An official website of the United States government
Automatic histopathological Whole Slide Image (WSI) analysis for cancer classification has been highlighted along with the advancements in microscopic imaging techniques, since manual examination and diagnosis with WSIs are time- and cost-consuming. Recently, deep convolutional neural networks have succeeded in histopathological image analysis. However, despite the success of the development, there are still opportunities for further enhancements. In this paper, we propose a novel cancer texture-based deep neural network (CAT-Net) that learns scalable morphological features from histopathological WSIs. The innovation of CAT-Net is twofold: (1) capturing invariant spatial patterns by dilated convolutional layers and (2) improving predictive performance while reducing model complexity. Moreover, CAT-Net can provide discriminative morphological (texture) patterns formed on cancerous regions of histopathological images comparing to normal regions. We elucidated how our proposed method, CAT-Net, captures morphological patterns of interest in hierarchical levels in the model. The proposed method out-performed the current state-of-the-art benchmark methods on accuracy, precision, recall, and F1 score.
more »
« less
Genetic mutation and biological pathway prediction based on whole slide images in breast carcinoma using deep learning
Abstract Breast carcinoma is the most common cancer among women worldwide that consists of a heterogeneous group of subtype diseases. The whole-slide images (WSIs) can capture the cell-level heterogeneity, and are routinely used for cancer diagnosis by pathologists. However, key driver genetic mutations related to targeted therapies are identified by genomic analysis like high-throughput molecular profiling. In this study, we develop a deep-learning model to predict the genetic mutations and biological pathway activities directly from WSIs. Our study offers unique insights into WSI visual interactions between mutation and its related pathway, enabling a head-to-head comparison to reinforce our major findings. Using the histopathology images from the Genomic Data Commons Database, our model can predict the point mutations of six important genes (AUC 0.68–0.85) and copy number alteration of another six genes (AUC 0.69–0.79). Additionally, the trained models can predict the activities of three out of ten canonical pathways (AUC 0.65–0.79). Next, we visualized the weight maps of tumor tiles in WSI to understand the decision-making process of deep-learning models via a self-attention mechanism. We further validated our models on liver and lung cancers that are related to metastatic breast cancer. Our results provide insights into the association between pathological image features, molecular outcomes, and targeted therapies for breast cancer patients.
more »
« less
- Award ID(s):
- 1747778
- PAR ID:
- 10307467
- Publisher / Repository:
- Nature Publishing Group
- Date Published:
- Journal Name:
- npj Precision Oncology
- Volume:
- 5
- Issue:
- 1
- ISSN:
- 2397-768X
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract BackgroundCrop improvement through cross-population genomic prediction and genome editing requires identification of causal variants at high resolution, within fewer than hundreds of base pairs. Most genetic mapping studies have generally lacked such resolution. In contrast, evolutionary approaches can detect genetic effects at high resolution, but they are limited by shifting selection, missing data, and low depth of multiple-sequence alignments. Here we use genomic annotations to accurately predict nucleotide conservation across angiosperms, as a proxy for fitness effect of mutations. ResultsUsing only sequence analysis, we annotate nonsynonymous mutations in 25,824 maize gene models, with information from bioinformatics and deep learning. Our predictions are validated by experimental information: within-species conservation, chromatin accessibility, and gene expression. According to gene ontology and pathway enrichment analyses, predicted nucleotide conservation points to genes in central carbon metabolism. Importantly, it improves genomic prediction for fitness-related traits such as grain yield, in elite maize panels, by stringent prioritization of fewer than 1% of single-site variants. ConclusionsOur results suggest that predicting nucleotide conservation across angiosperms may effectively prioritize sites most likely to impact fitness-related traits in crops, without being limited by shifting selection, missing data, and low depth of multiple-sequence alignments. Our approach—Prediction of mutation Impact by Calibrated Nucleotide Conservation (PICNC)—could be useful to select polymorphisms for accurate genomic prediction, and candidate mutations for efficient base editing. The trained PICNC models and predicted nucleotide conservation at protein-coding SNPs in maize are publicly available in CyVerse (https://doi.org/10.25739/hybz-2957).more » « less
-
