An Investigation into the Use of Spatially-Filtered Fourier Transforms to Classify Mammary Lesions.
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OHIO SCHOOL OF ENGINEERING
Pagination or Media Count:
Lesions of the female breast are analyzed using one- and two-dimensional Fourier transform techniques and power spectrum analysis. Attempts are made to perform a computer classification between lesioned and non-lesioned breast tissue. Additionally, attempts are made to ascertain the interclass difference in Fourier space between lesioned breast tissue which would enable accurate computer classification of benign and malignant lesions. Low frequency spatial filtering techniques are used in conjunction with the Euclidean distance and a nearest neighbor clustering metric in an attempt to accurately separate the various classes. Some success was observed when using the lower frequencies generated by the one-dimensional Fourier transform to separate benign and malignant breast tissue. However, no success was achieved when using two-dimensional Fourier transform and power spectrum analysis. Author
- Medicine and Medical Research