Selected publications

Superpixel-Based Conditional Random Fields (SuperCRF): Incorporating Global and Local Context for Enhanced Deep Learning in Melanoma Histopathology
Zormpas-Petridis K, Failmezger H, Raza SEA, Roxanis I., Jamin Y, Yuan Y
Frontiers in Oncology (2019)

The novel general framework SuperCRF improves cell classification by introducing global and local context-based information much like pathologists do. SuperCRF can be implemented in combination with any single-cell classifier and represent valuable tools to study the cancer-stroma-immune interface, which we used to identify predictors of survival in melanoma patients from conventional H&E stained histopathology.

Unmasking the tissue microecology of ductal carcinoma in situ with deep learning
Narayanan P, et al.

Automated identification and segmentation of DCIS components have been challenging due to their high morphological diversity, and we developed a deep learning approach to accomplish these tasks.

Breast cancer nuclear integrity regulator

Through bioinformatics integration of morphology and genomics in 1,000 primary breast tumours, we discovered genes acting as regulators of cancer nuclear morphology. Their function as regulators of nuclear integrity has been validated in vitro.

ConCORDe-Net: Cell Count Regularized Convolutional Neural Network for Cell Detection in Multiplex Immunohistochemistry Images
Hagos YB, Narayanan P, Akarca AU, Marafioti T, Yuan Y
MICCAI: Medical Image Computing and Computer Assisted Interventions (2019)

Inspired by inception-v3 , we developed Cell COunt RegularizeD Convolutional neural Network (ConCORDe-Net) which integrates conventional dice overlap and a new cell count loss function for optimizing cell detection.

Deconvolving convolutional neural network for cell detection
Raza SEA, AbdulJabbar K, Jamal-Hanjani M, Veeriah S, Le Quesne J, Swanton C, Yuan Y
ISBI: IEEE International Symposium on Biomedical Imaging (2019)

We present an alternate approach to conventional local maxima detection for cell nucleus identification in a histology image. A convolutional neural network (CNN) is modified to convolve it's output with the same mapping filter and is trained for the mapped labels.

Microenvironmental niche divergence shapes BRCA1-dysregulated ovarian cancer morphological plasticity
Heindl A, Khan AM, Rodrigues DN, Eason K, Sadanandam A, Orbegoso C, Punta M, Sottoriva A, Lise S, Banerjee S, Yuan Y
Nature Communications (2018) [Online Sweave]

We provide empirical evidence supporting the selective advantage of cancer cell subpopulation with morphological diversification in locally immunosuppressive microenvironment, signifying morphological diversification as an ecological process of cancer with profound clinical implications.

DeepSDCS: Dissecting cancer proliferation heterogeneity in Ki67 digital whole slide images
Narayanan, P, Dodson A, Gusterson B, Dowsett M, Yuan Y
Medical Imaging with Deep Learning (2018)

Simultaneous Detection and Cell Segmentation (DeepSDCS) performs automated cell segmentation and detection in Ki67-stained immunohistochemistry images. This network combines VGG16 with hypercolumn descriptors to form the vector of activation from specific layers to capture features at different granularity.

Relevance of spatial heterogeneity of immune infiltration for predicting risk of recurrence after endocrine therapy of ER+ breast cancer.
Heindl A, Sestak I, Naidoo R, Cuzick J, Dowsett M, Yuan Y*
JNCI (2018) [Additional ppt]

We provide a missing link between tumour immunity and disease outcome in ER+ breast cancer subtype by examining tumour spatial architecture. The association between immune spatial scores and late recurrence suggests a lasting memory of protumour immunity that may impact disease progression and evolution of endocrine treatment resistance, which may be exploited for therapeutic advances.

Biopsy variability of lymphocytic infiltration in breast cancer subtypes and the ImmunoSkew score.
Khan AM, Yuan Y*
Nature Scientific Reports (2016)

We systematically investigate biopsy variability for the lymphocytic infiltrate in 998 breast tumours using a novel virtual biopsy method. Biopsy variability of lymphocytic infiltrate differs considerably among breast cancer subtypes, with the HER2+ subtype having the highest variability.

Similarity and diversity of the tumor microenvironment in multiple metastases: critical implications for overall and progression-free survival of serous ovarian cancer.
Heindl A, Lan C, Rodrigues DN, Koelble K, Yuan Y*
Oncotarget (2016) [Sweave; Digital slides]

Diverse cell populations at the sites of cancer spread could be a clinically important feature of particularly aggressive ovarian cancers. Based on this, we have developed a new test to assess the diversity of metastatic sites, and use it to predict a patient's chances of surviving their diseases.