Approximately 75% of diagnosed breast cancer tumors are estrogen-receptor-positive tumors and are associated with a better prognosis due to response to hormonal therapies. However, around 40% of patients relapse after hormonal therapies. Genomic analysis of gene expression profiles in primary breast cancers and tamoxifen-resistant cell lines suggested the potential role of miR-489 in the regulation of estrogen signaling and development of tamoxifen resistance. Our in vitro analysis showed that loss of miR-489 expression promoted tamoxifen resistance, while overexpression of miR-489 in tamoxifen-resistant cells restored tamoxifen sensitivity. Mechanistically, we found that miR-489 is an estrogen-regulated miRNA that negatively regulates estrogen receptor signaling by using at least the following two mechanisms: (i) modulation of the ER phosphorylation status by inhibiting MAPK and AKT kinase activities; (ii) regulation of nuclear-to-cytosol translocation of estrogen receptor α (ERα) by decreasing p38 expression and consequently ER phosphorylation. In addition, miR-489 can break the positive feed-forward loop between the estrogen-Erα axis and p38 MAPK in breast cancer cells, which is necessary for its function as a transcription factor. Overall, our study unveiled the underlying molecular mechanism by which miR-489 regulates an estrogen signaling pathway through a negative feedback loop and uncovered its role in both the development of and overcoming of tamoxifen resistance in breast cancers.more » « less
-
Known genes in the breast cancer study literature could not be confirmed whether they are vital to breast cancer formations due to lack of convincing accuracy, although they may be biologically directly related to breast cancer based on present biological knowledge. It is hoped vital genes can be identified with the highest possible accuracy, for example, 100% accuracy and convincing causal patterns beyond what has been known in breast cancer. One hope is that finding gene-gene interaction signatures and functional effects may solve the puzzle. This research uses a recently developed competing linear factor analysis method in differentially expressed gene detection to advance the study of breast cancer formation. Surprisingly, 3 genes are detected to be differentially expressed in TNBC and non-TNBC (Her2, Luminal A, Luminal B) samples with 100% sensitivity and 100% specificity in 1 study of triple-negative breast cancers (TNBC, with 54 675 genes and 265 samples). These 3 genes show a clear signature pattern of how TNBC patients can be grouped. For another TNBC study (with 54 673 genes and 66 samples), 4 genes bring the same accuracy of 100% sensitivity and 100% specificity. Four genes are found to have the same accuracy of 100% sensitivity and 100% specificity in 1 breast cancer study (with 54 675 genes and 121 samples), and the same 4 genes bring an accuracy of 100% sensitivity and 96.5% specificity in the fourth breast cancer study (with 60 483 genes and 1217 samples). These results show the 4-gene-based classifiers are robust and accurate. The detected genes naturally classify patients into subtypes, for example, 7 subtypes. These findings demonstrate the clearest gene-gene interaction patterns and functional effects with the smallest numbers of genes and the highest accuracy compared with findings reported in the literature. The 4 genes are considered to be essential for breast cancer studies and practice. They can provide focused, targeted researches and precision medicine for each subtype of breast cancer. New breast cancer disease types may be detected using the classified subtypes, and hence new effective therapies can be developed.more » « less
-
Stabler, Cherie L. (Ed.)The unavailability of reliable models for studying breast cancer bone metastasis is the major challenge associated with poor prognosis in advanced-stage breast cancer patients. Breast cancer cells tend to preferentially disseminate to bone and colonize within the remodeling bone to cause bone metastasis. To improve the outcome of patients with breast cancer bone metastasis, we have previously developed a 3D in vitro breast cancer bone metastasis model using human mesenchymal stem cells (hMSCs) and primary breast cancer cell lines (MCF-7 and MDAMB231), recapitulating late-stage of breast cancer metastasis to bone. In the present study, we have tested our model using hMSCs and patient-derived breast cancer cell lines (NT013 and NT023) exhibiting different characteristics. We investigated the effect of breast cancer metastasis on bone growth using this 3D in vitro model and compared our results with previous studies. The results showed that NT013 and NT023 cells exhibiting hormone-positive and triple-negative characteristics underwent mesenchymal to epithelial transition (MET) and formed tumors in the presence of bone microenvironment, in line with our previous results with MCF-7 and MDAMB231 cell lines. In addition, the results showed upregulation of Wnt-related genes in hMSCs, cultured in the presence of excessive ET-1 cytokine released by NT013 cells, while downregulation of Wnt-related genes in the presence of excessive DKK-1, released by NT023 cells, leading to stimulation and abrogation of the osteogenic pathway, respectively, ultimately mimicking different types of bone lesions in breast cancer patients.more » « less
