RESUMEN
Crohn's disease (CD) is a complex chronic inflammatory disorder with both gastrointestinal and extra-intestinal manifestations associated immune dysregulation. Analyzing 202,359 cells from 170 specimens across 83 patients, we identify a distinct epithelial cell type in both terminal ileum and ascending colon (hereon as 'LND') with high expression of LCN2, NOS2, and DUOX2 and genes related to antimicrobial response and immunoregulation. LND cells, confirmed by in-situ RNA and protein imaging, are rare in non-IBD controls but expand in active CD, and actively interact with immune cells and specifically express IBD/CD susceptibility genes, suggesting a possible function in CD immunopathogenesis. Furthermore, we discover early and late LND subpopulations with different origins and developmental potential. A higher ratio of late-to-early LND cells correlates with better response to anti-TNF treatment. Our findings thus suggest a potential pathogenic role for LND cells in both Crohn's ileitis and colitis.
Asunto(s)
Colon , Enfermedad de Crohn , Oxidasas Duales , Células Epiteliales , Íleon , Lipocalina 2 , Enfermedad de Crohn/patología , Enfermedad de Crohn/genética , Enfermedad de Crohn/inmunología , Humanos , Células Epiteliales/metabolismo , Células Epiteliales/patología , Colon/patología , Íleon/patología , Lipocalina 2/metabolismo , Lipocalina 2/genética , Oxidasas Duales/genética , Oxidasas Duales/metabolismo , Masculino , Óxido Nítrico Sintasa de Tipo II/metabolismo , Óxido Nítrico Sintasa de Tipo II/genética , Femenino , Adulto , Factor de Necrosis Tumoral alfa/metabolismo , Mucosa Intestinal/patología , Mucosa Intestinal/metabolismo , Persona de Mediana EdadRESUMEN
Multiplex immunofluorescence (MxIF) is an advanced molecular imaging technique that can simultaneously provide biologists with multiple (i.e., more than 20) molecular markers on a single histological tissue section. Unfortunately, due to imaging restrictions, the more routinely used hematoxylin and eosin (H&E) stain is typically unavailable with MxIF on the same tissue section. As biological H&E staining is not feasible, previous efforts have been made to obtain H&E whole slide image (WSI) from MxIF via deep learning empowered virtual staining. However, the tiling effect is a long-lasting problem in high-resolution WSI-wise synthesis. The MxIF to H&E synthesis is no exception. Limited by computational resources, the cross-stain image synthesis is typically performed at the patch-level. Thus, discontinuous intensities might be visually identified along with the patch boundaries assembling all individual patches back to a WSI. In this work, we propose a deep learning based unpaired high-resolution image synthesis method to obtain virtual H&E WSIs from MxIF WSIs (each with 27 markers/stains) with reduced tiling effects. Briefly, we first extend the CycleGAN framework by adding simultaneous nuclei and mucin segmentation supervision as spatial constraints. Then, we introduce a random walk sliding window shifting strategy during the optimized inference stage, to alleviate the tiling effects. The validation results show that our spatially constrained synthesis method achieves a 56% performance gain for the downstream cell segmentation task. The proposed inference method reduces the tiling effects by using 50% fewer computation resources without compromising performance. The proposed random sliding window inference method is a plug-and-play module, which can be generalized for other high-resolution WSI image synthesis applications. The source code with our proposed model are available at https://github.com/MASILab/RandomWalkSlidingWindow.git.
