The Cell Tracking Challenge: 10 years of objective benchmarking.

The Cell Tracking Challenge is an ongoing benchmarking initiative that has become a reference in cell segmentation and tracking algorithm development. Here, we present a significant number of improvements introduced in the challenge since our 2017 report. These include the creation of a new segmentation-only benchmark, the enrichment of the dataset repository with new datasets that increase its diversity and complexity, and the creation of a silver standard reference corpus based on the most competitive results, which will be of particular interest for data-hungry deep learning-based strategies. Furthermore, we present the up-to-date cell segmentation and tracking leaderboards, an in-depth analysis of the relationship between the performance of the state-of-the-art methods and the properties of the datasets and annotations, and two novel, insightful studies about the generalizability and the reusability of top-performing methods. These studies provide critical practical conclusions for both developers and users of traditional and machine learning-based cell segmentation and tracking algorithms.

Automatic Segmentation and Quantification of Nigrosome-1 Neuromelanin and Iron in MRI: A Candidate Biomarker for Parkinson's Disease.

BACKGROUND: There is a lack of automated tools for the segmentation and quantification of neuromelanin (NM) and iron in the nigrosome-1 (N1). Existing tools evaluate the N1 sign, i.e., the presence or absence of the "swallow-tail" in iron-sensitive MRI, or globally analyze the MRI signal in an area containing the N1, without providing a volumetric delineation. PURPOSE: Present an automated method to segment the N1 and quantify differences in N1's NM and iron content between Parkinson's disease (PD) patients and healthy controls (HCs). Study whether N1 degeneration is clinically related to PD and could be used as a biomarker of the disease. STUDY TYPE: Prospective. SUBJECTS: Seventy-one PD (65.3 ± 10.3 years old, 34 female/37 male); 30 HC (62.7 ± 7.8 years old, 17 female/13 male). FIELD STRENGTH/SEQUENCE: 3 T Anatomical T1-weighted MPRAGE, NM-MRI T1-weighted gradient with magnetization transfer, susceptibility-weighted imaging (SWI). ASSESSMENT: N1 was automatically segmented in SWI images using a multi-image atlas, populated with healthy N1 structures manually annotated by a neurologist. Relative NM and iron content were quantified and their diagnostic performance assessed and compared with the substantia nigra pars compacta (SNc). The association between image parameters and clinically relevant variables was studied. STATISTICAL TESTS: Nonparametric tests were used (Mann-Whitney's U, chi-square, and Friedman tests) at P = 0.05. RESULTS: N1's relative NM content decreased and relative iron content increased in PD patients compared with HCs (NM-CRHC = 22.55 ± 1.49; NM-CRPD = 19.79 ± 1.92; NM-nVolHC = 2.69 × 10-5 ± 1.02 × 10-5 ; NM-nVolPD = 1.18 × 10-5 ± 0.96 × 10-5 ; Iron-CRHC = 10.51 ± 2.64; Iron-CRPD = 19.35 ± 7.88; Iron-nVolHC = 0.72 × 10-5 ± 0.81 × 10-5 ; Iron-nVolPD = 2.82 × 10-5 ± 2.04 × 10-5 ). Binary logistic regression analyses combining N1 and SNc image parameters yielded a top AUC = 0.955. Significant correlation was found between most N1 parameters and both disease duration (ρNM-CR = -0.31; ρiron-CR = 0.43; ρiron-nVol = 0.46) and the motor status (ρNM-nVol = -0.27; ρiron-CR = 0.33; ρiron-nVol = 0.28), suggesting NM reduction along with iron accumulation in N1 as the disease progresses. DATA CONCLUSION: This method provides a fully automatic N1 segmentation, and the analyses performed reveal that N1 relative NM and iron quantification improves diagnostic performance and suggest a relative NM reduction along with a relative iron accumulation in N1 as the disease progresses. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 1.

Brain ventricular enlargement in human and murine acute intermittent porphyria.