Spatial Heterogeneity in the Tumor Microenvironment.
Yuan Y*
Cold Spring Harb Perspect Med. (2016) [Limited reprints]

There is a desperate need to understand influence of the tumor microenvironment on cancer development and evolution. Applying principles and quantitative methods from ecology can suggest novel solutions to fulfil this need. We discuss spatial heterogeneity as a fundamental biological feature of the microenvironment, which has been largely ignored.

Microenvironmental heterogeneity parallels breast cancer progression: A histology-genomics integration analysis
Natrajan R, Sailem H, Mardakheh FM, Arias MG, Dowsett M, Bakal C, Yuan Y*
PLOS Medicine (2016) [Sweave; R package beta version]

We propose a clinically relevant role of tumour microenvironmental diversity for advanced breast tumors and highlight that ecological statistics can be translated into medical advances for identifying new biomarkers and for understanding the synergistic interplay between the microenvironment and cancer genomics.

Computational pathology: Exploring the spatial dimension of tumor ecology
Nawaz S, Yuan Y*
Cancer Letters (2015)

Cancer and normal cells exhibit both co-operative and competitive relationships, analogous to living organisms. Considering the tumour as an ecological habitat, we present current and potential applications of methods developed for analysing ecological relationships in nature to data derived from tumour histology. We discuss how this novel way of studying the tumour microenvironment can expand our knowledge of cancer progression and reveal new clinical prognosticators.

Quantitative histology analysis of the ovarian tumour microenvironment
Lan C^, Heindl A^, Huang X, Xi S, Banerjee S, Liu J*, Yuan Y*
Nature Scientific Reports (2015) [Sweave; data; Digital slides]

By developing automated histology analysis as a cost-efficient subtyping technology for ovarian cancer, we reveal a strong effect of the tumour microenvironment on ovarian cancer progression and highlight the potential of therapeutic interventions that target the stromal compartment.

An ecological measure of immune-cancer colocalization as a prognostic factor for breast cancer
Maley CC, Koelble K, Natrajan R, Aktipis A, Yuan Y*
Breast Cancer Research (2015) [Sweave; data]

We demonstrate how ecological methods applied to the study of the tumour microenvironment using routine histology samples can provide reproducible, quantitative biomarkers for identifying high-risk breast cancer patients.

Beyond immune density: critical role of spatial heterogeneity in estrogen receptor-negative breast cancer
Nawaz S, Heindl A, Koelble K, Yuan Y*
Modern Pathology (2015) [Sweave; data]

Using a statistical method commonly applied in crime mapping, we quantify the spatial clustering of cancer and immune cells in breast tumours, which led to the discovery that unexpectedly high levels of co-clustering of cancer and immune cells is indicative of favourable long-term prognosis in ER-negative breast cancer, an aggressive breast cancer subtype.

Modelling the spatial heterogeneity and molecular correlates of lymphocytic infiltration in triple-negative breast cancer
Yuan Y*
Journal of the Royal Society Interface (2015) [Sweave; data and functions]

This paper reports the first fully automated, computer-assisted method for scoring intratumour lymphocytes in breast cancer samples with the fusion of high-throughput image processing and robust statistical analysis.

Mapping spatial heterogeneity in the tumor microenvironment: a new era for digital pathology
Heindl A, Nawaz S, Yuan Y*
Laboratory Investigation (2015)

We review existing methods for analysing spatial heterogeneity in the tumour microenvironment and how they may be integrated with molecular profiling to help further our understanding of the complex relationship between cancer cells and the surrounding healthy tissue that can play a pivotal role in tumour progression.



FISHalyseR is a highthroughput tool for quantitative analysis of molecular heterogeneity at 
single-cell resolution in FISH images with an unlimited number of probes. [Download from Bioconductor]

Download statistics
Download stats

CRImage provides image analysis tools for segmentation, classification, and downstream analysis of H&E images. One application is for cellularity scoring of tumours by counting the number of cancer cells and other cells. The package also comes with a novel algorithm for copy-number data correction for SNP microarray data using estimates of tumours cellularity from pathological image analysis. [Download from Bioconductor]

Download statistics
Download stats