PURPOSE: To assess the usefulness of fractal geometry in quantitatively evaluating the convergence of peripheral vessels on peripheral lung tumors in maximum intensity projection (MIP) images. MATERIALS AND METHODS: We studied the MIP images of 34 pathologically proved small peripheral lung tumors (lung cancer in 21, hamartoma in 13) in 34 patients. To obtain MIP images, spiral CT (SOMATOM PLUS; Siemens) was performed during a single breath hold (24-second scan time, 2-mm section thickness, and 2 mm/sec table feed time, reconstructed at 1-mm increments). To evaluate the convergence of the peripheral vessels and bronchi towards the tumor, we fixed a region of interest (ROI) on the hilar side of the lung tumor, parallel to the chest wall, which consisted of 64 x 64 square pixels, in the images that divided at the center of the window width. We counted the overlapping pixels by the two-dimensional box-counting method and obtained fractal dimensional data on lung cancers and hamartomas. RESULTS: There was a statistically significant difference in the fractal dimension (D) between lung cancers (D = 1.81 +/- 0.13) and hamartomas (D = 1.67 +/- 0.10) (P = 0.0067). CONCLUSION: Fractal geometry could be useful in the diagnosis of small peripheral lung tumors.
PURPOSE: To assess the usefulness of fractal geometry in quantitatively evaluating the convergence of peripheral vessels on peripheral lung tumors in maximum intensity projection (MIP) images. MATERIALS AND METHODS: We studied the MIP images of 34 pathologically proved small peripheral lung tumors (lung cancer in 21, hamartoma in 13) in 34 patients. To obtain MIP images, spiral CT (SOMATOM PLUS; Siemens) was performed during a single breath hold (24-second scan time, 2-mm section thickness, and 2 mm/sec table feed time, reconstructed at 1-mm increments). To evaluate the convergence of the peripheral vessels and bronchi towards the tumor, we fixed a region of interest (ROI) on the hilar side of the lung tumor, parallel to the chest wall, which consisted of 64 x 64 square pixels, in the images that divided at the center of the window width. We counted the overlapping pixels by the two-dimensional box-counting method and obtained fractal dimensional data on lung cancers and hamartomas. RESULTS: There was a statistically significant difference in the fractal dimension (D) between lung cancers (D = 1.81 +/- 0.13) and hamartomas (D = 1.67 +/- 0.10) (P = 0.0067). CONCLUSION: Fractal geometry could be useful in the diagnosis of small peripheral lung tumors.
Authors: Javed Iqbal; Ranjitkumar Patil; Vikram Khanna; Anurag Tripathi; Vandana Singh; M A I Munshi; Rahul Tiwari Journal: J Family Med Prim Care Date: 2020-05-31