A controlled evaluation of tuned-aperture computed tomography applied to digital spot mammography

The purpose of this work was to compare the detection accuracy of 3-dimensi onal (3D) modalities of tuned-aperture computed tomography (TACT) with that of conventional 2-dimensional (2D) digital spot mammograms. A standardized mammographic phantom was placed beneath cadaveric breast tissues of varyin g densities. Five radiologists were asked to detect as many objects (specks , fibers, and low-contrast masses) as possible from 90 displays in a contro lled and factorially balanced multivariate experiment. Radiographic exposur e was varied systematically, and projections were averaged to ensure stocha stic comparability. Scores were weighted to eliminate task-specific bias an d were analyzed by multivariate analyses of variance. All display modalitie s based on the linear application of the 3D TACT reconstruction method yiel ded significantly higher detection scores for all tasks than did convention al 2D digital spot mammography, which served as the scientific control moda lity. This effect was found to be statistically significant (P < .001) in s pite of significant variations between tissues (P < .001), observers (P < . 001), and exposures (P < .01). TACT may be a promising alternative or enhan cement to conventional 2D digital mammography for Tasks well simulated by t his experimental design. Copyright (C) 2000 by W.B. Saunders Company.