RESUMEN
MOTIVATION: Multiplexed immunofluorescence (mIF) is an emerging assay for multichannel protein imaging that can decipher cell-level spatial features in tissues. However, existing automated cell phenotyping methods, such as clustering, face challenges in achieving consistency across experiments and often require subjective evaluation. As a result, mIF analyses often revert to marker gating based on manual thresholding of raw imaging data. RESULTS: To address the need for an evaluable semi-automated algorithm, we developed GammaGateR, an R package for interactive marker gating designed specifically for segmented cell-level data from mIF images. Based on a novel closed-form gamma mixture model, GammaGateR provides estimates of marker-positive cell proportions and soft clustering of marker-positive cells. The model incorporates user-specified constraints that provide a consistent but slide-specific model fit. We compared GammaGateR against the newest unsupervised approach for annotating mIF data, employing two colon datasets and one ovarian cancer dataset for the evaluation. We showed that GammaGateR produces highly similar results to a silver standard established through manual annotation. Furthermore, we demonstrated its effectiveness in identifying biological signals, achieved by mapping known spatial interactions between CD68 and MUC5AC cells in the colon and by accurately predicting survival in ovarian cancer patients using the phenotype probabilities as input for machine learning methods. GammaGateR is a highly efficient tool that can improve the replicability of marker gating results, while reducing the time of manual segmentation. AVAILABILITY AND IMPLEMENTATION: The R package is available at https://github.com/JiangmeiRubyXiong/GammaGateR.
Asunto(s)
Algoritmos , Análisis de la Célula Individual , Humanos , Análisis de la Célula Individual/métodos , Programas Informáticos , Procesamiento de Imagen Asistido por Computador/métodos , Femenino , Neoplasias Ováricas/metabolismo , Técnica del Anticuerpo Fluorescente/métodos , Biomarcadores/metabolismoRESUMEN
Analyzing high resolution whole slide images (WSIs) with regard to information across multiple scales poses a significant challenge in digital pathology. Multi-instance learning (MIL) is a common solution for working with high resolution images by classifying bags of objects (i.e. sets of smaller image patches). However, such processing is typically performed at a single scale (e.g., 20× magnification) of WSIs, disregarding the vital inter-scale information that is key to diagnoses by human pathologists. In this study, we propose a novel cross-scale MIL algorithm to explicitly aggregate inter-scale relationships into a single MIL network for pathological image diagnosis. The contribution of this paper is three-fold: (1) A novel cross-scale MIL (CS-MIL) algorithm that integrates the multi-scale information and the inter-scale relationships is proposed; (2) A toy dataset with scale-specific morphological features is created and released to examine and visualize differential cross-scale attention; (3) Superior performance on both in-house and public datasets is demonstrated by our simple cross-scale MIL strategy. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.
Asunto(s)
Algoritmos , Diagnóstico por Imagen , HumanosRESUMEN
The membrane (M) glycoprotein of coronaviruses (CoVs) serves as the nidus for virion assembly. Using a yeast two-hybrid screen, we identified the interaction of the cytosolic tail of Murine Hepatitis Virus (MHV-CoV) M protein with Myosin Vb (MYO5B), specifically with the alternative splice variant of cellular MYO5B including exon D (MYO5B+D), which mediates interaction with Rab10. When co-expressed in human lung epithelial A549 and canine kidney epithelial MDCK cells, MYO5B+D co-localized with the MHV-CoV M protein, as well as with the M proteins from Porcine Epidemic Diarrhea Virus (PEDV-CoV), Middle East Respiratory Syndrome (MERS-CoV) and Severe Acute Respiratory Syndrome 2 (SARS-CoV-2). Co-expressed M proteins and MYO5B+D co-localized with endogenous Rab10 and Rab11a. We identified point mutations in MHV-CoV M that blocked the interaction with MYO5B+D in yeast 2-hybrid assays. One of these point mutations (E121K) was previously shown to block MHV-CoV virion assembly and its interaction with MYO5B+D. The E to K mutation at homologous positions in PEDV-CoV, MERS-CoV and SARS-CoV-2 M proteins also blocked colocalization with MYO5B+D. The knockdown of Rab10 blocked the co-localization of M proteins with MYO5B+D and was rescued by re-expression of CFP-Rab10. Our results suggest that CoV M proteins traffic through Rab10-containing systems, in association with MYO5B+D.
