RENAL TUMOR QUANTIFICATION AND CLASSIFICATION IN TRIPLE-PHASE CONTRAST-ENHANCED ABDOMINAL CT

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Kidney cancer affects approximately 250,000 people in the USA.
• Current manual tumor size measurements are time-consuming and prone to significant variability.
• Objective and reproducible tumor assessment is crucial for effective management.

Purpose of the Study:

• To develop and evaluate a computer-assisted radiology tool for renal tumor assessment.
• To improve the accuracy and efficiency of tumor diagnosis and treatment monitoring.
• To enable quantitative analysis of tumor characteristics, including size, volume, and enhancement.

Main Methods:

• Development of a computer-assisted tool utilizing anisotropic diffusion, fast-marching, and geodesic level-sets.
• Incorporation of a novel statistical refinement step for adaptive lesion shape analysis.
• Quantification of 3D tumor size, volume, and enhancement for serial management.

Main Results:

• Semi-automated quantifications demonstrated disparity within inter-observer variability limits compared to manual measurements.
• Automated tumor classification achieved significant separation between cysts, von Hippel-Lindau syndrome (VHL) lesions, and hereditary papillary renal carcinomas (HPRC) (p < 0.004).
• The tool enables quantitative analysis and serial management of renal tumors.

Conclusions:

• The proposed computer-assisted radiology tool offers a reliable method for renal tumor assessment.
• This technology can enhance diagnostic accuracy and streamline treatment management for kidney cancer.
• The automated classification capability shows promise for differentiating various renal lesion types.