Deep Learning-based Image Analysis Research Articles

Adversarial robustness improvement for X-ray bone segmentation using synthetic data created from computed tomography scans

Deep learning-based image analysis offers great potential in clinical practice. However, it faces mainly two challenges: scarcity of large-scale annotated clinical data for training and susceptibility to adversarial data in inference. As an example, an artificial intelligence (AI) system could check patient positioning, by segmenting and evaluating relative positions of anatomical structures in medical images. Nevertheless, data to train such AI system might be highly imbalanced with mostly well-positioned images being available. Thus, we propose the use of synthetic X-ray images and annotation masks forward projected from 3D photon-counting CT volumes to create realistic non-optimally positioned X-ray images for training. An open-source model (TotalSegmentator) was used to annotate the clavicles in 3D CT volumes. We evaluated model robustness with respect to the internal (simulated) patient rotation α\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\alpha$$\\end{document} on real-data-trained models and real&synthetic-data-trained models. Our results showed that real&synthetic- data-trained models have Dice score percentage improvements of 3% to 15% across different α\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\alpha$$\\end{document} groups compared to the real-data-trained model. Therefore, we demonstrated that synthetic data could be supplementary used to train and enrich heavily underrepresented conditions to increase model robustness.

Enhanced powder characteristics of succinic acid through crystallization techniques for food industry application

During crystallization, succinic acid, an important food additive, exhibit severe agglomeration behaviour, uneven crystal size distribution (CSD), and irregular crystal morphology, resulting in poor powder flow properties. This paper presents an innovative approach to optimize succinic acid crystallization through investigation of novel blade milling method, seeding, and inclusion of green additive, cetyltrimethylammonium bromide (CTAB). CTAB for crystal habit regulation and formation of desired cubic-like morphology. We employ Mask R-CNN, a deep learning-based image analysis tool to quantitatively assess crystal and powder flow properties. Results demonstrate that the inclusion of 0.5 wt% CTAB with two-stage cooling method yields uniform CSD ranging from 30 μm to 160 μm and improves powder properties, reducing the agglomeration degree by 46% and caking ratio to only 6.78%, indicating a decreased tendency for clumping. The integration of crystallization with the proposed deep learning framework not only improves succinic acid food functionality but also showcases highly efficient method for optimizing crystal properties, setting new standard for food crystallization processes. It has been adequately presented with real-time high-precision analysis of CSD, agglomeration, etc, facilitating rapid screening for optimized food quality improvements.

Estimating body weight of caged sea cucumbers (Apostichopus japonicus) using an underwater time-lapse camera and image analysis by semantic segmentation

Image analysis is being developed to improve the efficiency of fishery and aquaculture technologies. Optical cameras are an easy and cost-effective method for monitoring fish and other species. In this study, a monitoring system that combines an underwater time-lapse camera and a deep learning-based image analysis was developed for utilization in integrated multi-trophic aquaculture (IMTA). The sea cucumber (Apostichopus japonicus) was used as a target species because the technology necessary for estimating growth, particularly in terms of weight, of caged sea cucumber using an underwater environment is still under study. Therefore, semantic segmentation was applied to classify the images into caged sea cucumbers and various underwater backgrounds. Multiple images of sea cucumbers were captured in a water tank that mimicked the box cage used in IMTA, and their body weights were measured simultaneously. For model development, approximately 1,300 images were prepared for the training and validation processes. The model then achieved an IoU (Intersection over Union) of approximately 94 % for the validation data. Next, the pixel numbers of sea cucumbers were converted into an area calculated using the size of the cage net as the background. The relationship between the area and weight of sea cucumbers yielded an approximate line for estimating body weight. As a result, the approximation line had a coefficient of determination of R2 = 0.87 for training and validation data and RMSE (Root Mean Square Error) =1.81 and 6.78 g for sea cucumbers less than 10 and 110 g, respectively. Using the model, test images in an actual IMTA situation were applied, and the estimated body weights were close to the measured values for small sea cucumbers. If we apply this model to images obtained over an extended period, the growth of sea cucumbers in a time series can be understood.

Well Plate-Based Localized Electroporation Workflow for Rapid Optimization of Intracellular Delivery.

