|
KMID : 1147120160220010015
|
|
Journal of the Korean Society of Imaging Informatics in Medicine
2016 Volume.22 No. 1 p.15 ~ p.22
|
|
|
Texture analysis-based Computer-aided Differential Diagnosis of CT Images for Renal Mass Characterization
|
|
Park Seung-Hyun
Lee Seung-Soo
Jung Dae-Chul
Lee Han-Sang
Hong Helen
Bae Hee-Jin
Kim Dae-Keun
Rha Koon-Ho
|
|
|
Abstract
|
|
|
Purpose: To determine the accuracy of texture analysis-based computer aided diagnosis to differentiate clear cell type renal cell cancer (less than 4 cm) from the renal cyst and angiomyolipoma on contrast-enhanced multi-detector computed tomography (MDCT) images.
Materials and Methods: In this retrospective study, pathologically proven cases of 10 clear cell type renal cell cancers, radiologically proven cases of 10 renal cysts and ten angiomyolipomas were included. For tumor segmentation, interactive graph cuts were adjusted, and texture features were extracted from these segmented tumors. Mean, standard deviation, skewness, kurtosis, and entropy were calculated for texture analysis. For comparison of texture analysis parameters, Kruskal-Wallis test and Dunn’s procedure as a second step were performed. For diagnostic performance evaluation between benign pathology and renal cell cancer, ROC analysis was performed.
Results: There was statistically significant difference in all measured textual parameters among three kinds of small renal mass (P-value < 0.05). The area under the receiver operating curve (AUC) was 0.85-1.00 to differentiate malignancy from benign pathology in this study population.
Conclusion: Texture analysis based computer aided diagnosis may be helpful to differentiate small renal masses (lower than 4 cm) objectively and quantitatively.
|
|
|
KEYWORD
|
|
|
Computed tomography (CT), Small Renal Mass (SRM), Renal cell carcinoma (RCC), Texture analysis, Comp
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|
|