Asunto(s)
Proteínas M de Coronavirus , Animales , Perros , Humanos , Células de Riñón Canino Madin Darby/metabolismo , Células de Riñón Canino Madin Darby/virología , Coronavirus del Síndrome Respiratorio de Oriente Medio , Miosinas , Proteínas de Unión al GTP rab/genética , Saccharomyces cerevisiae , Porcinos , Proteínas de la Matriz Viral , SARS-CoV-2/metabolismo , Virus de la Hepatitis Murina/metabolismo , Células A549/metabolismo , Células A549/virología , Virus de la Diarrea Epidémica Porcina/metabolismoRESUMEN
Although amplifications and mutations in receptor tyrosine kinases (RTKs) act as bona fide oncogenes, in most cancers, RTKs maintain moderate expression and remain wild-type. Consequently, cognate ligands control many facets of tumorigenesis, including resistance to anti-RTK therapies. Herein, we show that the ligands for the RTKs MET and RON, HGF and HGFL, respectively, are synthesized as inactive precursors that are activated by cellular proteases. Our newly generated HGF/HGFL protease inhibitors could overcome both de novo and acquired cetuximab resistance in colorectal cancer (CRC). Conversely, HGF overexpression was necessary and sufficient to induce cetuximab resistance and loss of polarity. Moreover, HGF-induced cetuximab resistance could be overcome by the downstream MET inhibitor, crizotinib, and upstream protease inhibitors. Additionally, HAI-1, an endogenous inhibitor of HGF proteases, (i) was downregulated in CRC, (ii) exhibited increased genomic methylation that correlated with poor prognosis, (iii) HAI-1 expression correlated with cetuximab response in a panel of cancer cell lines, and (iv) exogenous addition of recombinant HAI-1 overcame cetuximab resistance in CC-HGF cells. Thus, we describe a targetable, autocrine HAI-1/Protease/HGF/MET axis in cetuximab resistance in CRC.
Asunto(s)
Neoplasias Colorrectales , Transducción de Señal , Humanos , Cetuximab/farmacología , Proteínas Proto-Oncogénicas c-met/genética , Proteínas Proto-Oncogénicas c-met/metabolismo , Resistencia a Antineoplásicos/genética , Neoplasias Colorrectales/tratamiento farmacológico , Neoplasias Colorrectales/genética , Neoplasias Colorrectales/metabolismo , Inhibidores de Proteasas/farmacología , Péptido Hidrolasas/metabolismo , Línea Celular Tumoral , Factor de Crecimiento de Hepatocito/genética , Factor de Crecimiento de Hepatocito/metabolismo , Factor de Crecimiento de Hepatocito/farmacologíaRESUMEN
Colorectal cancer exhibits dynamic cellular and genetic heterogeneity during progression from precursor lesions toward malignancy. Analysis of spatial multi-omic data from 31 human colorectal specimens enabled phylogeographic mapping of tumor evolution that revealed individualized progression trajectories and accompanying microenvironmental and clonal alterations. Phylogeographic mapping ordered genetic events, classified tumors by their evolutionary dynamics, and placed clonal regions along global pseudotemporal progression trajectories encompassing the chromosomal instability (CIN+) and hypermutated (HM) pathways. Integrated single-cell and spatial transcriptomic data revealed recurring epithelial programs and infiltrating immune states along progression pseudotime. We discovered an immune exclusion signature (IEX), consisting of extracellular matrix regulators DDR1, TGFBI, PAK4, and DPEP1, that charts with CIN+ tumor progression, is associated with reduced cytotoxic cell infiltration, and shows prognostic value in independent cohorts. This spatial multi-omic atlas provides insights into colorectal tumor-microenvironment co-evolution, serving as a resource for stratification and targeted treatments.