Efficient and nontoxic delivery of foreign cargo into cells is a critical step in many biological studies and cell engineering workflows with applications in areas such as biomanufacturing and cell-based therapeutics. However, effective molecular delivery into cells involves optimizing several experimental parameters. In the case of electroporation-based intracellular delivery, there is a need to optimize parameters like pulse voltage, duration, buffer type, and cargo concentration for each unique application. Here, we present the protocol for fabricating and utilizing a high-throughput multi-well localized electroporation device (LEPD) assisted by deep learning-based image analysis to enable rapid optimization of experimental parameters for efficient and nontoxic molecular delivery into cells. The LEPD and the optimization workflow presented herein are relevant to both adherent and suspended cell types and different molecular cargo (DNA, RNA, and proteins). The workflow enables multiplexed combinatorial experiments and can be adapted to cell engineering applications requiring in vitro delivery. Key features • A high-throughput multi-well localized electroporation device (LEPD) that can be optimized for both adherent and suspended cell types. • Allows for multiplexed experiments combined with tailored pulse voltage, duration, buffer type, and cargo concentration. • Compatible with various molecular cargoes, including DNA, RNA, and proteins, enhancing its versatility for cell engineering applications. • Integration with deep learning-based image analysis enables rapid optimization of experimental parameters.

Deep learning-based image analysis for filamentous and floc-forming bacteria in wastewater treatment

In municipal wastewater treatment, effective secondary clarification relies on the balance between floc-forming bacteria and filamentous bacteria. Consequently, comprehensive and real-time monitoring of this balance will enable reliable operation of biological wastewater treatment. This research presents an artificial intelligence (AI)-based approach for the classification of filamentous and floc-forming bacteria in microscopic images using deep learning. To provide ground truth labeling, an automated rule-based segmentation algorithm was developed using color and morphology criteria along with supplementary filtration steps to enhance the precision of filamentous and floc-forming bacteria identification. The segmentation algorithm demonstrated reliable detection and categorization of bacteria across varying background intensities and effectively recognized intricate microbial configurations. Subsequently, the supervised deep learning model was trained on the segmented images and constructed with an encoder/decoder architecture. Machine training with only 68 microscopic images demonstrated successful classification of the filamentous and floc-forming bacteria with a 97.8 % accuracy. In addition, qualitative evaluation demonstrated that the deep learning model could generalize machine understanding across diverse scenarios and discern misclassified filamentous bacteria accurately. The proposed model stands as a promising automated tool for real-time quantification of filamentous and floc-forming bacteria in bioreactors and clarifiers, offering the potential for reliable operation as well as immediate actions for sludge bulking and membrane fouling problems.

GeoAI Reproducibility and Replicability: A Computational and Spatial Perspective

GeoAI has emerged as an exciting interdisciplinary research area that combines spatial theories and data with cutting-edge AI models to address geospatial problems in a novel, data-driven manner. Although GeoAI research has flourished in the GIScience literature, its reproducibility and replicability (R&R), fundamental principles that determine the reusability, reliability, and scientific rigor of research findings, have rarely been discussed. This article aims to provide an in-depth analysis of this topic from both computational and spatial perspectives. We first categorize the major goals for reproducing GeoAI research, namely, validation (repeatability), learning and adapting the method for solving a similar or new problem (reproducibility), and examining the generalizability of the research findings (replicability). Each of these goals requires different levels of understanding of GeoAI, as well as different methods to ensure its success. We then discuss the factors that might cause the lack of R&R in GeoAI research, with an emphasis on (1) the selection and use of training data; (2) the uncertainty that resides in the GeoAI model design, training, deployment, and inference processes; and more important (3) the inherent spatial heterogeneity of geospatial data and processes. We use a deep learning-based image analysis task as an example to demonstrate the results’ uncertainty and spatial variance caused by different factors. The findings reiterate the importance of knowledge sharing, as well as the generation of a “replicability map” that incorporates spatial autocorrelation and spatial heterogeneity into consideration in quantifying the spatial replicability of GeoAI research.

