Biology subjects

Hruban, R. H.

Publications and source records attributed to Hruban, R. H..

At least 19 recordsLinked to original sources

bioRxiv2026

Virtual multiplex staining of the pancreatic islets across type 1 diabetes progression using a Schroedinger bridge

Classical hematoxylin and eosin (H&E) staining enables review of tissue morphology but lacks information regarding the molecular state of cells. Immunohistochemical (IHC) techniques label specific proteins in tissue, allowing differentiation of relevant structures that may go undetectable in H&E. However, the IHC process is complex, expensive, and time-consuming, especially for multiplex IHC (mIHC) limiting its use in large cohorts. Stain conversion of H&E to IHC using generative artificial intelligence models such as generative adversarial networks (GANs) represent one solution to this problem. However, GANs are unstable during out of distribution sampling and are prone to hallucinations or mode collapse, limiting their accuracy in challenging image conversion tasks. To address this, the field has recently turned to diffusion models. Here, we introduce Schrodinger-bridge for Multiplex ImmunoLabel Estimation (SMILE). Unlike conventional diffusion models that map from source to target through an intermediate Gaussian noise, Schrodinger-bridge diffusion models skip this step and have been shown to better preserve structures during image translation. To test the performance of SMILE, we generated a large cohort of high-fidelity H&E-mIHC image pairs from pancreatic organ donors, targeting insulin, glucagon, and CD3. Our dataset well-sampled across type-1 diabetes status, pancreas anatomical location, age, and sex. Using this cohort, we demonstrate the superiority of SMILE compared to GANs via a comprehensive evaluation framework incorporating texture, distribution, and antibody-specific metrics, as well as blinded pathologist reviews. We further confirmed the ability of SMILE to generate accurate mIHC images from H&Es generated at an external site, to perform whole slide image conversion, and to generate realistic three-dimensional maps of the pancreatic islets in non-diabetic, auto-antibody positive, and type-1 diabetic donor tissue. Finally, we performed stain conversion of paired H&E to HER2 and Ki67 images in breast cancer, confirming the superiority of SMILE in diverse stain conversion applications. Collectively, this framework provides a scalable pipeline for high-throughput proteomic inference from archival H&Es, providing transformative potential for pancreatic research and digital pathology.

bioinformatics↗

bioRxiv2025

CODAvision: best practices and a user-friendly interface for rapid, customizable segmentation of medical images

Image-based machine learning tools have emerged as powerful resources for analyzing medical images, with deep learning-based semantic segmentation commonly utilized to enable spatial quantification of structures in images. However, customization and training of segmentation algorithms requires advanced programming skills and intricate workflows, limiting their accessibility to many investigators. Here, we present a protocol and software for automatic segmentation of medical images guided by a graphical user interface (GUI) using the CODAvision algorithm. This workflow simplifies the process of semantic segmentation of microanatomical structures by enabling users to train highly customizable deep learning models without extensive coding expertise. The protocol outlines best practices for creating robust training datasets, configuring model parameters, and optimizing performance across diverse biomedical image modalities. CODAvision enhances the usability of the CODA algorithm (Nature Methods, 2022) by streamlining parameter configuration, model training, and performance evaluation, automatically generating quantitative results and comprehensive reports. We expand beyond the original implementation of CODA to serial histology by demonstrating robust performance across numerous medical image modalities and diverse biological questions. We provide sample results in data types including histology, magnetic resonance imaging (MRI), and computed tomography (CT). We demonstrate the diverse use of this tool in applications including quantification of metastatic burden in in vivo models and deconvolution of spot-based spatial transcriptomics datasets. This protocol is designed for researchers with interest in rapid design of highly customizable semantic segmentation algorithms and a basic understanding of programming and anatomy.

bioinformatics↗

bioRxiv2024

3D histology reveals that immune response to pancreatic precancers is heterogeneous and depends on global pancreas structure