The morphological changes that occur in the central nervous system of patients with severe acute intermittent porphyria (AIP) have not yet been clearly established. The aim of this work was to analyze brain involvement in patients with severe AIP without epileptic seizures or clinical posterior reversible encephalopathy syndrome, as well as in a mouse model receiving or not liver-directed gene therapy aimed at correcting the metabolic disorder. We conducted neuroradiologic studies in 8 severely affected patients (6 women) and 16 gender- and age-matched controls. Seven patients showed significant enlargement of the cerebral ventricles and decreased brain perfusion was observed during the acute attack in two patients in whom perfusion imaging data were acquired. AIP mice exhibited reduced cerebral blood flow and developed chronic dilatation of the cerebral ventricles even in the presence of slightly increased porphyrin precursors. While repeated phenobarbital-induced attacks exacerbated ventricular dilation in AIP mice, correction of the metabolic defect using liver-directed gene therapy restored brain perfusion and afforded protection against ventricular enlargement. Histological studies revealed no signs of neuronal loss but a denser neurofilament pattern in the periventricular areas, suggesting compression probably caused by imbalance in cerebrospinal fluid dynamics. In conclusion, severely affected AIP patients exhibit cerebral ventricular enlargement. Liver-directed gene therapy protected against the morphological consequences of the disease seen in the brain of AIP mice. The observational study was registered at Clinicaltrial.gov as NCT02076763.

Heterogenous presence of neutrophil extracellular traps in human solid tumours is partially dependent on IL-8.

Neutrophil extracellular traps (NETs) are webs of extracellular nuclear DNA extruded by dying neutrophils infiltrating tissue. NETs constitute a defence mechanism to entrap and kill fungi and bacteria. Tumours induce the formation of NETs to the advantage of the malignancy via a variety of mechanisms shown in mouse models. Here, we investigated the presence of NETs in a variety of human solid tumours and their association with IL-8 (CXCL8) protein expression and CD8+ T-cell density in the tumour microenvironment. Multiplex immunofluorescence panels were developed to identify NETs in human cancer tissues by co-staining with the granulocyte marker CD15, the neutrophil marker myeloperoxidase and citrullinated histone H3 (H3Cit), as well as IL-8 protein and CD8+ T cells. Three ELISA methods to detect and quantify circulating NETs in serum were optimised and utilised. Whole tumour sections and tissue microarrays from patients with non-small cell lung cancer (NSCLC; n = 14), bladder cancer (n = 14), melanoma (n = 11), breast cancer (n = 31), colorectal cancer (n = 20) and mesothelioma (n = 61) were studied. Also, serum samples collected retrospectively from patients with metastatic melanoma (n = 12) and NSCLC (n = 34) were ELISA assayed to quantify circulating NETs and IL-8. NETs were detected in six different human cancer types with wide individual variation in terms of tissue density and distribution. At least in NSCLC, bladder cancer and metastatic melanoma, NET density positively correlated with IL-8 protein expression and inversely correlated with CD8+ T-cell densities. In a series of serum samples from melanoma and NSCLC patients, a positive correlation between circulating NETs and IL-8 was found. In conclusion, NETs are detectable in formalin-fixed human biopsy samples from solid tumours and in the circulation of cancer patients with a considerable degree of individual variation. NETs show a positive association with IL-8 and a trend towards a negative association with CD8+ tumour-infiltrating lymphocytes. © 2021 The Authors. The Journal of Pathology published by John Wiley & Sons, Ltd. on behalf of The Pathological Society of Great Britain and Ireland.

NMF-RI: blind spectral unmixing of highly mixed multispectral flow and image cytometry data.

MOTIVATION: Recent advances in multiplex immunostaining and multispectral cytometry have opened the door to simultaneously visualizing an unprecedented number of biomarkers both in liquid and solid samples. Properly unmixing fluorescent emissions is a challenging task, which normally requires the characterization of the individual fluorochromes from control samples. As the number of fluorochromes increases, the cost in time and use of reagents becomes prohibitively high. Here, we present a fully unsupervised blind spectral unmixing method for the separation of fluorescent emissions in highly mixed spectral data, without the need for control samples. To this end, we extend an existing method based on non-negative Matrix Factorization, and introduce several critical improvements: initialization based on the theoretical spectra, automated selection of 'sparse' data and use of a re-initialized multilayer optimizer. RESULTS: Our algorithm is exhaustively tested using synthetic data to study its robustness against different levels of colocalization, signal to noise ratio, spectral resolution and the effect of errors in the initialization of the algorithm. Then, we compare the performance of our method to that of traditional spectral unmixing algorithms using novel multispectral flow and image cytometry systems. In all cases, we show that our blind unmixing algorithm performs robust unmixing of highly spatially and spectrally mixed data with an unprecedently low computational cost. In summary, we present the first use of a blind unmixing method in multispectral flow and image cytometry, opening the door to the widespread use of our method to efficiently pre-process multiplex immunostaining samples without the need of experimental controls. AVAILABILITY AND IMPLEMENTATION: https://github.com/djimenezsanchez/Blind_Unmixing_NMF_RI/ contains the source code and all datasets used in this manuscript. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

An objective comparison of cell-tracking algorithms.