A deep learning-based image analysis for assessing the extent of abduction in abducens nerve palsy patients before and after strabismus surgery

PurposeThis study aimed to propose a novel deep learning-based approach to assess the extent of abduction in patients with abducens nerve palsy before and after strabismus surgery. MethodsThis study included 13 patients who were diagnosed with abducens nerve palsy and underwent strabismus surgery in a tertiary hospital. Photographs of primary, dextroversion and levoversion position were collected before and after strabismus surgery. The eye location and eye segmentation network were trained via recurrent residual convolutional neural networks with attention gate connection based on U-Net (R2AU-Net). Facial images of abducens nerve palsy patients were used as the test set and parameters were measured automatically based on the masked images. Absolute abduction also was measured manually, and relative abduction was calculated. Agreements between manual and automatic measurements, as well as repeated automatic measurements were analyzed. Preoperative and postoperative results were compared. ResultsThe intraclass correlation coefficients (ICCs) between manual and automatic measurements of absolute abduction ranged from 0.985 to 0.992 (P＜0.001), and the bias ranged from −0.25 ​mm to −0.05 ​mm. The ICCs between two repeated automatic measurements ranged from 0.994 to 0.997 (P＜0.001), and the bias ranged from −0.11 ​mm to 0.05 ​mm. After strabismus surgery, absolute abduction of affected eye increased from 2.18 ​± ​1.40 ​mm to 3.36 ​± ​1.93 ​mm (P＜0.05). The relative abduction was improved in 76.9% patients (10/13) after surgery (P＜0.01). ConclusionsThis image analysis technique demonstrated excellent accuracy and repeatability for automatic measurements of ocular abduction, which has promising application prospects in objectively assessing surgical outcomes in patients with abducens nerve palsy.

Modeling heart disease using human cells, bio-inspired cell culture systems and deep learning-based image analysis to discovery new therapeutic candidates

Modeling heart disease using human cells, bio-inspired cell culture systems and deep learning-based image analysis to discovery new therapeutic candidates

Abstract PO4-07-05: Clinical evaluation of image-based risk profiling in breast cancer histopathology and comparison to an established gene expression assay

Abstract Background: Among oestrogen receptor (ER)-positive and human epidermal growth factor receptor 2 (HER2)-negative early breast cancer, a significant proportion of patients are categorised as intermediate risk, based on classic clinico-pathological variables, thus providing limited information to guide adjuvant chemotherapy decisions. Prognostic risk profiling is an integrated part of modern breast cancer diagnostics to provide additional risk information for this patient group. Among the established prognostic assays based on gene expression, the Prosigna assay is widely used and provides an individual risk of recurrence (ROR) score and identifies intrinsic subtypes with associated survival outcomes. Pioneering artificial intelligence-based precision diagnostics, Stratipath Breast, is a deep learning-based image analysis tool that utilises digitised histopathological whole slide images to stratify intermediate risk patients in terms of risk of recurrence. Materials and methods: The study included 234 invasive breast tumours from patients with early ER-positive HER2-negative breast cancer, clinically assessed as intermediate risk tumours and eligible for chemotherapy. All tumours had therefore previously been analysed by the Prosigna assay in clinical routine at point of diagnosis between 2020 and 2022 at the Karolinska University Hospital and Södersjukhuset, Stockholm, Sweden. Clinicopathological data including Prosigna results (ROR score, risk group and intrinsic subtype) were extracted from medical records, along with the corresponding archived haematoxylin and eosin (HE)-stained formalin-fixed paraffin-embedded tissue slides. The HE slides were subsequently digitised and analysed by the Stratipath Breast tool. The agreement between the two tests for risk stratification was evaluated in this real-world breast cancer case series. Results: The Prosigna assay classified 76 (32.5%), 110 (47.0%) and 48 (20.5%) tumours as low, intermediate and high risk, respectively. The Stratipath Breast analysis identified 116 (49.6%) tumours as low risk and 118 (50.4%) as high risk. Among Prosigna intermediate risk tumours, 52 (47.3%) were stratified as low risk and 58 (52.7%) as high risk by Stratipath Breast. The overall agreement between the two tests for low risk and high risk groups was 71.0%, with a Cohen’s linear kappa of 0.42. Twelve of the 48 Prosigna high risk cases were classified as Stratipath low risk. ROR scores were higher in the Stratipath high risk group compared to the low risk group (p < 0.001), across all cases as well as in the Prosigna intermediate group. Among the 176 histological grade (NHG) 2 tumours, 97 (55.1%) and 79 (44.9%) were stratified as Stratipath low risk and high risk, respectively, whereas 66 (37.5%), 83 (47.2%) and 27 (15.3%) were stratified as Prosigna low, intermediate and high risk, respectively. The majority of NHG1 (10 of 12) and NHG3 (37 of 46) tumours were stratified as Stratipath low risk and high risk, respectively. For both risk profiling tests, NHG and Ki67 proliferation index differed between risk groups. Conclusions: In this study of clinically assessed intermediate risk ER-positive HER2-negative breast cancer, we observed a moderate agreement between Prosigna and Stratipath Breast for low risk and high risk groups. In addition, image-based risk profiling stratified more of the NHG2 tumours as high risk. Citation Format: Stephanie Robertson, Yinxi Wang, Wenwen Sun, Emelie Karlsson, Sandy Kang Lövgren, Mattias Rantalainen, Johan Hartman. Clinical evaluation of image-based risk profiling in breast cancer histopathology and comparison to an established gene expression assay [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-07-05.