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal cancer for which few effective therapies exist. Immunotherapies specifically are ineffective in pancreatic cancer, in part due to its unique stromal and immune microenvironment. Pancreatic intraepithelial neoplasia, or PanIN, is the main precursor lesion to PDAC. Recently it was discovered that PanINs are remarkably abundant in the grossly normal pancreas, suggesting that the vast majority will never progress to cancer. Here, through construction of 48 samples of cm3-sized human pancreas tissue, we profiled the immune microenvironment of 1,476 PanINs in 3D and at single-cell resolution to better understand the early evolution of the pancreatic tumor microenvironment and to determine how inflammation may play a role in cancer progression. We found that bulk pancreatic inflammation strongly correlates to PanIN cell fraction. We found that the immune response around PanINs is highly heterogeneous, with distinct immune hotspots and cold spots that appear and disappear in a span of tens of microns. Immune hotspots generally mark locations of higher grade of dysplasia or locations near acinar atrophy. The immune composition at these hotspots is dominated by naive, cytotoxic, and regulatory T cells, cancer associated fibroblasts, and tumor associated macrophages, with little similarity to the immune composition around less-inflamed PanINs. By mapping FOXP3+ cells in 3D, we found that regulatory T cells are present at higher density in larger PanIN lesions compared to smaller PanINs, suggesting that the early initiation of PanINs may not exhibit an immunosuppressive response. This analysis demonstrates that while PanINs are common in the pancreases of most individuals, inflammation may play a pivotal role, both at the bulk and the microscopic scale, in demarcating regions of significance in cancer progression.

cancer biology↗

bioRxiv2024

Multi-omic profiling of intraductal papillary neoplasms of the pancreas reveals distinct expression patterns and potential markers of progression

In order to advance our understanding of precancers in the pancreas, 69 pancreatic intraductal papillary neoplasms (IPNs), including 64 intraductal papillary mucinous neoplasms (IPMNs) and 5 intraductal oncocytic papillary neoplasms (IOPNs), 32 pancreatic cyst fluid samples, 104 invasive pancreatic ductal adenocarcinomas (PDACs), 43 normal adjacent tissues (NATs), and 76 macro-dissected normal pancreatic ducts (NDs) were analyzed by mass spectrometry. A total of 10,246 proteins and 22,284 glycopeptides were identified in all tissue samples, and 756 proteins with more than 1.5-fold increase in abundance in IPMNs relative to NDs were identified, 45% of which were also identified in cyst fluids. The over-expression of selected proteins was validated by immunolabeling. Proteins and glycoproteins overexpressed in IPMNs included those involved in glycan biosynthesis and the immune system. In addition, multiomics clustering identified two subtypes of IPMNs. This study provides a foundation for understanding tumor progression and targets for earlier detection and therapies. SignificanceThis multilevel characterization of intraductal papillary neoplasms of the pancreas provides a foundation for understanding the changes in protein and glycoprotein expression during the progression from normal duct to intraductal papillary neoplasm, and to invasive pancreatic carcinoma, providing a foundation for informed approaches to earlier detection and treatment.

cancer biology↗

bioRxiv2024

Islands of genomic stability in the face of genetically unstable metastatic cancer

AbstractO_ST_ABSIntroductionC_ST_ABSMetastatic cancer affects millions of people worldwide annually and is the leading cause of cancer-related deaths. Most patients with metastatic disease are not eligible for surgical resection, and current therapeutic regimens have varying success rates, some with 5-year survival rates below 5%. Here we test the hypothesis that metastatic cancer can be genetically targeted by exploiting single base substitution mutations unique to individual cells that occur as part of normal aging prior to transformation. These mutations are targetable because [~]10% of them form novel tumor-specific "NGG" protospacer adjacent motif (PAM) sites targetable by CRISPR-Cas9. MethodsWhole genome sequencing was performed on five rapid autopsy cases of patient-matched primary tumor, normal and metastatic tissue from pancreatic ductal adenocarcinoma decedents. CRISPR-Cas9 PAM targets were determined by bioinformatic tumor-normal subtraction for each patient and verified in metastatic samples by high-depth capture-based sequencing. ResultsWe found that 90% of PAM targets were maintained between primary carcinomas and metastases overall. We identified rules that predict PAM loss or retention, where PAMs located in heterozygous regions in the primary tumor can be lost in metastases (private LOH), but PAMs occurring in regions of loss of heterozygosity (LOH) in the primary tumor were universally conserved in metastases. ConclusionsRegions of truncal LOH are strongly retained in the presence of genetic instability, and therefore represent genetic vulnerabilities in pancreatic adenocarcinomas. A CRISPR-based gene therapy approach targeting these regions may be a novel way to genetically target metastatic cancer.

