Role of Computer-Aided Detection in Very Small Screening Detected Invasive Breast Cancers

Abstract

This study aims to assess computer-aided detection (CAD) performance with full-field digital mammography (FFDM) in very small (equal to or less than 1 cm) invasive breast cancers. Sixty-eight invasive breast cancers less than or equal to 1 cm were retrospectively studied. All cases were detected with FFDM in women aged 49-69 years from our breast cancer screening program. Radiological characteristics of lesions following BI-RADS descriptors were recorded and compared with CAD sensitivity. Age, size, BI-RADS classification, breast density type, histological type of the neoplasm, and role of the CAD were also assessed. Per-study specificity and mass false-positive rate were determined by using 100 normal consecutive studies. Thirty-seven (54.4 %) masses, 17 (25 %) calcifications, 6 (8.8 %) masses with calcifications, 7 (10.3 %) architectural distortions, and 1 asymmetry (1.5 %) were found. CAD showed an overall sensitivity of 86.7 % (masses, 86.5 %; calcifications, 100 %; masses with calcifications, 100 %; and architectural distortion, 57.14 %), CAD failed to detect 9 out of 68 cases: 5 of 37 masses, 3 of 7 architectural distortions, and 1 of 1 asymmetry. Fifteen out of 37 masses were hyperdense, and all of them were detected by CAD. No association was seen among mass morphology or margins and detectability. Per-study specificity and CAD false-positive rate was 26 % and 1.76 false marks per study. In conclusion, CAD shows a high sensitivity and a low specificity. Lesion size, histology, and breast density do not influence sensitivity. Mammographic features, mass density, and thickness of the spicules in architectural distortions do influence.

Interpretation time for screening mammography as a function of the number of computer-aided detection marks

“…þ β 4 ðBI − RADS ij ¼ 1Þ þ β 5 ðdensity ij ¼ "Extremely"Þ þ β 6 ðdensity ij ¼ "Heterogeneous"Þ þ β 7 ðdensity ij ¼ "Fibroglandular"Þ þ ε ij ; (1) where "ln" is the natural logarithmic function; β 0i is the intercept term for reader i; CAD ij is the number of CAD marks with corresponding reader-specific coefficient β 1i ; images ij is the number of images; BI-RADS ij is the BIRADS category where, e.g., ðBI-RADS ij ¼ 0Þ is an indicator variable that takes the value 1 if BI-RADS ij ¼ 0 and otherwise is 0; density ij is the breast density category; and ε ij is the error term. We assume that the ðβ 0i ; β 1i Þ pairs have a bivariate normal distribution across readers with means β 0 and β 1 , variances σ 2 β 0i and σ 2 β 1i , and covariance σ β 0 β 1 , and we assume that the ε ij are independent and normally distributed with zero mean and variance σ 2 ε . Note that only the intercept and CAD coefficient vary across readers; i.e., for the other independent variables, the corresponding regression coefficients are assumed constant across readers.…”

“…CAD is notorious for creating a large number of false-positive marks. 2 – 4 Its overall value in screening has been questioned, with a study in 2011 finding that in a large cohort of women undergoing screening mammography, the use of CAD decreased specificity for breast cancer detection but did not have a statistically significant improvement in detection rate or decrease in breast cancer stage, size, or nodal status. 5 A later study by an overlapping group of authors of another large cohort of patients showed CAD to increase detection of ductal carcinoma in situ and to allow detection of invasive breast cancer at earlier stages.…”