Abstract
We present a novel method for evaluating the spatial correlation structure in two-dimensional (2D) mammograms and evaluate its merits for risk prediction. Two matched case-control studies were analyzed. Study 1 included women (N = 588 pairs) with mammograms acquired with either Hologic Selenia full field digital mammography (FFDM) units or Hologic Dimensions digital breast tomosynthesis units. Study 2 included women (N = 180 pairs) with mammograms acquired with a General Electric Senographe 2000D FFDM unit. Matching variables included age, HRT usage/duration, screening history, and mammography unit. The local autocorrelation function was determined with Fourier analysis and compared with template defined as 2D double-sided exponential function with one spatial extent parameter: n = 4, 12, 24, 50, 74, 100, and 124 defined in pixel widths. The difference between local correlation and template was gauged within a kernel with an adjustable parameter and summarized, producing two measures: the mean (mn+1), and standard (sn+1). Both adjustable parameters were varied in Study 1. Select measures that produced significant associations with breast cancer were translated to Study 2. Breast cancer associations were evaluated with conditional logistic regression, adjusted for body mass index and ethnicity. Odds ratios (ORs) were estimated as per standard increment with 95% confidence intervals (CIs). Two measures were selected for breast cancer association analysis in Study 1: m75 and s25. Both measures revealed significant associations with breast cancer: OR = 1.45 (1.23, 1.66) for m75 and OR = 1.30 (1.14, 1.49) for s25. When translating to Study 2, these measures also revealed significant associations: OR = 1.49 (1.12, 1.96) for m75 and OR = 1.34 (1.06, 1.69) for s25. Novel correlation metrics presented in this work revealed significant associations with breast cancer risk. This approach is general and may have applications beyond mammography.