Abstract 2319: Deep learning-enabled dynamic infiltration and response to NK therapies in solid tumor organoids

Abstract Patient-derived organoid co-culture models have become a state of the art platform to evaluate the effectiveness of immunotherapeutic agents in preclinical studies. However, preclinical development as well as quantification of therapeutic potency and resistance require differential labeling of cells with fluorescent dyes or transgenic fluorescent proteins, which may dysregulate the effector cell function or produce undesired cytotoxicity. Deep learning-based image analysis allows the use of brightfield images to co-localize tumor organoids (TOs) with effector cells, and measure TO-specific responses to novel candidate immunotherapies in a label-free manner; this can accelerate their evaluation as therapeutic candidates in cancer patients and shed light on tumor-immune interaction mechanisms. Here we record multi-day time-lapse confocal microscopy images of TOs co-cultured with NK cells at increasing concentrations of target to effector cells. Two pre-trained U-Net convolutional neural networks are used to separately segment tumor and immune cells from the brightfield channel, and quantify immune cell infiltration over time. TO segmentation masks are registered with a vital dye (caspase 3/7) channel to selectively quantify TO cell death and correlate it with estimated immune cell infiltration across a large cohort of patient-derived TO models (including breast, colorectal, endometrial, gastric, head and neck, liver, lung and pancreatic cancer), at increasing concentrations of 6 different NK cell types (identified as A through F). We find that the estimated density of infiltrating immune cells is highly correlated with TO death, as quantified by fluorescence intensity of caspase over time (median Pearson’s r=0.89, P<0.001). Higher ratios of effector to target cells lead to higher activation of NK cells as measured by infiltrating cell density. Differential infiltration dynamics are observed across TOs and immune cell lines, as peak infiltration density is affected by co-culture time, ratio of target to effector cells, TO line, and NK cell type. These findings highlight that the paired segmentation models are able to measure varying degrees of infiltration across different types of immune cells and TO models. The approach described here is highly scalable and only requires capturing brightfield images of the assay, therefore eliminating the need to label cells and avoiding phototoxic effects or alteration of the immune cells activity. Time series measurements enable quantification of patterns of immune cell activation, including infiltration, migration and co-localization dynamics, which provide insights into the pharmacokinetics and mechanisms for specific immune therapies. Overall, this methodology enables high throughput screening of many therapeutic candidates across dozens to hundreds of unique TO-models, thereby facilitating targeted precision therapy. Citation Format: Luca Lonini, Sonal Khare, Timothy Don Lopez, Stanislaw Szydlo, Michael Streit, Olga A. Karginova, Mary K. Flaherty, Ameen Salhaudeen, Andre Kunert, Anna Oja, Kate Sasser, Martin C. Stumpe, Chi-Sing Ho. Deep learning-enabled dynamic infiltration and response to NK therapies in solid tumor organoids [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2319.