Asunto(s)
Neoplasias Colorrectales , Inestabilidad de Microsatélites , Microambiente Tumoral , Humanos , Inestabilidad Cromosómica/genética , Neoplasias Colorrectales/patología , Perfilación de la Expresión Génica , Quinasas p21 Activadas/genética , Filogenia , Mutación , Progresión de la Enfermedad , PronósticoRESUMEN
Multiplex immunofluorescence (MxIF) is an emerging imaging technology whose downstream molecular analytics highly rely upon the effectiveness of cell segmentation. In practice, multiple membrane markers (e.g., NaKATPase, PanCK and ß-catenin) are employed to stain membranes for different cell types, so as to achieve a more comprehensive cell segmentation since no single marker fits all cell types. However, prevalent watershed-based image processing might yield inferior capability for modeling complicated relationships between markers. For example, some markers can be misleading due to questionable stain quality. In this paper, we propose a deep learning based membrane segmentation method to aggregate complementary information that is uniquely provided by large scale MxIF markers. We aim to segment tubular membrane structure in MxIF data using global (membrane markers z-stack projection image) and local (separate individual markers) information to maximize topology preservation with deep learning. Specifically, we investigate the feasibility of four SOTA 2D deep networks and four volumetric-based loss functions. We conducted a comprehensive ablation study to assess the sensitivity of the proposed method with various combinations of input channels. Beyond using adjusted rand index (ARI) as the evaluation metric, which was inspired by the clDice, we propose a novel volumetric metric that is specific for skeletal structure, denoted as clDiceSKEL. In total, 80 membrane MxIF images were manually traced for 5-fold cross-validation. Our model outperforms the baseline with a 20.2% and 41.3% increase in clDiceSKEL and ARI performance, which is significant (p<0.05) using the Wilcoxon signed rank test. Our work explores a promising direction for advancing MxIF imaging cell segmentation with deep learning membrane segmentation. Tools are available at https://github.com/MASILab/MxIF_Membrane_Segmentation.
RESUMEN
Crohn's disease (CD) is a complex chronic inflammatory disorder that may affect any part of gastrointestinal tract with extra-intestinal manifestations and associated immune dysregulation. To characterize heterogeneity in CD, we profiled single-cell transcriptomics of 170 samples from 65 CD patients and 18 non-inflammatory bowel disease (IBD) controls in both the terminal ileum (TI) and ascending colon (AC). Analysis of 202,359 cells identified a novel epithelial cell type in both TI and AC, featuring high expression of LCN2, NOS2, and DUOX2, and thus is named LND. LND cells, confirmed by high-resolution in-situ RNA imaging, were rarely found in non-IBD controls, but expanded significantly in active CD. Compared to other epithelial cells, genes defining LND cells were enriched in antimicrobial response and immunoregulation. Moreover, multiplexed protein imaging demonstrated that LND cell abundance was associated with immune infiltration. Cross-talk between LND and immune cells was explored by ligand-receptor interactions and further evidenced by their spatial colocalization. LND cells showed significant enrichment of expression specificity of IBD/CD susceptibility genes, revealing its role in immunopathogenesis of CD. Investigating lineage relationships of epithelial cells detected two LND cell subpopulations with different origins and developmental potential, early and late LND. The ratio of the late to early LND cells was related to anti-TNF response. These findings emphasize the pathogenic role of the specialized LND cell type in both Crohn's ileitis and Crohn's colitis and identify novel biomarkers associated with disease activity and treatment response.
RESUMEN
Motivation: Multiplexed immunofluorescence (mIF) is an emerging assay for multichannel protein imaging that can decipher cell-level spatial features in tissues. However, existing automated cell phenotyping methods, such as clustering, face challenges in achieving consistency across experiments and often require subjective evaluation. As a result, mIF analyses often revert to marker gating based on manual thresholding of raw imaging data. Results: To address the need for an evaluable semi-automated algorithm, we developed GammaGateR, an R package for interactive marker gating designed specifically for segmented cell-level data from mIF images. Based on a novel closed-form gamma mixture model, GammaGateR provides estimates of marker-positive cell proportions and soft clustering of marker-positive cells. The model incorporates user-specified constraints that provide a consistent but slide-specific model fit. We compared GammaGateR against the newest unsupervised approach for annotating mIF data, employing two colon datasets and one ovarian cancer dataset for the evaluation. We showed that GammaGateR produces highly similar results to a silver standard established through manual annotation. Furthermore, we demonstrated its effectiveness in identifying biological signals, achieved by mapping known spatial interactions between CD68 and MUC5AC cells in the colon and by accurately predicting survival in ovarian cancer patients using the phenotype probabilities as input for machine learning methods. GammaGateR is a highly efficient tool that can improve the replicability of marker gating results, while reducing the time of manual segmentation. Availability and Implementation: The R package is available at https://github.com/JiangmeiRubyXiong/GammaGateR.