We present a combined report on the results of three editions of the Cell Tracking Challenge, an ongoing initiative aimed at promoting the development and objective evaluation of cell segmentation and tracking algorithms. With 21 participating algorithms and a data repository consisting of 13 data sets from various microscopy modalities, the challenge displays today's state-of-the-art methodology in the field. We analyzed the challenge results using performance measures for segmentation and tracking that rank all participating methods. We also analyzed the performance of all of the algorithms in terms of biological measures and practical usability. Although some methods scored high in all technical aspects, none obtained fully correct solutions. We found that methods that either take prior information into account using learning strategies or analyze cells in a global spatiotemporal video context performed better than other methods under the segmentation and tracking scenarios included in the challenge.

Objective comparison of particle tracking methods.

Particle tracking is of key importance for quantitative analysis of intracellular dynamic processes from time-lapse microscopy image data. Because manually detecting and following large numbers of individual particles is not feasible, automated computational methods have been developed for these tasks by many groups. Aiming to perform an objective comparison of methods, we gathered the community and organized an open competition in which participating teams applied their own methods independently to a commonly defined data set including diverse scenarios. Performance was assessed using commonly defined measures. Although no single method performed best across all scenarios, the results revealed clear differences between the various approaches, leading to notable practical conclusions for users and developers.

Complementary Effects of Interleukin-15 and Alpha Interferon Induce Immunity in Hepatitis B Virus Transgenic Mice.

UNLABELLED: In chronic hepatitis B (CHB), failure to control hepatitis B virus (HBV) is associated with T cell dysfunction. HBV transgenic mice mirror many features of the human disease, including T cell unresponsiveness, and thus represent an appropriate model in which to test novel therapeutic strategies. To date, the tolerant state of CD8(+) T cells in these animals could be altered only by strong immunogens or by immunization with HBV antigen-pulsed dendritic cells; however, the effectors induced were unable to suppress viral gene expression or replication. Because of the known stimulatory properties of alpha interferon (IFN-α) and interleukin-15 (IL-15), this study explored the therapeutic potential of liver-directed gene transfer of these cytokines in a murine model of CHB using adeno-associated virus (AAV) delivery. This combination not only resulted in a reduction in the viral load in the liver and the induction of an antibody response but also gave rise to functional and specific CD8(+) immunity. Furthermore, when splenic and intrahepatic lymphocytes from IFN-α- and IL-15-treated animals were transferred to new HBV carriers, partial antiviral immunity was achieved. In contrast to previous observations made using either cytokine alone, markedly attenuated PD-L1 induction in hepatic tissue was observed upon coadministration. An initial study with CHB patient samples also gave promising results. Hence, we demonstrated synergy between two stimulating cytokines, IL-15 and IFN-α, which, given together, constitute a potent approach to significantly enhance the CD8(+) T cell response in a state of immune hyporesponsiveness. Such an approach may be useful for treating chronic viral infections and neoplastic conditions. IMPORTANCE: With 350 million people affected worldwide and 600,000 annual deaths due to HBV-induced liver cirrhosis and/or hepatocellular carcinoma, chronic hepatitis B (CHB) is a major health problem. However, current treatment options are costly and not very effective and/or need to be administered for life. The unprecedented efficacy of the strategy described in our paper may offer an alternative and is relevant for a broad spectrum of readers because of its clear translational importance to other chronic viral infections in which a hyporesponsive antigen-specific T cell repertoire prevents clearance of the pathogen.

A benchmark for comparison of cell tracking algorithms.

