Automatic Lung Nodule Detection Using Deep Learning WEEK

Automatic Lung Nodule Detection Using Deep Learning WEEK 4: BAG OF FREQUENCIES IMPLEMENTATION REU STUDENT: WINONA RICHEY GRADUATE STUDENT: NAJI KHOSRAVAN PROFESSOR: DR. BAGCI

Bag of Frequencies: Function �Input whole image, nodule center coordinates, radius of nodule �Output Feature Vector: spectral signatures from intensity profiles �Overview: Take 2 D surfaces from 3 D voxel � Find number of surfaces to sample in X, Y, and Z directions Sample intensities � Taken counterclockwise Get spectral signatures with FFT � Take the first ½ of output due to symmetry Images: Ciompi, et al. 2015. Bag-of-Frequencies: A Descriptor of Pulmonary Nodules in Computed Tomography Images.

Bag of Frequencies Parameter Description Number of slices #of 2 D surfaces extracted from nodule Size of slices Important if sampling outside of radius Example 10 [4 slices in XY plane, 3 slices in YZ plane, 3 slices in ZX plane] 1. 2 slices will be 120% the size of the nodule Slice below is 40 x 40 Radius is 17 Each radius is shown in a separate color (RBG) in the image below Sample Radii Determines #and location of intensity profiles in relation to the radius [. 5, . 9, 1. 05] three intensity profiles will be taken at 50%, 90% and 105% of the radius Number of intensity Samples # of points sampled; determines angles b/t points of intensity profile 8 points will be taken every 45 degrees, counter clockwise (as shown; first sample is black, last is bright)

Frequency Analysis of Intensity Profiles of 2 D surfaces �Intensity values taken at regular angular intervals, Δω Intensity profiles stored as matrix rows �Fourier Transform applied to each intensity profile (below) Periodic signals have a discrete Fourier Transform (spectrum) �absolute value of Ψ represents a spectral signature Sample from 0 to M/2 due to symmetry of absolute value of the spectrum Images: Ciompi, et al. 2015. Bag-of-Frequencies: A Descriptor of Pulmonary Nodules in Computed Tomography Images.

Progress �Completion of 3 D implementation of Bag of Frequencies to obtain feature vectors Images show sampling pattern �Next Steps: Implement Taxonomic Indices to obtain feature vectors

References �Ciompi, F. , Jacobs, C. , Scholten, E. T. , Wille, M. W. , de Jong, P. A. , Prokop, M. , & van Ginneken, B. (2015). Bag-of-Frequencies: A Descriptor of Pulmonary Nodules in Computed Tomography Images. IEEE Transactions On Medical Imaging, 34(4), 962 -973.