RESUMEN
Microvillus inclusion disease (MVID), caused by loss-of-function mutations in the motor protein myosin Vb (MYO5B), is a severe infantile disease characterized by diarrhea, malabsorption, and acid/base instability, requiring intensive parenteral support for nutritional and fluid management. Human patient-derived enteroids represent a model for investigation of monogenic epithelial disorders but are a rare resource from MVID patients. We developed human enteroids with different loss-of function MYO5B variants and showed that they recapitulated the structural changes found in native MVID enterocytes. Multiplex immunofluorescence imaging of patient duodenal tissues revealed patient-specific changes in localization of brush border transporters. Functional analysis of electrolyte transport revealed profound loss of Na+/H+ exchange (NHE) activity in MVID patient enteroids with near-normal chloride secretion. The chloride channel-blocking antidiarrheal drug crofelemer dose-dependently inhibited agonist-mediated fluid secretion. MVID enteroids exhibited altered differentiation and maturation versus healthy enteroids. γ-Secretase inhibition with DAPT recovered apical brush border structure and functional Na+/H+ exchange activity in MVID enteroids. Transcriptomic analysis revealed potential pathways involved in the rescue of MVID cells including serum/glucocorticoid-regulated kinase 2 (SGK2) and NHE regulatory factor 3 (NHERF3). These results demonstrate the utility of patient-derived enteroids for developing therapeutic approaches to MVID.
Asunto(s)
Síndromes de Malabsorción , Mucolipidosis , Miosina Tipo V , Humanos , Microvellosidades/genética , Cadenas Pesadas de Miosina/genética , Miosina Tipo V/genética , Enterocitos/metabolismo , Síndromes de Malabsorción/genética , Síndromes de Malabsorción/terapia , Síndromes de Malabsorción/metabolismo , Mucolipidosis/genética , Mucolipidosis/terapia , Mucolipidosis/metabolismoRESUMEN
Crohn's disease (CD) is a debilitating inflammatory bowel disease with no known cure. Computational analysis of hematoxylin and eosin (H&E) stained colon biopsy whole slide images (WSIs) from CD patients provides the opportunity to discover unknown and complex relationships between tissue cellular features and disease severity. While there have been works using cell nuclei-derived features for predicting slide-level traits, this has not been performed on CD H&E WSIs for classifying normal tissue from CD patients vs active CD and assessing slide label-predictive performance while using both separate and combined information from pseudo-segmentation labels of nuclei from neutrophils, eosinophils, epithelial cells, lymphocytes, plasma cells, and connective cells. We used 413 WSIs of CD patient biopsies and calculated normalized histograms of nucleus density for the six cell classes for each WSI. We used a support vector machine to classify the truncated singular value decomposition representations of the normalized histograms as normal or active CD with four-fold cross-validation in rounds where nucleus types were first compared individually, the best was selected, and further types were added each round. We found that neutrophils were the most predictive individual nucleus type, with an AUC of 0.92 ± 0.0003 on the withheld test set. Adding information improved cross-validation performance for the first two rounds and on the withheld test set for the first three rounds, though performance metrics did not increase substantially beyond when neutrophils were used alone.
RESUMEN
The Tangram algorithm is a benchmarking method of aligning single-cell (sc/snRNA-seq) data to various forms of spatial data collected from the same region. With this data alignment, the annotation of the single-cell data can be projected to spatial data. However, the cell composition (cell-type ratio) of the single-cell data and spatial data might be different because of heterogeneous cell distribution. Whether the Tangram algorithm can be adapted when the two data have different cell-type ratios has not been discussed in previous works. In our practical application that maps the cell-type classification results of single-cell data to the Multiplex immunofluorescence (MxIF) spatial data, cell-type ratios were different, though they were sampled from adjacent areas. In this work, both simulation and empirical validation were conducted to quantitatively explore the impact of the mismatched cell-type ratio on the Tangram mapping in different situations. Results show that the cell-type difference has a negative influence on classification accuracy.