cancer biology↗

bioRxiv2023

Three-dimensional assessments are necessary to determine the true spatial tissue composition of diseased tissues

Methods for spatially resolved cellular profiling using thinly cut sections have enabled in-depth quantitative tissue mapping to study inter-sample and intra-sample differences in normal human anatomy and disease onset and progression. These methods often profile extremely limited regions, which may impact the evaluation of heterogeneity due to tissue sub-sampling. Here, we applied CODA, a deep learning-based tissue mapping platform, to reconstruct the three-dimensional (3D) microanatomy of grossly normal and cancer-containing human pancreas biospecimens obtained from individuals who underwent pancreatic resection. To compare inter-and intra-sample heterogeneity, we assessed bulk and spatially resolved tissue composition in a cohort of two-dimensional (2D) whole slide images (WSIs) and a cohort of thick slabs of pancreas tissue that were digitally reconstructed in 3D from serial sections. To demonstrate the marked under sampling of 2D assessments, we simulated the number of WSIs and tissue microarrays (TMAs) necessary to represent the compositional heterogeneity of 3D data within 10% error to reveal that tens of WSIs and hundreds of TMA cores are sometimes needed. We show that spatial correlation of different pancreatic structures decay significantly within a span of microns, demonstrating that 2D histological sections may not be representative of their neighboring tissues. In sum, we demonstrate that 3D assessments are necessary to accurately assess tissue composition in normal and abnormal specimens and in order to accurately determine neoplastic content. These results emphasize the importance of intra-sample heterogeneity in tissue mapping efforts.

cancer biology↗

bioRxiv2023

Power-law growth models explain incidences and sizes of pancreatic cancer precursor lesions and confirm spatial genomic findings

Pancreatic ductal adenocarcinoma is a rare but lethal cancer. Recent evidence reveals that pancreatic intraepithelial neoplasms (PanINs), the microscopic precursor lesions in the pancreatic ducts that can give rise to invasive pancreatic cancer, are significantly larger and more prevalent than previously believed. Better understanding of the growth law dynamics of PanINs may improve our ability to understand how a miniscule fraction of these lesions makes the transition to invasive cancer. Here, using artificial intelligence (AI)-based three-dimensional (3D) tissue mapping method, we measured the volumes of >1,000 PanIN and found that lesion size is distributed according to a power law with a fitted exponent of -1.7 over > 3 orders of magnitude. Our data also suggest that PanIN growth is not very sensitive to the pancreatic microenvironment or an individuals age, family history, and lifestyle, and is rather shaped by general growth behavior. We analyze several models of PanIN growth and fit the predicted size distributions to the observed data. The best fitting models suggest that both intraductal spread of PanIN lesions and fusing of multiple lesions into large, highly branched structures drive PanIN growth patterns. This work lays the groundwork for future mathematical modeling efforts integrating PanIN incidence, morphology, genomic, and transcriptomic features to understand pancreas tumorigenesis, and demonstrates the utility of combining experimental measurement of human tissues with dynamic modeling for understanding cancer tumorigenesis.

biophysics↗

bioRxiv2023

Somatic protospacer adjacent motifs are numerous and selectively targetable in cancers

Somatic mutations are desirable targets for selective elimination of cancer, yet most are found within the noncoding regions. We propose a novel, cancer-specific killing approach using CRISPR-Cas9 which exploits the requirement of a protospacer adjacent motif (PAM) for Cas9 activity. Through whole genome sequencing (WGS) of paired tumor minus normal (T-N) samples from three pancreatic cancer patients (Panc480, Panc504, and Panc1002), we identified an average of 417 somatic PAMs per tumor produced from single base substitutions. We analyzed 591 paired T-N samples from The International Cancer Genome Consortium and discovered medians of [~]455 somatic PAMs per tumor in pancreatic, [~]2800 in lung, and [~]3200 in esophageal cancer cohorts. Finally, we demonstrated >80% selective cell death of two targeted pancreatic cancer cell lines in co-cultures using 4-9 sgRNAs, targeting noncoding regions, designed from the somatic PAM discovery approach. We also showed no off-target activity from these tumor-specific sgRNAs through WGS. Statement of significanceThis study demonstrates the potential of CRISPR-Cas9 as a novel and selective anti-cancer strategy. It requires just a few targets to induce double strand breaks for significant cytotoxicity. Our findings markedly expand the repertoire of targetable mutations in cancers and support genetically targeting other adult solid tumor types.