MOTIVATION: Automatic tracking of cells in multidimensional time-lapse fluorescence microscopy is an important task in many biomedical applications. A novel framework for objective evaluation of cell tracking algorithms has been established under the auspices of the IEEE International Symposium on Biomedical Imaging 2013 Cell Tracking Challenge. In this article, we present the logistics, datasets, methods and results of the challenge and lay down the principles for future uses of this benchmark. RESULTS: The main contributions of the challenge include the creation of a comprehensive video dataset repository and the definition of objective measures for comparison and ranking of the algorithms. With this benchmark, six algorithms covering a variety of segmentation and tracking paradigms have been compared and ranked based on their performance on both synthetic and real datasets. Given the diversity of the datasets, we do not declare a single winner of the challenge. Instead, we present and discuss the results for each individual dataset separately. AVAILABILITY AND IMPLEMENTATION: The challenge Web site (http://www.codesolorzano.com/celltrackingchallenge) provides access to the training and competition datasets, along with the ground truth of the training videos. It also provides access to Windows and Linux executable files of the evaluation software and most of the algorithms that competed in the challenge.

Automated neuromelanin imaging as a diagnostic biomarker for Parkinson's disease.

BACKGROUND: We aimed to analyze the diagnostic accuracy of an automated segmentation and quantification method of the SNc and locus coeruleus (LC) volumes based on neuromelanin (NM)-sensitive MRI (NM-MRI) in patients with idiopathic (iPD) and monogenic (iPD) Parkinson's disease (PD). METHODS: Thirty-six patients (23 idiopathic and 13 monogenic PARKIN or LRRK2 mutations) and 37 age-matched healthy controls underwent 3T-NM-MRI. SNc and LC volumetry were performed using fully automated multi-image atlas segmentation. The diagnostic performance to differentiate PD from controls was measured using the area under the curve (AUC) and likelihood ratios based on receiver operating characteristic (ROC) analyses. RESULTS: We found a significant reduction of SNc and LC volumes in patients, when compared to controls. ROC analysis showed better diagnostic accuracy when using SNc volume than LC volume. Significant differences between ipsilateral and contralateral SNc volumes, in relation to the more clinically affected side, were found in patients with iPD (P = 0.007). Contralateral atrophy in the SNc showed the highest power to discriminate PD subjects from controls (AUC, 0.93-0.94; sensitivity, 91%-92%; specificity, 89%; positive likelihood ratio: 8.4-8.5; negative likelihood ratio: 0.09-0.1 at a single cut-off point). Interval likelihood ratios for contralateral SNc volume improved the diagnostic accuracy of volumetric measurements. CONCLUSION: SNc and LC volumetry based on NM-MRI resulting from the automated segmentation and quantification technique can yield high diagnostic accuracy for differentiating PD from health and might be an unbiased disease biomarker. © 2015 International Parkinson and Movement Disorder Society.

Spatially resolved tissue imaging to analyze the tumor immune microenvironment: beyond cell-type densities.

INTRODUCTION: The tissue immune microenvironment is associated with key aspects of tumor biology. The interaction between the immune system and cancer cells has predictive and prognostic potential across different tumor types. Spatially resolved tissue-based technologies allowed researchers to simultaneously quantify different immune populations in tumor samples. However, bare quantification fails to harness the spatial nature of tissue-based technologies. Tumor-immune interactions are associated with specific spatial patterns that can be measured. In recent years, several computational tools have been developed to increase our understanding of these spatial patterns. TOPICS COVERED: In this review, we cover standard techniques as well as new advances in the field of spatial analysis of the immune microenvironment. We focused on marker quantification, spatial intratumor heterogeneity analysis, cell‒cell spatial interaction studies and neighborhood analyses.

Multimodal evaluation of drug antibacterial activity reveals cinnamaldehyde analog anti-biofilm effects against Haemophilus influenzae.

