Physical & Psychophysical Evaluation of a New Digital Specimen Radiography System for Use in Mammography Elizabeth Krupinski, PhD Hans Roehrig, PhD University of Arizona William Schempp, PhD MedOptics Corporation

Rationale w Specimen radiography of excised breast tissue is critical to ensure that an appropriate tissue sample has been obtained or that complete removal of the lesion has been successful.

Objectives w To physically characterize the new high- resolution digital specimen radiography system from MedOptics Corporation. w To evaluate human observer detection performance using images acquired with the new system.

The Digital System w Camera 1024 x 1024, 1” x 1” large area high-res CCD 2:1 fiber optic taper bonded to CCD (2”x2” FOV) standard Lanex Regular screen coupled to large end of taper image data digitized at 500 Kpix/sec, 12-bits w Faxitron MICRO 50 x-ray source kVp range 1-90 sec time selection SID = 50 cm

Faxitron HVL & Eeff

Camera X-radiation Response Exposure [mR] C o r r e c t e d M e a n [ A D U ] o r R M S [ A D U ] Camera Response (Mean and Stdv (rms) in Digital Units [ADU]) to X-radiation (mR) 30 kVp 20 kVp 30 kVp 20 kVp 25 kVp Corrected Mean Curve Fit for 30 kVp log(Y) = *log(X) Corrected RMS Curve Fit for 30 kVp log(Y) = *log(X)

Camera Dark Signal & Noise Time [sec] M e a n o r S t d e v [ G L ] Dark Signal Dark Noise Uncorrected Dark-Signal and Dark-Noise of Uncooled CCD as Function of Time

Noise Power Spectrum 25 kVp Spatial Frequency (lp/mm) N o i s e P o w e r ( m m ^ 2 ) Noise Power Spectra for Exposures at 25 kVp Exposure Time 1 sec Exposure Time 4 sec

Noise Power Spectrum 30 kVp Spatial Frequency (lp/mm) N o i s e P o w e r ( m m ^ 2 ) Noise Power Spectra for Exposures at 30 kVp Exposure Time 1 sec Exposure Time 3 sec

Comparison NPS & Square MTF Spatial Frequency (lp/mm) N o i s e P o w e r ( m m ^ 2 ) o r M T F ^ 2 Scaled MTF^2 from Curve Fit Noise Power Spectrum 25 kVp, 4 sec Noise Power Spectrum and MTF for Specimen Radiography System

MTF Spatial Frequency (lp/mm) M T F ( r e l. U n i t s ) MTF of the CCD-based Camera of the MedOptics Specimen Radiography System; derived from the Fourier Transform of profiles of x-ray images of a 10 micron wide x-ray slit ("Line-Spread-Function") Dashed Line 25 kVp Solid Line 30 kVp Polynomial Coefficients Curve Fit: Polynomial of degree 2 Degree 0: Degree 1: Degree 2:

MTF from Square Wave Response Spatial Frequency (lp/mm) M T F ( r e l. U n i t s ) MTF of the CCD-based Camera of the MedOptics Specimen Radiography System; derived from the Fourier Transform of x-ray images of a lead-bar pattern ("Squarewave Response") Dashed Line 25 kVp Solid Line 30 kVp

Observer Performance Study w 2 CDMAM contrast-detail phantoms Commercially available large version Small experimental version w Imaging conditions Traditional plain film Fuji Computed Radiography (CR) MedOptics Digital System w 25 kV & 30 High (3600 GL), medium (2400 GL) & low (1200 GL) exposures

Film Image Optical Densities

Viewing Conditions w Traditional film & Fuji CR film standard 1100 ft-L extraneous light masked w Fuji CR digital & MedOptics Digital DataRay DR x ft-L Perceptually linearized with Barten curve

Observers & Task w 10 undergrad students in each condition w On outline of phantom indicate location of all disks visible in the phantom(s) w Window/Level & zoom available on monitor w Ambient lights turned off in all conditions

Large Phantom Results F = 68.44, p <

Small Phantom Results F = 52.62, p <

Performance Summary w Increases monotonically low to high exposures w 25 kV > 30 kV w MedOptics Digital System Best w Traditional plain film worst w Fuji CR monitor & film in middle w Minor differences between phantoms

Conclusions w MedOptics Digital Biopsy System yields high detection performance w Currently mammographers image core biopsy samples with either traditional systems or stereotactic units Traditional takes too long Stereotactic requires patient to get out of unit to image sample w Digital avoids these problems & is fast