| (英) |
To develop a highly accurate differentiation method for malignant and benign tumors by precisely analyzing renal and renal tumors from multi-temporal 3D contrast-enhanced CT images. For this purpose, we will construct a database of 200 cases of kidney and renal tumors in multi-temporal contrast-enhanced CT images. Using this database, five types of kidney and renal tumors ( Clear cell renal cell carcinoma, Papillary renal cell carcinoma, Achoromatic renal cell carcinoma, Angiomyolipoma, Oncocytoma) were extracted with high speed and high accuracy using 3D U-Net, and the images were clustered by gender, age, volume, shape, type, non-contrast, arterial phase, portal vein phase, and equilibrium phase images. phase, arterial phase, portal vein phase, and equilibrium phase images. The circulatory dynamics of small-diameter renal tumors are analyzed to demonstrate a highly accurate method for differentiating small-diameter renal tumors. |