Biofilm formation by the pathobiont Haemophilus influenzae is associated with human nasopharynx colonization, otitis media in children, and chronic respiratory infections in adults suffering from chronic respiratory diseases such as chronic obstructive pulmonary disease (COPD). ß-lactam and quinolone antibiotics are commonly used to treat these infections. However, considering the resistance of biofilm-resident bacteria to antibiotic-mediated killing, the use of antibiotics may be insufficient and require being replaced or complemented with novel strategies. Moreover, unlike the standard minimal inhibitory concentration assay used to assess antibacterial activity against planktonic cells, standardization of methods to evaluate anti-biofilm drug activity is limited. In this work, we detail a panel of protocols for systematic analysis of drug antimicrobial effect on bacterial biofilms, customized to evaluate drug effects against H. influenzae biofilms. Testing of two cinnamaldehyde analogs, (E)-trans-2-nonenal and (E)-3-decen-2-one, demonstrated their effectiveness in both H. influenzae inhibition of biofilm formation and eradication or preformed biofilms. Assay complementarity allowed quantifying the dynamics and extent of the inhibitory effects, also observed for ampicillin resistant clinical strains forming biofilms refractory to this antibiotic. Moreover, cinnamaldehyde analog encapsulation into poly(lactic-co-glycolic acid) (PLGA) polymeric nanoparticles allowed drug vehiculization while maintaining efficacy. Overall, we demonstrate the usefulness of cinnamaldehyde analogs against H. influenzae biofilms, present a test panel that can be easily adapted to a wide range of pathogens and drugs, and highlight the benefits of drug nanoencapsulation towards safe controlled release.

Synplex: In Silico Modeling of the Tumor Microenvironment From Multiplex Images.

Multiplex immunofluorescence is a novel, high-content imaging technique that allows simultaneous in situ labeling of multiple tissue antigens. This technique is of growing relevance in the study of the tumor microenvironment, and the discovery of biomarkers of disease progression or response to immune-based therapies. Given the number of markers and the potential complexity of the spatial interactions involved, the analysis of these images requires the use of machine learning tools that rely for their training on the availability of large image datasets, extremely laborious to annotate. We present Synplex, a computer simulator of multiplexed immunofluorescence images from user-defined parameters: i. cell phenotypes, defined by the level of expression of markers and morphological parameters; ii. cellular neighborhoods based on the spatial association of cell phenotypes; and iii. interactions between cellular neighborhoods. We validate Synplex by generating synthetic tissues that accurately simulate real cancer cohorts with underlying differences in the composition of their tumor microenvironment and show proof-of-principle examples of how Synplex could be used for data augmentation when training machine learning models, and for the in silico selection of clinically relevant biomarkers. Synplex is publicly available at https://github.com/djimenezsanchez/Synplex.

A hybrid physics-based and data-driven framework for cellular biological systems: Application to the morphogenesis of organoids.

How cells orchestrate their cellular functions remains a crucial question to unravel how they organize in different patterns. We present a framework based on artificial intelligence to advance the understanding of how cell functions are coordinated spatially and temporally in biological systems. It consists of a hybrid physics-based model that integrates both mechanical interactions and cell functions with a data-driven model that regulates the cellular decision-making process through a deep learning algorithm trained on image data metrics. To illustrate our approach, we used data from 3D cultures of murine pancreatic ductal adenocarcinoma cells (PDAC) grown in Matrigel as tumor organoids. Our approach allowed us to find the underlying principles through which cells activate different cell processes to self-organize in different patterns according to the specific microenvironmental conditions. The framework proposed here expands the tools for simulating biological systems at the cellular level, providing a novel perspective to unravel morphogenetic patterns.

Prussian Blue nanoparticles: An FDA-approved substance that may quickly degrade at physiological pH.

Prussian blue (PB) is a coordination polymer based on the Fe2+…CN…Fe3+ sequence. It is an FDA-approved drug, intended for oral use at the acidic pH of the stomach and of most of the intestine track. However, based on FDA approval, a huge number of papers proposed the use of PB nanoparticles (PBnp) under "physiological conditions", meaning pH buffered at 7.4 and high saline concentration. While most of these papers report that PBnp are stable at this pH, a small number of papers describes instead PBnp degradation at the same or similar pH values, i.e. in the 7-8 range. Here we give a definitively clear picture: PBnp are intrinsically unstable at pH ≥ 7, degrading with the fast disappearance of their 700 nm absorption band, due to the formation of OH- complexes from the labile Fe3+ centers. However, we show also that the presence of a polymeric coating (PVP) can protect PBnp at pH 7.4 for over 24 h. Moreover, we demonstrate that when "physiological conditions" include serum, a protein corona is rapidly formed on PBnp, efficiently avoiding degradation. We also show that the viability of PBnp-treated EA.hy926, NCI-H1299, and A549 cells is not affected in a wide range of conditions that either prevent or promote PBnp degradation.

Weakly supervised deep learning to predict recurrence in low-grade endometrial cancer from multiplexed immunofluorescence images.

