RSNA 2013 

Abstract Archives of the RSNA, 2013


LL-GUS-SU2A

Image Texture Analysis on Primary Tumor Can Stratify Differences in Overall Survival, Distant Metastases, Tumor Stage, and Fuhrman Grade in Patients with Clear Cell Renal Carcinoma: Results of a Feasibility Study

Scientific Informal (Poster) Presentations

Presented on December 1, 2013
Presented as part of LL-GUS-SUA: Genitourinary/Uroradiology - Sunday Posters and Exhibits (12:30pm - 1:00pm)

Participants

Raghunandan Vikram MBBS, FRCR, Presenter: Nothing to Disclose
Payel Ghosh, Abstract Co-Author: Nothing to Disclose
Pheroze Tamboli MD, Abstract Co-Author: Nothing to Disclose
Arvind Rao, Abstract Co-Author: Nothing to Disclose

PURPOSE

To identify textural features on contrast-enhanced CT derived on primary tumors associated with overall survival, distant metastases, tumor stage and Fuhrman grade in patients with Clear cell Renal carcinoma (CRCC).

METHOD AND MATERIALS

The study was performed on CT scans of 39 CRCC patients from The Cancer Genome Atlas (TCGA) clear cell carcinoma database. The primary tumor was segmented using the Medical Image Interaction Tool Kit (mitk.org). 73684 imaging features were extracted on each phase (non-contrast, nephrographic, cortico-medullary and excretory phases) on MATLAB. 3-D textural algorithms (Laws’, Wavelet, Haralick), volumetric features and ratios at different image resolutions. These were correlated with overall survival, metastases, tumor stage and Fuhrman grade using the Wilcoxon-Ranksum test. Cox proportional hazards model was used for obtaining survival-associated univariate p-values. Multiple testing corrections for p-values were done using Benjamini-Hochberg FDR correction.

RESULTS

The number of features that could stratify survival were: 2099 in non-contrast, 1424 in cortico-medullary, 551 in nephrographic and 345 in excretory phase. The number of features that could stratify presence of metastases (M1vsM0) were 1935 in non-contrast, 312 in cortico-medullary, 1468 in nephrographic and 3822 in excretory phase. Only nephrographic and excretory phases had 273 significant features each after FDR correction (q-value < 0.05). The number of features that could stratify T stage (T1&T2 versus T3&4) were: 10453 in non-contrast phase, 2919 in cortico-medullary, 5103 in nephrographic and 2936 in excretory phase. 290 nephrographic and 180 excretory phase features could be ratified after FDR correction (q-value < 0.05). The number of image features that could stratify Fuhrman grade were: 18467 in non-contrast, 12081 in cortico-medullary, 3086 in the nephrographic and 2936 in the excretory phase.

CONCLUSION

Image texture features are associated with overall survival, distant metastases, tumor stage and Fuhrman grade in patients with clear cell renal carcinoma. Non-contrast phase had strongest correlation with survival and Fuhrman grade. Features from nephrographic and excretory phases correlated strongly with metastasis and TNM stage in this initial analysis.

CLINICAL RELEVANCE/APPLICATION

Image texture analysis on routine staging CT is a promising risk assessment tool in patients with clear cell renal carcinoma. 

Cite This Abstract

Vikram, R, Ghosh, P, Tamboli, P, Rao, A, Image Texture Analysis on Primary Tumor Can Stratify Differences in Overall Survival, Distant Metastases, Tumor Stage, and Fuhrman Grade in Patients with Clear Cell Renal Carcinoma: Results of a Feasibility Study.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13024424.html