Abstract 3650: Deep learning-based image cytometry and co-localization index in tumor immune microenvironment

Abstract Background: In pathology, digitizing tissue slides has prompted a remarkable development in image analysis using deep learning. This technological advancement is anticipated to aid in pathological diagnosis and to enhance patient management. Deep learning-based image cytometry (DL-IC) enables accurate cell identification and counting and the acquisition of vast amounts of location information from tissue slides. DL-IC can capture information about the diverse and complex tumor immune microenvironment(TIME) and its constituent cells and help identify biomarkers to predict patient treatment efficacy and prognosis. This study will introduce a spatial interaction map and co-localization index (CLI) for the analysis of TIME using DL-IC. Materials and Methods: Cu-Cyto, a deep learning-based image analysis technology, was used in this study; bit-pattern kernel filtering technology, which can accurately count cells while avoiding the determination of multiple cell counts, was used by Cu-Cyto (Abe T, et al. Anticancer Res. 43:3755, 2023). First, the accuracy of cell counting using Cu-Cyto was evaluated. Second, tumor tissue slides with immunohistochemical (IHC) and hematoxylin-eosin (H&E) staining were prepared from surgical specimens of patients with rectal cancer who had undergone neoadjuvant chemoradiotherapy (NACRT), and the relationship between the co-localization index (CLI) of cancer cells and CD8+T cells and prognosis was investigated. CLI was defined to predict cell- cell interactions on the basis of the relative distances between different cell types (Nagasaka T. PCT/JP 2021/021455). Results: The performances of three versions of Cu-Cyto were evaluated according to their learning stages. In the early stage of learning, the F1 score for immunostained CD8+ T cells (0.343) was higher than that for non-immunostained cells (adenocarcinoma cells [0.040] and lymphocytes [0.002]). In the latest stage of learning, the F1 scores for adenocarcinoma cells, lymphocytes, and CD8+ T cells were 0.589, 0.889, and 0.911, respectively. Next, we examined the correlation of CLI between cancer cells and CD8+ T cells with prognosis: patients with a higher CLI significantly prolonged five-year disease-free survival (P=0.038), while there was no substantial difference in five-year overall survival (P=0.57). Conclusion]: Cu-Cyto performed well in cell determination. In particular, IHC was able to increase the learning efficiencies in the early stages of learning. The CLI calculated using Cu-Cyto ts an objective, reproducible, and innovative quantitative approach for assessing cell-cell interactions, which has been shown to be associated with recurrence-free survival in patients with rectal cancer after NACRT. Its performance is expected to improve even further with continuous learning, and the DL-IC can contribute to the implementation of precision oncology. Citation Format: Tomoki Abe, Kimihiro Yamashita, Toru Nagasaka, Tomosuke Mukoyama, Souichirou Miyake, Yasuhiro Ueda, Masayuki Ando, Yuki Okazoe, Takao Tsuneki, Yukari Adachi, Ryunosuke Konaka, Ryuichiro Sawada, Hironobu Goto, Hiroshi Hasegawa, Shingo Kanaji, Takeru Matsuda, Takumi Fukumoto, Yoshihiro Kakeji. Deep learning-based image cytometry and co-localization index in tumor immune microenvironment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3650.

Analysis of Osteosarcoma Cell Lines and Patient Tissue Using a 3D In Vivo Tumor Model—Possible Effects of Punicalagin

Osteosarcomas are the most common primary malignant bone tumors and mostly affect children, adolescents, and young adults. Despite current treatment options such as surgery and polychemotherapy, the survival of patients with metastatic disease remains poor. In recent studies, punicalagin has reduced the cell viability, angiogenesis, and invasion in cell culture trials. The aim of this study was to examine the effects of punicalagin on osteosarcomas in a 3D in vivo tumor model. Human osteosarcoma biopsies and SaOs-2 and MG-63 cells, were grown in a 3D in vivo chorioallantoic membrane (CAM) model. After a cultivation period of up to 72 h, the tumors received daily treatment with punicalagin for 4 days. Weight measurements of the CAM tumors were performed, and laser speckle contrast imaging (LSCI) and a deep learning-based image analysis software (CAM Assay Application v.3.1.0) were used to measure angiogenesis. HE, Ki-67, and Caspase-3 staining was performed after explantation. The osteosarcoma cell lines SaOs-2 and MG-63 and osteosarcoma patient tissue displayed satisfactory growth patterns on the CAM. Treatment with punicalagin decreased tumor weight, proliferation, and tumor-induced angiogenesis, and the tumor tissue showed pro-apoptotic characteristics. These results provide a robust foundation for the implementation of further studies and show that punicalagin offers a promising supplementary treatment option for osteosarcoma patients. The 3D in vivo tumor model represents a beneficial model for the testing of anti-cancer therapies.

