Child pages
  • Integration of CT-based Qualitative and Radiomic Features with Proteomic Variables in Patients with High-Grade Serous Ovarian Cancer: An Exploratory Analysis
Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 3 Next »

Data Citation

Lucian Beer1*, Hilal Sahin1*,  Ivana Blazic2, Hebert Alberto Vargas3, Harini Veeraraghavan4,  Justin Kirby, Brenda Fevrier-Sullivan, John Freymann, Carl Jaffe, Thomas Conrads, George Maxwell, Kathleen Darcy, Erich Huang*, Evis Sala1* Radiogenomics of Clear Cell Renal Cell Carcinoma: Preliminary Findings of The Cancer Genome Atlas-Renal Cell Carcinoma (TCGA-RCC) Research Group. The Cancer Imaging Archive. Data DOI Here

Description

PURPOSE:

To investigate the association between CT imaging traits and texture metrics with proteomic data in patients with high grade serous ovarian cancer (HGSOC).

MATERIALS AND METHODS:

This retrospective multi-institutional study enrolled 20 patients with HGSOC prior to primary cytoreductive surgery. Two readers independently assessed the contrast enhanced computed tomography (CT) and extracted 33 imaging traits. In addition all sites of suspected HGSOC were manually segmentedand grey-level correlation matrix-based texture features were computed from each tumor site. Three texture features representing inter-site tumour heterogeneity were used for further analysis. Combined analysis of transcriptomics proteomics was used to identify stably expressed proteins between primary tumour sites and metastasis. The correlation between the different imaging traitsand texture features with measurement of protein abundance were assessedusing Kendall tau rank correlation coefficient and Mann-Whitney U test, whereby the area under the receiver operating characteristic curve (AUC) was reported as a metric of strength and direction of the association. P values < 0.05 were considered significant. 

RESULTS:

Expression of eight proteins were significantly associated with CT-based imaging traits. The strongest positive correlations was observed between peritoneal diseasein the liver/ right upper quadrant (P<0.001, AUC=0.940). Four proteins were associated with texture features representing inter-site tumor heterogeneity, the strongest positive correlation was between protein abundance of GSTM1 and the feature cluster dissimilarity. 

CONCLUSIONS:

This study provides first insights on potentially strong associations between standard of care CT imaging traits and CT-based texture measures of tumour burden inter-site heterogeneity and abundance of several associated proteins.

Publication Citation

Lucian Beer1*, Hilal Sahin1*,  Ivana Blazic2, Hebert Alberto Vargas3, Harini Veeraraghavan4,  Justin Kirby, Brenda Fevrier-Sullivan, John Freyman, Carl Jaffe, Thomas Conrads, George Maxwell, Kathleen Darcy, Erich Huang*, Evis Sala1*

Integration of CT-based Qualitative and Radiomic Features with Proteomic Variables in Patients with High-Grade Serous Ovarian Cancer: An Exploratory Analysis (PUBLICATION DOI HERE)

Download

  • 20 patients
  • Image Feature:  (Summary table provided in publication)

    36 image features, including the 33 radiologists-scored imaging traits used in the study of Vargas et al such as lesion size and laterality, locations of peritoneal disease, nodal stations involved, and locations of metastases, and three computer-extracted texture metrics were obtained[11].

  • Image Data (subset of TCGA-OV)
  • Segmentations - Segmentation was performed using 3DSlicer [13] by tracing the contour of each lesion on each slice to produce the volume of interests (VOI). Voxel-wise Haralick textures (energy, entropy, contrast, and homogeneity) were computed from within the manually delineated VOIs using in-house software implemented in C++ using the Insight ToolKit[14]. Site specific sub-regions were computed by voxel-wise clustering of the Haralick textures using kernel K-means method[15]. Following clustering, tumor sites were divided into distinct sub-regions and summarized using average of Haralicktexture measures of all voxels within that region.
  • Proteomic Data - protein relative abundance measurements  (link to paper and/or provide spreadsheet of data used)

 

  • No labels