Corpus ID: 220280491

Towards Explainable Graph Representations in Digital Pathology

@article{Jaume2020TowardsEG,
  title={Towards Explainable Graph Representations in Digital Pathology},
  author={Guillaume Jaume and Pushpak Pati and Antonio Foncubierta-Rodr{\'i}guez and Florinda Feroce and Giosu{\`e} Scognamiglio and Anna Maria Anniciello and Jean-Philippe Thiran and Orcun Goksel and Maria Gabrani},
  journal={ArXiv},
  year={2020},
  volume={abs/2007.00311}
}
Explainability of machine learning (ML) techniques in digital pathology (DP) is of great significance to facilitate their wide adoption in clinics. Recently, graph techniques encoding relevant biological entities have been employed to represent and assess DP images. Such paradigm shift from pixel-wise to entity-wise analysis provides more control over concept representation. In this paper, we introduce a post-hoc explainer to derive compact per-instance explanations emphasizing diagnostically… Expand
Quantifying Explainers of Graph Neural Networks in Computational Pathology
TLDR
A set of novel quantitative metrics based on statistics of class separability using pathologically measurable concepts to characterize graph explainers are proposed and validated by expert pathologists to explain Cell-Graph representations for Breast Cancer Subtyping. Expand
HistoCartography: A Toolkit for Graph Analytics in Digital Pathology
TLDR
HistoCartography is developed, a standardized python API with necessary preprocessing, machine learning and explainability tools to facilitate graph-analytics in computational pathology and benchmarks the computational time and performance on multiple datasets across different imaging types and histopathology tasks. Expand
A Survey on Graph-Based Deep Learning for Computational Histopathology
TLDR
This review provides a conceptual grounding of graph-based deep learning and discusses its current success for tumor localization and classification, tumor invasion and staging, image retrieval, and survival prediction, and outlines the limitations of existing techniques, and suggests potential future advances. Expand
IA-GCN: Interpretable Attention based Graph Convolutional Network for Disease prediction
TLDR
An interpretable graph learning-based model which interprets the clinical relevance of the input features towards the task, uses the explanation to improve the model performance and learns a population level latent graph that may be used to interpret the cohort’s behavior is proposed. Expand
Hierarchical Graph Representations in Digital Pathology
TLDR
A novel multi-level hierarchical entity-graph representation of tissue specimens is proposed to model hierarchical compositions that encode histological entities as well as their intraand interentity level interactions to map the tissue structure to tissue functionality. Expand
Hierarchical Cell-to-Tissue Graph Representations for Breast Cancer Subtyping in Digital Pathology
TLDR
This work considers a tissue as a hierarchical composition of multiple types of histological entities from fine to coarse level, that depicts multivariate tissue information at multiple levels, and proposes a novel multi-level hierarchical entity-graph representation to depict a tissue specimen. Expand
Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future
TLDR
This survey thoroughly review the different types of graph architectures and their applications in healthcare in a systematic manner, organized by their domain of application including functional connectivity, anatomical structure, and electrical-based analysis. Expand
Predictive modelling of highly multiplexed tumour tissue images by graph neural networks
TLDR
This paper proposes and evaluates MULTIPLAI, a novel framework to predict clinical biomarkers from IMC data that relies on attention-based graph neural networks (GNNs) that integrate both the phenotypic and spatial dimensions of IMC images to capture important TME features with clinical importance. Expand
HACT-Net: A Hierarchical Cell-to-Tissue Graph Neural Network for Histopathological Image Classification
TLDR
A novel hierarchical cell-to-tissue-graph (HACT) representation is proposed to improve the structural depiction of the tissue and outperformed recent convolutional neural network and graph neural network approaches for breast cancer multi-class subtyping. Expand
Hierarchical Phenotyping and Graph Modeling of Spatial Architecture in Lymphoid Neoplasms
TLDR
This study is among the first to hybrid local and global graph methods to profile orchestration and interaction of cellular components to address the challenge in hematolymphoid cancers where the cell classes in TME are unclear. Expand
...
1
2
...

References

SHOWING 1-10 OF 24 REFERENCES
Regression Concept Vectors for Bidirectional Explanations in Histopathology
TLDR
This work proposes a methodology to exploit continuous concept measures as Regression Concept Vectors (RCVs) in the activation space of a layer to exploit network sensitivity to increasing values of a given concept measure. Expand
Resolving challenges in deep learning-based analyses of histopathological images using explanation methods
TLDR
This work shows how heatmaps generated by explanation methods allow to resolve common challenges encountered in deep learning-based digital histopathology analyses, and advocate pixel-wise heatmaps, which offer a more precise and versatile diagnostic instrument. Expand
Weakly Supervised Prostate Tma Classification Via Graph Convolutional Networks
TLDR
This work proposes a weakly-supervised approach for grade classification in tissue micro-arrays (TMA) using graph convolutional networks (GCNs), in which the spatial organization of cells as a graph to better capture the proliferation and community structure of tumor cells. Expand
Towards the Augmented Pathologist: Challenges of Explainable-AI in Digital Pathology
TLDR
Some (incomplete) research issues which should be addressed in an integrated and concerted effort for paving the way towards the augmented pathologist are described. Expand
Looking Under the Hood: Deep Neural Network Visualization to Interpret Whole-Slide Image Analysis Outcomes for Colorectal Polyps
TLDR
This work proposes a deep-learning-based image analysis approach that not only can accurately classify different types of polyps in whole-slide images, but also generates major regions and features on the slide through a model visualization approach. Expand
Histographs: Graphs in Histopathology
TLDR
This work proposes to classify cancers using graph convolutional networks (GCNs) by modeling a tissue section as a multi-attributed spatial graph of its constituent cells, and demonstrates the utility of this approach by obtaining classification accuracy that is competitive with CNNs. Expand
Cell nuclei attributed relational graphs for efficient representation and classification of gastric cancer in digital histopathology
TLDR
The proposed graph-based method provides a promising way for computer-based analysis of histopathological images of gastric cancer and is compared with state of the art methods. Expand
A Multi-Organ Nucleus Segmentation Challenge
TLDR
Several of the top techniques compared favorably to an individual human annotator and can be used with confidence for nuclear morphometrics as well as heavy data augmentation in the MoNuSeg 2018 challenge. Expand
CGC-Net: Cell Graph Convolutional Network for Grading of Colorectal Cancer Histology Images
TLDR
A novel cell-graph convolutional neural network (CGC-Net) that converts each large histology image into a graph, where each node is represented by a nucleus within the original image and cellular interactions are denoted as edges between these nodes according to node similarity. Expand
A Deep Learning Architecture for Image Representation, Visual Interpretability and Automated Basal-Cell Carcinoma Cancer Detection
TLDR
A deep learning architecture for automated basal cell carcinoma cancer detection that integrates image representation learning, image classification and result interpretability is presented and Experimental results show that the representation learned from a large histology image data set has the best overall performance. Expand
...
1
2
3
...