Direct Observation and Automated Measurement of Stomatal Responses to Pseudomonas syringae pv. tomato DC3000 in Arabidopsis thaliana.

Stomata are microscopic pores found in the plant leaf epidermis. Regulation of stomatal aperture is pivotal not only for balancing carbon dioxide uptake for photosynthesis and transpirational water loss but also for restricting bacterial invasion. While plants close stomata upon recognition of microbes, pathogenic bacteria, such as Pseudomonas syringae pv. tomato DC3000 (Pto), reopen the closed stomata to gain access into the leaf interior. In conventional assays for assessing stomatal responses to bacterial invasion, leaf epidermal peels, leaf discs, or detached leaves are floated on bacterial suspension, and then stomata are observed under a microscope followed by manual measurement of stomatal aperture. However, these assays are cumbersome and may not reflect stomatal responses to natural bacterial invasion in a leaf attached to the plant. Recently, a portable imaging device was developed that can observe stomata by pinching a leaf without detaching it from the plant, together with a deep learning-based image analysis pipeline designed to automatically measure stomatal aperture from leaf images captured by the device. Here, building on these technical advances, a new method to assess stomatal responses to bacterial invasion in Arabidopsis thaliana is introduced. This method consists of three simple steps: spray inoculation of Pto mimicking natural infection processes, direct observation of stomata on a leaf of the Pto-inoculated plant using the portable imaging device, and automated measurement of stomatal aperture by the image analysis pipeline. This method was successfully used to demonstrate stomatal closure and reopening during Pto invasion under conditions that closely mimic the natural plant-bacteria interaction.

Applying traffic camera and deep learning-based image analysis to predict PM2.5 concentrations

BackgroundAir pollution has caused a significant burden in terms of mortality and mobility worldwide. However, the current coverage of air quality monitoring networks is still limited. ObjectiveThis study aims to apply a novel approach to convert the existing traffic cameras into sensors measuring particulate matter with a diameter of 2.5 μm or less (PM2.5) so that the coverage of PM2.5 monitoring could be expanded without extra cost. MethodsIn our study, the traffic camera images were collected at a rate of 4 images/h and the corresponding hourly PM2.5 concentration was collected from the reference grade PM2.5 station 3 km away. A customized neural network model was trained to obtain the PM2.5 concentration from images followed by a random forest model to predict the hourly PM2.5 concentration. The saliency maps and the feature importance were utilized to interpret the neural network. ResultsProposed novel approach has a high prediction performance to predict hourly PM2.5 from traffic camera images, with a root mean square error (RMSE) of 0.76 μg/m3 and a coefficient of determination (R2) of 0.98. The saliency map shows neural network focuses on unobstructed far-end road surfaces while the random forest feature importance highlights the first quarter image's significance. The model performance is robust whether weather conditions are controlled or not. ConclusionOur study provided a practical approach to converting the existing traffic cameras into PM2.5 sensors. The deep learning method based on the Resnet architecture in our study can broaden the coverage of PM2.5 monitoring with no additional infrastructure needed.