RESUMEN
BACKGROUND & AIMS: Elements of field cancerization, including atrophic gastritis, metaplasia, and dysplasia, promote gastric cancer development in association with chronic inflammation. However, it remains unclear how stroma changes during carcinogenesis and how the stroma contributes to progression of gastric preneoplasia. Here we investigated heterogeneity of fibroblasts, one of the most important elements in the stroma, and their roles in neoplastic transformation of metaplasia. METHODS: We used single-cell transcriptomics to evaluate the cellular heterogeneity of mucosal cells from patients with gastric cancer. Tissue sections from the same cohort and tissue microarrays were used to identify the geographical distribution of distinct fibroblast subsets. We further evaluated the role of fibroblasts from pathologic mucosa in dysplastic progression of metaplastic cells using patient-derived metaplastic gastroids and fibroblasts. RESULTS: We identified 4 subsets of fibroblasts within stromal cells defined by the differential expression of PDGFRA, FBLN2, ACTA2, or PDGFRB. Each subset was distributed distinctively throughout stomach tissues with different proportions at each pathologic stage. The PDGFRα+ subset expanded in metaplasia and cancer compared with normal, maintaining a close proximity with the epithelial compartment. Co-culture of metaplasia- or cancer-derived fibroblasts with gastroids showing the characteristics of spasmolytic polypeptide-expressing metaplasia-induced disordered growth, loss of metaplastic markers, and increases in markers of dysplasia. Culture of metaplastic gastroids with conditioned media from metaplasia- or cancer-derived fibroblasts also promoted dysplastic transition. CONCLUSIONS: These findings indicate that fibroblast associations with metaplastic epithelial cells can facilitate direct transition of metaplastic spasmolytic polypeptide-expressing metaplasia cell lineages into dysplastic lineages.
Asunto(s)
Mucosa Gástrica , Neoplasias Gástricas , Humanos , Mucosa Gástrica/patología , Neoplasias Gástricas/patología , Hiperplasia , Metaplasia/patología , Fibroblastos/metabolismoRESUMEN
Environmental enteric dysfunction (EED) is characterized by malabsorption and diarrhea that result in irreversible deficits in physical and intellectual growth. We sought to define the expression of transport and tight junction proteins by quantitative analysis of duodenal biopsies from patients with EED. Biopsies from Pakistani children with confirmed EED diagnoses were compared to those from age-matched North American healthy controls, patients with celiac disease, and patients with nonceliac disease with villous atrophy or intraepithelial lymphocytosis. Expression of brush border digestive and transport proteins and paracellular (tight junction) proteins was assessed by quantitative multiplex immunofluorescence microscopy. EED was characterized by partial villous atrophy and marked intraepithelial lymphocytosis. Epithelial proliferation and enteroendocrine, tuft, and Paneth cell numbers were unchanged, but there was significant goblet cell expansion in EED biopsies. Expression of proteins involved in nutrient and water absorption and that of the basolateral Cl- transport protein NKCC1 were also increased in EED. Finally, the barrier-forming tight junction protein claudin-4 (CLDN4) was significantly upregulated in EED, particularly within villous enterocytes. In contrast, expression of CFTR, CLDN2, CLDN15, JAM-A, occludin, ZO-1, and E-cadherin was unchanged. Upregulation of a barrier-forming tight junction protein and brush border and basolateral membrane proteins that support nutrient and water transport in EED is paradoxical, as their increased expression would be expected to be correlated with increased intestinal barrier function and enhanced absorption, respectively. These data suggest that EED activates adaptive intestinal epithelial responses to enhance nutrient absorption but that these changes are insufficient to restore health.