cancer biology↗

bioRxiv2023

Mechanism of delayed cell death following simultaneous CRISPR-Cas9 targeting in pancreatic cancers

While radiation is an effective oncologic therapy, killing cancer by inducing DNA double-strand breaks (DSBs), it lacks specificity for neoplastic cells. We have previously adapted the CRISPR-Cas9 gene-editing technology as a cancer-specific treatment modality targeting somatic mutations in pancreatic cancer (PC). However, its tumoricidal potential remains unclear, especially in comparison to therapeutic doses of radiation. Here, we demonstrate that CRISPR-Cas9-induced DSBs are more cytotoxic in PCs than a comparable number of radiation-induced DSBs. We observed >90% tumor growth inhibition by targeting 9 sites with cancer-specific single-guide RNAs (sgRNAs). Through both bioinformatics and cytogenetics analyses, we found that CRISPR-Cas9-induced DSBs triggered ongoing chromosomal rearrangements, with 87% of structural variants not directly produced from the initial CRISPR-Cas9-induced DSBs, and chromosomal instability (CIN) peaking before cell death. By comparing the cytotoxicity of CRISPR-Cas9- to radiation-induced DSBs, we demonstrate that the number of DSBs required to achieve equitoxic effects was [~]3 times higher for radiation than CRISPR-Cas9. Finally, we show that PC cells that had survived CRISPR-Cas9 targeting retained susceptibility to subsequent CRISPR-Cas9-induced DSBs at different genomic sites with >87% growth inhibition. Together, our data support the therapeutic potential of CRISPR-Cas9 as an anti-cancer strategy.

cancer biology↗

bioRxiv2023

MYC-driven increases in mitochondrial DNA copy number occur early and persist throughout prostatic cancer progression

Increased mitochondrial function may render some cancers vulnerable to mitochondrial inhibitors. Since mitochondrial function is regulated partly by mitochondrial DNA copy number (mtDNAcn), accurate measurements of mtDNAcn could help reveal which cancers are driven by increased mitochondrial function and may be candidates for mitochondrial inhibition. However, prior studies have employed bulk macrodissections that fail to account for cell type-specific or tumor cell heterogeneity in mtDNAcn. These studies have often produced unclear results, particularly in prostate cancer. Herein, we developed a multiplex in situ method to spatially quantify cell type specific mtDNAcn. We show that mtDNAcn is increased in luminal cells of high-grade prostatic intraepithelial neoplasia (HGPIN), is increased in prostatic adenocarcinomas (PCa), and is further elevated in metastatic castration-resistant prostate cancer. Increased PCa mtDNAcn was validated by two orthogonal methods and is accompanied by increases in mtRNAs and enzymatic activity. Mechanistically, MYC inhibition in prostate cancer cells decreases mtDNA replication and expression of several mtDNA replication genes, and MYC activation in the mouse prostate leads to increased mtDNA levels in the neoplastic prostate cells. Our in situ approach also revealed elevated mtDNAcn in precancerous lesions of the pancreas and colon/rectum, demonstrating generalization across cancer types using clinical tissue samples.

cancer biology↗

bioRxiv2023

Three-dimensional genomic mapping of human pancreatic tissue reveals striking multifocality and genetic heterogeneity in precancerous lesions

Pancreatic intraepithelial neoplasia (PanIN) is a precursor to pancreatic cancer and represents a critical opportunity for cancer interception. However, the number, size, shape, and connectivity of PanINs in human pancreatic tissue samples are largely unknown. In this study, we quantitatively assessed human PanINs using CODA, a novel machine-learning pipeline for 3D image analysis that generates quantifiable models of large pieces of human pancreas with single-cell resolution. Using a cohort of 38 large slabs of grossly normal human pancreas from surgical resection specimens, we identified striking multifocality of PanINs, with a mean burden of 13 spatially separate PanINs per cm3 of sampled tissue. Extrapolating this burden to the entire pancreas suggested a median of approximately 1000 PanINs in an entire pancreas. In order to better understand the clonal relationships within and between PanINs, we developed a pipeline for CODA-guided multi-region genomic analysis of PanINs, including targeted and whole exome sequencing. Multi-region assessment of 37 PanINs from eight additional human pancreatic tissue slabs revealed that almost all PanINs contained hotspot mutations in the oncogene KRAS, but no gene other than KRAS was altered in more than 20% of the analyzed PanINs. PanINs contained a mean of 13 somatic mutations per region when analyzed by whole exome sequencing. The majority of analyzed PanINs originated from independent clonal events, with distinct somatic mutation profiles between PanINs in the same tissue slab. A subset of the analyzed PanINs contained multiple KRAS mutations, suggesting a polyclonal origin even in PanINs that are contiguous by rigorous 3D assessment. This study leverages a novel 3D genomic mapping approach to describe, for the first time, the spatial and genetic multifocality of human PanINs, providing important insights into the initiation and progression of pancreatic neoplasia.