Deep learning-based image analysis for in situ microscopic imaging of cell culture process

Mammalian cell culture is an important bioprocess that directly affects the quality and yield of biopharmaceuticals. Traditionally, condition monitoring of the operation is based on sampling periodically and off-line analysis, which is labor intensive, time consuming, and causing time delays. In this work, in situ microscope is investigated for on-line real-time monitoring of the culture process of Chinese hamster ovary cells with focus on investigation of deep learning-based Mask R-CNN algorithm for image analysis. The model is trained by 184 images with 183,040 cells using data augmentation methods and transfer learning technique. Mask R-CNN segmented the clustered cells more effectively than the conventional one combining edge detection, intensity thresholding, and advanced watershed method as well as the multi-scale edge detection method. Its Dice score, accuracy, precision, sensitivity, F1 score, specificity, and relative volume difference reach 0.862, 0.945, 0.901, 0.827, 0.862, 0.977, and 0.082, respectively. The evolution of geometrical features of cells were further analyzed, including equivalent diameter, circularity, aspect ratio, and eccentricity. The result demonstrated the great potential of deep learning technology in analysis of on-line images for optimization and control of the cell culture process.

Upregulation of TIM3 and reduced expression of PD-1 on immune cell subsets in advanced prostate cancers

Abstract Introduction/Objective Although most prostate cancers behave in an indolent manner, a small proportion is highly aggressive. Both primary and advanced prostate cancer is widely known as a non-inflamed cancer that is characterized by a paucity of immune infiltration. Methods/Case Report To assess the spatial interplay of more than 30 TIM3, CTLA-4, PD-1/-L1 expressing leukocyte subpopulations in 453 prostate cancers, tissue microarrays were stained with 21 antibodies using our BLEACH&STAIN multiplex fluorescence immunohistochemistry approach and analyzed using a deep learning-based image analysis framework. Results (if a Case Study enter NA) The immune cell density of CD8+ cytotoxic T-cells, CD4+ T-helper cells, FOXP3+ regulatory T-cells, M1/ M2 macrophages, as well as CD11c+ dendritic cells increased consistently along with the Gleason grade in primary prostate cancer (p≤ 0.034 each). In recurrent prostate cancers under therapy, the density of FOXP3+regulatory T-cells and M1 macrophages further increased, while the density of CD8+ cytotoxic T-cells, CD4+ T-helper cells, as well as CD11c+ dendritic cells decreased (p≤0.017 each). Although the immune checkpoint expression of TIM3 on T-cell subsets, macrophages and dendritic cells was upregulated in advanced/ recurrent tumors, the expression level of PD-1 was downregulated in all analyzed T-cell subsets. Conclusion Although prostate cancer is a generally considered a low inflamed tumor, the degree of tumor infiltration by various immune cell subtypes increases markedly along with tumor progression and in recurrent tumors under therapy. Taken together, these data suggest that the evaluation of the spatial distribution of immune cell types along with their immune checkpoint expression can provide relevant clinical information in prostate cancer.

Deep learning-based image analysis identifies a DAT-negative subpopulation of dopaminergic neurons in the lateral Substantia nigra

Here we present a deep learning-based image analysis platform (DLAP), tailored to autonomously quantify cell numbers, and fluorescence signals within cellular compartments, derived from RNAscope or immunohistochemistry. We utilised DLAP to analyse subtypes of tyrosine hydroxylase (TH)-positive dopaminergic midbrain neurons in mouse and human brain-sections. These neurons modulate complex behaviour, and are differentially affected in Parkinson’s and other diseases. DLAP allows the analysis of large cell numbers, and facilitates the identification of small cellular subpopulations. Using DLAP, we identified a small subpopulation of TH-positive neurons (~5%), mainly located in the very lateral Substantia nigra (SN), that was immunofluorescence-negative for the plasmalemmal dopamine transporter (DAT), with ~40% smaller cell bodies. These neurons were negative for aldehyde dehydrogenase 1A1, with a lower co-expression rate for dopamine-D2-autoreceptors, but a ~7-fold higher likelihood of calbindin-d28k co-expression (~70%). These results have important implications, as DAT is crucial for dopamine signalling, and is commonly used as a marker for dopaminergic SN neurons.

Assessment of deep learning-based image analysis for disaster waste identification

Accurately predicting different types and quantities of wastes is crucial for proper waste management and sustainable growth and development. High levels of disaster waste (DW) can negatively impact emergency responses, public health, and recovery and rebuilding processes, making the effective management of DW highly important. Most existing waste identification methods cannot suitably identify DW since it is highly subjective in form. In this study, we examined the applicability of different deep learning-based image recognition techniques for DW recognition. A DW image dataset was collected and categorized into normal and hard cases. Then, experiments were conducted using these three network models: DeepLabV3+, ResNeSt-50, and ResNeSt-101. The experimental results indicated that the mean intersection over union (mIoU) of the DeepLabV3+ model was 0.802 for the normal case and that of the ResNeSt-101 model was 0.676 for the hard case, indicating favorable performances. Overall, the deep learning-based image analysis method is more robust than existing methods, particularly in terms of its ability to accurately identify waste in various forms, such as mixed or piled up waste.