Asunto(s)
Mucosa Intestinal , Linfocitosis , Niño , Humanos , Mucosa Intestinal/metabolismo , Linfocitosis/metabolismo , Linfocitosis/patología , Uniones Estrechas/metabolismo , Proteínas de Uniones Estrechas/metabolismo , Atrofia/metabolismo , Atrofia/patologíaRESUMEN
Microvillus Inclusion Disease (MVID), caused by loss-of-function mutations in the motor protein Myosin Vb (MYO5B), is a severe infantile disease characterized by diarrhea, malabsorption, and acid-base instability, requiring intensive parenteral support for nutritional and fluid management. Human patient-derived enteroids represent a model for investigation of monogenic epithelial disorders but are a rare resource from MVID patients. We developed human enteroids with different loss-of function MYO5B variants and showed that they recapitulated the structural changes found in native MVID enterocytes. Multiplex Immunofluorescence imaging of patient duodenal tissues revealed patient-specific changes in localization of brush border transporters. Functional analysis of electrolyte transport revealed profound loss of Na + /H + exchange (NHE) activity in MVID patient enteroids with near-normal chloride secretion. The chloride channel-blocking anti-diarrheal drug, Crofelemer, dose-dependently inhibited agonist-mediated fluid secretion. MVID enteroids exhibited altered differentiation and maturation versus healthy enteroids. Inhibition of Notch signaling with the γ-secretase inhibitor, DAPT, recovered apical brush border structure and functional Na + /H + exchange activity in MVID enteroids. Transcriptomic analysis revealed potential pathways involved in the rescue of MVID cells including serum- and glucocorticoid-induced protein kinase 2 (SGK2), and NHE regulatory factor 3 (NHERF3). These results demonstrate the utility of patient-derived enteroids for developing therapeutic approaches to MVID. Conflict-of-interest statement: The authors have declared that no conflict of interest exists.
RESUMEN
Many anomaly detection approaches, especially deep learning methods, have been recently developed to identify abnormal image morphology by only employing normal images during training. Unfortunately, many prior anomaly detection methods were optimized for a specific "known" abnormality (e.g., brain tumor, bone fraction, cell types). Moreover, even though only the normal images were used in the training process, the abnormal images were often employed during the validation process (e.g., epoch selection, hyper-parameter tuning), which might leak the supposed "unknown" abnormality unintentionally. In this study, we investigated these two essential aspects regarding universal anomaly detection in medical images by (1) comparing various anomaly detection methods across four medical datasets, (2) investigating the inevitable but often neglected issues on how to unbiasedly select the optimal anomaly detection model during the validation phase using only normal images, and (3) proposing a simple decision-level ensemble method to leverage the advantage of different kinds of anomaly detection without knowing the abnormality. The results of our experiments indicate that none of the evaluated methods consistently achieved the best performance across all datasets. Our proposed method enhanced the robustness of performance in general (average AUC 0.956).
RESUMEN
Multi-instance learning (MIL) is widely used in the computer-aided interpretation of pathological Whole Slide Images (WSIs) to solve the lack of pixel-wise or patch-wise annotations. Often, this approach directly applies "natural image driven" MIL algorithms which overlook the multi-scale (i.e. pyramidal) nature of WSIs. Off-the-shelf MIL algorithms are typically deployed on a single-scale of WSIs (e.g., 20× magnification), while human pathologists usually aggregate the global and local patterns in a multi-scale manner (e.g., by zooming in and out between different magnifications). In this study, we propose a novel cross-scale attention mechanism to explicitly aggregate inter-scale interactions into a single MIL network for Crohn's Disease (CD), which is a form of inflammatory bowel disease. The contribution of this paper is two-fold: (1) a cross-scale attention mechanism is proposed to aggregate features from different resolutions with multi-scale interaction; and (2) differential multi-scale attention visualizations are generated to localize explainable lesion patterns. By training ~250,000 H&E-stained Ascending Colon (AC) patches from 20 CD patient and 30 healthy control samples at different scales, our approach achieved a superior Area under the Curve (AUC) score of 0.8924 compared with baseline models. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.
RESUMEN
Multi-modal learning (e.g., integrating pathological images with genomic features) tends to improve the accuracy of cancer diagnosis and prognosis as compared to learning with a single modality. However, missing data is a common problem in clinical practice, i.e., not every patient has all modalities available. Most of the previous works directly discarded samples with missing modalities, which might lose information in these data and increase the likelihood of overfitting. In this work, we generalize the multi-modal learning in cancer diagnosis with the capacity of dealing with missing data using histological images and genomic data. Our integrated model can utilize all available data from patients with both complete and partial modalities. The experiments on the public TCGA-GBM and TCGA-LGG datasets show that the data with missing modalities can contribute to multi-modal learning, which improves the model performance in grade classification of glioma cancer.