cancer biology↗

bioRxiv2020

In situ characterization of the 3D microanatomy of the pancreas and pancreatic cancer at single cell resolution

Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest forms of cancer. Accumulating evidence indicates the tumor microenvironment is highly associated with tumorigenesis through regulation of cellular physiology, signaling systems, and gene expression profiles of cancer cells. Yet the mechanisms by which the microenvironment evolves from normal pancreas architecture to precursor lesions and invasive cancer is poorly understood. Obtaining high-content and high-resolution information from a complex tumor microenvironment in large volumetric landscapes represents a key challenge in the field of cancer biology. To address this challenge, we established a novel method to reconstruct three-dimensional (3D) centimeter-scale tissues containing billions of cells from serially sectioned histological samples, utilizing deep learning approaches to recognize eight distinct tissue subtypes from hematoxylin and eosin stained sections at micrometer and single-cell resolution. Using samples from a range of normal, precancerous, and invasive pancreatic cancer tissue, we map in 3D modes of cancer invasion in the tumor microenvironment, and emphasize the need for further 3D quantification of biological systems.

cancer biology↗

bioRxiv2020

DETECTION OF INCIPIENT PANCREATIC CANCER WITH NOVEL TUMOR-SPECIFIC ANTIBODIES IN MOUSE MODELS

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy, as 90% of patients do not survive beyond five years from diagnosis. This dismal prognosis is largely due to the advanced stage of the disease at diagnosis, which precludes potentially curative surgical resection. Although early detection strategies hold significant promise for improving patient outcomes, there is still no accurate diagnostic tool to detect incipient PDAC. Here, we sought to develop antibodies for the early detection of PDAC by positron-emission tomography (PET) imaging. Accordingly, we establish a pipeline to generate novel tumor-specific monoclonal antibodies (mAbs) against cell-surface proteins of PDAC patient-derived organoids (PDOs). We identify a panel of 16 tumor organoid-binding antibodies (TOBi-bodies) that display high reactivity to human PDAC tissues but not to matched adjacent normal pancreas. We then employ biochemical, flow cytometric, mass spectrometric, and CRISPR/Cas9-mediated knockout methods to determine the cognate antigens of these TOBi-bodies. We identify two mAbs that bind to tumor-specific variants of the surface protein CEACAM6 and show minimal binding to normal tissues. PET imaging in mouse models using these TOBi-bodies enables the detection of incipient human organoid-derived PDAC tumors that are rather undetectable by palpation or high-resolution ultrasound imaging techniques. We propose that further development of these mAbs as PET radiotracers could facilitate the early detection and accurate staging of PDAC.

cancer biology↗

bioRxiv2020

Multiscale label-free volumetric holographic histopathology of thick-tissue slides with subcellular resolution

Histopathology relies upon the staining and sectioning of biological tissues, which can be laborious and may cause artefacts and distort tissues. Here, we demonstrate label-free volumetric imaging of thick-tissue slides, exploiting refractive index distributions as intrinsic imaging contrast. The present method systematically exploits label-free quantitative phase imaging techniques, volumetric reconstruction of intrinsic refractive index distributions in tissues, and numerical algorithms for the seamless stitching of multiple 3D tomograms and for reducing scattering-induced image distortion. We demonstrate demonstrated label-free volumetric imaging of thick tissues with the field of view of 2 mm x 1.75 mm x 0.2 mm with a spatial resolution of 170 nm x 170 nm x 1200 nm. The number of optical modes, calculated as the reconstructed volume divided by the size of the point spread function, was approximately 20 Giga voxels. We have also demonstrated that different tumour types, and a variety of precursor lesions and pathologies can be visualized with the present method. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=150 SRC="FIGDIR/small/205633v1_ufig1.gif" ALT="Figure 1"> View larger version (98K): org.highwire.dtl.DTLVardef@dcf15corg.highwire.dtl.DTLVardef@16a05d3org.highwire.dtl.DTLVardef@1a48860org.highwire.dtl.DTLVardef@f57430_HPS_FORMAT_FIGEXP M_FIG C_FIG

