additionally, the model identifies key factors influencing the progression of the lesions. RESULTS: Model comparisons highlighted the superior performance of Random Forests (RF) and Support Vector Machines (SVM), both in clinical parameter and pathological image-based predictions. Notably, the RF model, which integrates pathological images and clinical multi-parameters, achieved the highest AUC of 0.866. Another significant finding was the substantial impact of sleep quality on the spontaneous clearance of HPV and regression of LSIL. CONCLUSIONS: In contrast to current cervical cancer prediction models, our model's prognostic capabilities extend to the spontaneous regression stage of cervical cancer. This model aids clinicians in real-time monitoring of lesions and in developing personalized treatment or follow-up plans by assessing individual risk factors, thus fostering lesion spontaneous reversal and aiding in cervical cancer prevention and reduction.

Deep learning-based pathological prediction of lymph node metastasis for patient with renal cell carcinoma from primary whole slide images.

BACKGROUND: Metastasis renal cell carcinoma (RCC) patients have extremely high mortality rate. A predictive model for RCC micrometastasis based on pathomics could be beneficial for clinicians to make treatment decisions. METHODS: A total of 895 formalin-fixed and paraffin-embedded whole slide images (WSIs) derived from three cohorts, including Shanghai General Hospital (SGH), Clinical Proteomic Tumor Analysis Consortium (CPTAC) and Cancer Genome Atlas (TCGA) cohorts, and another 588 frozen section WSIs from TCGA dataset were involved in the study. The deep learning-based strategy for predicting lymphatic metastasis was developed based on WSIs through clustering-constrained-attention multiple-instance learning method and verified among the three cohorts. The performance of the model was further verified in frozen-pathological sections. In addition, the model was also tested the prognosis prediction of patients with RCC in multi-source patient cohorts. RESULTS: The AUC of the lymphatic metastasis prediction performance was 0.836, 0.865 and 0.812 in TCGA, SGH and CPTAC cohorts, respectively. The performance on frozen section WSIs was with the AUC of 0.801. Patients with high deep learning-based prediction of lymph node metastasis values showed worse prognosis. CONCLUSIONS: In this study, we developed and verified a deep learning-based strategy for predicting lymphatic metastasis from primary RCC WSIs, which could be applied in frozen-pathological sections and act as a prognostic factor for RCC to distinguished patients with worse survival outcomes.