Lesion Detectability in Contrast-Enhanced Breast CT Using Model Observers
S Lyu1*, A Hernandez2, C Abbey3, J Boone4, (1) University of California Davis, Davis, CA, (2) University of California Davis Medical Center, Sacramento, CA, (3) University of California Santa Barbara, Santa Barbara, CA, (4) University of California Davis Medical Center, Sacramento, CA
Presentations
TU-E-201-4 (Tuesday, 7/12/2022) 11:00 AM - 12:15 PM [Eastern Time (GMT-4)]
Room 201
Purpose: To quantitatively evaluate the improvement in lesion detectability due to contrast enhancement across lesion diameter, section thickness, and breast density using a pre-whitened matched filter (PWMF) model observer.
Methods: The relationship between iodine concentration and HU was measured using a rod phantom containing known iodine concentrations placed in a polyethylene breast phantom and scanned on a prototype breast CT (bCT) system. Mathematically generated spherical mass lesions of varying diameters (1, 3, 5, 9, 15 mm) and contrast enhancement levels (0, 7, 14, 21, 28 HU) were inserted at random locations in 20 actual patient bCT datasets. Images with varying thicknesses (0.39 to 11.64 mm) were generated by slice averaging. The volumetric glandular fraction (VGF) of the patient datasets was quantified using a previously reported algorithm. A PWMF was generated for combinations of lesion diameter, contrast enhancement, section thickness, and breast density. The PWMF detection performance was assessed using the area under the ROC curve for 200 lesions and non-lesions per breast and parameter combination.
Results: In lower density breasts (VGF < 30%), detection performance improves with the addition of contrast before plateauing around 7 ∆HU. In higher density breasts (VGF > 30%), 28 ∆HU is necessary to maximize detectability for lesions greater than 5 mm, and more contrast addition is necessary to maximize detectability for lesions smaller than 5 mm. Detection performance generally improves as section thickness decreases, with the exception of the unenhancing 1 mm lesion, which is best detected using a 2 mm section thickness.
Conclusion: This study demonstrates quantitatively how lesion detectability improves with contrast enhancement in bCT across relevant parameters. Optimal injected contrast agent volume and display section thickness for detection is dependent on the size of the lesion being detected as well as breast density.
Funding Support, Disclosures, and Conflict of Interest: Authors Boone, Hernandez, and Abbey have conflicts of interest. Author Boone is a stockholder for Izotropic Imaging Corp. of Canada. Funding support: R01 CA181081
Keywords
Breast, Cone-beam CT, Contrast
Taxonomy
IM- Breast X-Ray Imaging: Breast CT
Contact Email