Predicting recurrence in low-grade, early-stage endometrial cancer (EC) is both challenging and clinically relevant. We present a weakly-supervised deep learning framework, NaroNet, that can learn, without manual expert annotation, the complex tumor-immune interrelations at three levels: local phenotypes, cellular neighborhoods, and tissue areas. It uses multiplexed immunofluorescence for the simultaneous visualization and quantification of CD68 + macrophages, CD8 + T cells, FOXP3 + regulatory T cells, PD-L1/PD-1 protein expression, and tumor cells. We used 489 tumor cores from 250 patients to train a multilevel deep-learning model to predict tumor recurrence. Using a tenfold cross-validation strategy, our model achieved an area under the curve of 0.90 with a 95% confidence interval of 0.83-0.95. Our model predictions resulted in concordance for 96,8% of cases (κ = 0.88). This method could accurately assess the risk of recurrence in EC, outperforming current prognostic factors, including molecular subtyping.

In vitro modeling of polyclonal infection dynamics within the human airways by Haemophilus influenzae differential fluorescent labeling.

IMPORTANCE: Genomic diversity of nontypeable H. influenzae strains confers phenotypic heterogeneity. Multiple strains of H. influenzae can be simultaneously isolated from clinical specimens, but we lack detailed information about polyclonal infection dynamics by this pathogen. A long-term barrier to our understanding of this host-pathogen interplay is the lack of genetic tools for strain engineering and differential labeling. Here, we present a novel plasmid toolkit named pTBH (toolbox for Haemophilus), with standardized modules for fluorescent or bioluminescent labeling, adapted to H. influenzae requirements but designed to be versatile so it can be utilized in other bacterial species. We present detailed experimental and quantitative image analysis methods, together with proof-of-principle examples, and show the ample possibilities of 3D microscopy, combined with quantitative image analysis, to model H. influenzae polyclonal infection lifestyles and unravel the co-habitation and co-infection dynamics of this respiratory pathogen.

Adiposoft: automated software for the analysis of white adipose tissue cellularity in histological sections.

The accurate estimation of the number and size of cells provides relevant information on the kinetics of growth and the physiological status of a given tissue or organ. Here, we present Adiposoft, a fully automated open-source software for the analysis of white adipose tissue cellularity in histological sections. First, we describe the sequence of image analysis routines implemented by the program. Then, we evaluate our software by comparing it with other adipose tissue quantification methods, namely, with the manual analysis of cells in histological sections (used as gold standard) and with the automated analysis of cells in suspension, the most commonly used method. Our results show significant concordance between Adiposoft and the other two methods. We also demonstrate the ability of the proposed method to distinguish the cellular composition of three different rat fat depots. Moreover, we found high correlation and low disagreement between Adiposoft and the manual delineation of cells. We conclude that Adiposoft provides accurate results while considerably reducing the amount of time and effort required for the analysis.

NaroNet: Discovery of tumor microenvironment elements from highly multiplexed images.

Understanding the spatial interactions between the elements of the tumor microenvironment -i.e. tumor cells. fibroblasts, immune cells- and how these interactions relate to the diagnosis or prognosis of a tumor is one of the goals of computational pathology. We present NaroNet, a deep learning framework that models the multi-scale tumor microenvironment from multiplex-stained cancer tissue images and provides patient-level interpretable predictions using a seamless end-to-end learning pipeline. Trained only with multiplex-stained tissue images and their corresponding patient-level clinical labels, NaroNet unsupervisedly learns which cell phenotypes, cell neighborhoods, and neighborhood interactions have the highest influence to predict the correct label. To this end, NaroNet incorporates several novel and state-of-the-art deep learning techniques, such as patch-level contrastive learning, multi-level graph embeddings, a novel max-sum pooling operation, or a metric that quantifies the relevance that each microenvironment element has in the individual predictions. We validate NaroNet using synthetic data simulating multiplex-immunostained images where a patient label is artificially associated to the -adjustable- probabilistic incidence of different microenvironment elements. We then apply our model to two sets of images of human cancer tissues: 336 seven-color multiplex-immunostained images from 12 high-grade endometrial cancer patients; and 382 35-plex mass cytometry images from 215 breast cancer patients. In both synthetic and real datasets, NaroNet provides outstanding predictions of relevant clinical information while associating those predictions to the presence of specific microenvironment elements.