Deep learning-assisted high-content screening identifies isoliquiritigenin as an inhibitor of DNA double-strand breaks for preventing doxorubicin-induced cardiotoxicity

BackgroundAnthracyclines including doxorubicin are essential components of many cancer chemotherapy regimens, but their cardiotoxicity severely limits their use. New strategies for treating anthracycline-induced cardiotoxicity (AIC) are still needed. Anthracycline-induced DNA double-strand break (DSB) is the major cause of its cardiotoxicity. However, DSB-based drug screening for AIC has not been performed possibly due to the limited throughput of common assays for detecting DSB. To discover new therapeutic candidates for AIC, here we established a method to rapidly visualize and accurately evaluate the intranuclear anthracycline-induced DSB, and performed a screening for DSB inhibitors.ResultsFirst, we constructed a cardiomyocyte cell line stably expressing EGFP-53BP1, in which the formation of EGFP-53BP1 foci faithfully marked the doxorubicin-induced DSB, providing a faster and visible approach to detecting DSB. To quantify the DSB, we used a deep learning-based image analysis method, which showed the better ability to distinguish different cell populations undergoing different treatments of doxorubicin or reference compounds, compared with the traditional threshold-based method. Subsequently, we applied the deep learning-assisted high-content screening method to 315 compounds and found three compounds (kaempferol, kaempferide, and isoliquiritigenin) that exert cardioprotective effects in vitro. Among them, the protective effect of isoliquiritigenin is accompanied by the up-regulation of HO-1, down-regulation of peroxynitrite and topo II, and the alleviation of doxorubicin-induced DSB and apoptosis. The results of animal experiments also showed that isoliquiritigenin maintained the myocardial tissue structure and cardiac function in vivo. Moreover, isoliquiritigenin did not affect the killing of HeLa and MDA-MB-436 cancer cells by doxorubicin and thus has the potential to be a lead compound to exert cardioprotective effects without affecting the antitumor effect of doxorubicin.ConclusionsOur findings provided a new method for the drug discovery for AIC, which combines phenotypic screening with artificial intelligence. The results suggested that isoliquiritigenin as an inhibitor of DSB may be a promising drug candidate for AIC.

Prediction of lymph node metastasis in primary gastric cancer from pathological images and clinical data by multimodal multiscale deep learning

Background:Gastric cancer is the third most common cause of cancer-related death. Accurate preoperative prediction of lymph node metastasis (LNM) in primary gastric cancer strongly influences the choice of surgical approach and the prognosis of gastric cancer patients. Objective:To develop and validate a deep learning-based model to analyze routine histological slides of patients with primary gastric cancer as well as clinical data to predict the occurrence of LNM preoperatively. Patients and Methods:Radical surgery slides from 309 patients and biopsy slides from 157 patients were collected from Fujian Cancer Hospital, along with radical surgery slides from 306 patients from The Cancer Genome Atlas (TCGA). Clinical data, including age, gender, lauren classification, and tumor location, were collected. These datasets were used to develop and validate a deep learning-based model. Results:Five models were trained via cross-validation, with a mean area under the receiver operating characteristic curve (AUC) (standard deviation [SD]) of 0.877 (0.048) achieved. There was a significant difference in scores between both classes (LNM positive [N+] and LNM negative [N0]) ( p<0.001). we validated the performance of the model on biopsy slides and achieved a mean AUC (SD) of 0.725 (0.020). In the analysis of clinical data, the lauren classification was demonstrated to be an independent risk factor for predicting LNM. Conclusion:Our study confirmed that deep learning-based image analysis could preoperatively predict LNM in patients with primary gastric cancer, combining histological slides at different magnification scales and relevant clinical data, showing superiority over individual modality prediction.