pathology↗

bioRxiv2019

Evolutionary Origins of Recurrent Pancreatic Cancer

Surgery is the only curative option for Stage I/II pancreatic cancer, nonetheless most patients will recur after surgery and die of their disease. To identify novel opportunities for management of recurrent pancreatic cancer we performed whole exome or targeted sequencing of 10 resected primary cancers and matched intrapancreatic recurrences or distant metastases. We identified that adjuvant or first-line platinum therapy corresponds to an increased mutational burden of recurrent disease. Recurrent disease is enriched for mutations that activate Mapk/Erk and PI3K/AKT signaling and develops from a monophyletic or polyphyletic origin. Treatment induced genetic bottlenecks lead to a modified genetic landscape and subclonal heterogeneity for driver gene alterations in part due to intermetastatic seeding. In one patient what was believed to be recurrent disease was an independent (second) primary tumor. These findings advocate for combination therapies with immunotherapy and routine post-treatment sampling as a component of management of recurrent pancreatic cancer.

cancer biology↗

bioRxiv2019

Assessing aneuploidy with repetitive element sequencing.

We report a sensitive PCR-based assay that can detect aneuploidy in samples containing as little as 3 picograms of DNA. Using a single primer pair, we amplified [~]750,000 amplicons distributed throughout the genome Aneuploidy was detected in 49% of liquid biopsies from a total of 883 non-metastatic cancers of eight different types. Combining aneuploidy with somatic mutation detection and eight standard protein biomarkers yielded a median sensitivity of 80% at 99% specificity.

cancer biology↗

bioRxiv2019

The Genetic Basis of Transcriptional and Spatial Heterogeneity of Squamous Features in Pancreatic Ductal Adenocarcinoma

Recent studies indicate that pancreatic cancer expression profiles are variable and largely reflect a classical or basal-type phenotype. We performed genetic sequencing, RNA-seq, and histologic review of multiregion sampled pancreatic cancers and found that squamous and squamoid features, indicators of poor prognosis, correlate with a "basal-like" expressional type. Cancers with squamous features were more likely to have truncal mutations in chromatin modifier genes and intercellular heterogeneity for MYC amplification that was associated with entosis. In most patients the basal phenotype coexisted with a glandular component, and phylogenetic studies indicated that it arose from a subclonal population in the tumor. These data provide a unifying paradigm for understanding the interrelationship of basal-type features, squamous histology, and somatic mutations in chromatin modifier genes in the context of the clonal evolution of pancreatic cancer.

cancer biology↗

bioRxiv2017

Computational discovery of tissue morphology biomarkers in very long-term survivors with pancreatic ductal adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest forms of cancer, with an average 5-year survival rate of only 8%. Within PDAC patients, however, there is a small subset of patients who survive >10 years. Deciphering underlying reasons behind prolonged survival could potentially provide new opportunities to treat PDAC; however, no genomic, transcriptomic, proteomic, or clinical signatures have been found to robustly separate this subset of patients. Digital pathology, in combination with machine learning, provides an opportunity to computationally search for tissue morphology patterns associated with disease outcomes. Here, we developed a computational framework to analyze whole-slide images (WSI) of PDAC patient tissue and identify tissue-morphology signatures for very long term surviving patients. Our results indicate that less tissue morphology heterogeneity is significantly linked to better patient survival and that the extra-tumoral space encodes prognostic information for survival. Based on information from morphological heterogeneity in the tumor and its adjacent area, we established a machine learning model with an AUC of 0.94. Our analysis workflow highlighted a quantitative visual-based tissue phenotype analysis that also allows direct interaction with pathology. This study demonstrates a pathway to accelerate the discovery of undetermined tissue morphology associated with pathogenesis states and prognosis and diagnosis of patients by utilizing new computational approaches.

bioinformatics↗