Moderated Poster

Poster, Podium & Video Sessions

MP18-13: Texture Analysis of enhancing, non-lipid containing solid renal masses: Differentiation of Malignant from Benign Renal Tumors.

Friday, May 12
3:30 PM - 5:30 PM
Location: BCEC: Room 153

Presentation Authors: Bino Varghese, Vinay Duddalwar*, Frank Chen, Darryl Hwang, Steven Cen, Bhushan Desai, Gangning Liang, Mihir Desai, Sameer Chopra, Manju Aron, Monish Aron, Inderbir Gill, Los Angeles, CA

Introduction: Contrast Enhanced Computed Tomography (CECT) is the most common modality of imaging a renal mass. While metrics including pixel enhancement have been described for differentiation of various types of tumors, we describe an additional technique of texture analysis.

Methods: In this Institutional Review Board (IRB) approved, Health Insurance Portability and Accountability Act (HIPAA) compliant, retrospective study, we identified 136 patients with solid, non-lipid containing enhancing renal tumors based on post-surgical pathology examination (94 Malignant, 42 Benign). Here, we test the feasibility using textural biomarkers, to objectively quantify and differentiate the textural heterogeneity of malignant subtypes, here, clear cell renal carcinoma, papillary renal carcinoma, and chromophobe from, benign subtypes, here, oncocytoma and lipid poor angiomyolipoma, using standard-of-care contrast-enhanced computed tomography (CECT) images.

Results: Three sets of stepwise logistic regression were used to select the best predictor among all candidate predictors from 2D GLCM, 3D GLCM and spectral (Table 1). The discrimination power gain from spectral metrics in addition to 2D and 3D GLCM combined was assessed using a one-degree freedom chi square test when comparing the area under the curve between the full model and the model without spectral metrics.
The full model with 2D, 3D GLCM and spectral predictors yielded an AUC of 0.92 (95% CI: 0.87-0.96), while the model with 2D and 3D only already reached almost the same AUC. The difference between the two model was less than 0.01 (p=0.89) (Figure 1).


Conclusions: CECT-based texture metrics can differentiate between malignant- and benign-renal tumors, with 2D and 3D GLCM metrics providing the most information for segregating malignant from benign renal tumors. In combination with other metrics such as contrast enhancement, shape metrics etc., texture metrics, have the potential to improve patient management and help stratify renal tumors using prostate CECT.



Source Of Funding: This project has received funding from the Whittier Foundation. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Foundation.

Send Email for Sameer Chopra


Assets

MP18-13: Texture Analysis of enhancing, non-lipid containing solid renal masses: Differentiation of Malignant from Benign Renal Tumors.



Attendees who have favorited this

Send Email for Texture Analysis of enhancing, non-lipid containing solid renal masses: Differentiation of Malignant from Benign Renal Tumors.