Switching from Analog to Digital Mammography

What are current FFDM technologies? How do you measure image quality in FFDM? What do you gain with FFDM? What do you lose? Many FFDM technologies avail: DR direct outperforms DR indirect and both outperform CR Image quality: Measured with MTF (resolution) and DQE (overall performance) Digital typically uses higher energy Dose reduction FFDM better: Increase in dynamic range, digital viewing (winlev/mag) Reduction in repeats, improved diagnosis in dense tissue Increased energy and image quality allow the option to reduce dose Immediate viewing, no film process, reduces repeats Advanced applications (CAD, contrast enhanced, teleradiology, tomo, etc.) FFDM worse: Greater cost (300-450k vs. <100k for SFM) Similar performance for some if not most women Pros/cons of FFDM FFDM technologies FFDM image quality FFDM technologies FFDM image quality Digital detector Covers full field of breast Early small-field or spot-view systems had around 10x10 cm for diagnostic or biopsy purposes only Performance optimization Independent acquisition, display, storage Greater dynamic range 1000:1 vs. 40:1 Eliminate repeats due to over/under-exposure Better Dx from ­improved penetration of dense breasts Faster throughput (?) Reduced dose “Advanced imaging” applications Computer-aided detection, teleradiology, quantitation, tomosynthesis, dual energy Greater cost Lower spatial resolution “Electronic magnification” is not the real thing! Significant advantage only for certain women: Peri/premenopausal, younger (<50 yrs), dense breasts Workflow issues Comparison with SFM priors Some sites still printing to hardcopy Some sites report slower acquisition Softcopy adjustments (window / level / mag) slower 2001-2003, ~50K women @ 33 sites Five systems NEJM 9/2005 Adoption rate: pre DMIST: <10% 2 yrs later (late 2007): 30% 4 yrs later (7/2009): 54% Who really benefits? Younger women < 50 years old (34%) Women with denser breasts (47%) Women who are pre or perimenopausal (37%) Later press quote: "About 65% of the population" Follow on paper in Radiology 2008 10 subgroups of women <50 yo, dense, AND pre-/perimenopausal was ONLY subgroup where FFDM>SFM (17%) Borderline significant for 65yo fatty breasts: SFM>FFDM (13%) “Trend” toward FFDM>SFM for <50yo, dense, OR pre-/perimenopausal Pros/cons of FFDM FFDM technologies FFDM image quality Pros/cons of FFDM FFDM image quality Flat panel or Digital Radiography (DR) Computed Radiography (CR) Young, Imaging 2006; Mahesh M Radiographics 2004;24 Cesium iodide (CsI) phosphor scintillator converts x-rays into light, then light is turned into electric signal by amorph silicon (a-Si) photodiode CE Senographe 2000D, Senographe DS, Essential X-rays captured by amorph selenium (a-Se), read out directly, no light spread http://www.analogic.com/products-medical-digital-adiography.htm Silicon detector converts each x-ray photon to electric signal Multi-slit scatter rejection Sectra MicroDose Mahesh M Radiographics 2004;24 http://www.analogic.com/products-medical-digital-digital-x-ray-imaging.htm   Photostimulable phosphor in cassette, replaces film cassette X-ray energy stored, then scanned by laser and released as light Fuji Computed Radiographyfor Mammography (FCRm) KC Young, Imaging 2006 (a) Comparison of threshold contrast for a 0.1 mm detail and mean glandular dose for Siemens Novation and Fuji Profect CR systems against the limiting values in European guidance. (Contrast was calculated for a nominal beam quality at 28 kV Mo/Mo.) (b) Mean glandular dose required to achieve the minimum and achievable image quality standard for a 0.25 mm detail size.  The Big Three Same: Detector Area (approx. 24 x 29 cm) Bit Depth (14 bits or 0-16,383) Different:     GE SonoGraphe Essential Hologic Selenia Siemens Inspiration Target/Filter Rh/Rh W/Rh W/Rh Detector Indirect Csl Direct a-Se Direct a-Se Pixel Pitch (µm) 100 70 70 FFDM technologies FFDM image quality Pros/cons of FFDM FFDM technologies Pros/cons of FFDM Image composed of different spatial frequencies Low frequencies for large, blurry, soft objects (mass body, fibroglandular tissue) High frequencies for small, focused, sharp objects (mass margin/spiculation, vessels) Plot how much info is preserved (y-axis) at each spatial frequency (x-axis) Contrast vs. Resolution Contrast from 100% to 2% of sine waves Pattern becomes harder to see MTF of multi-component system MTF: SFM, indirect DR, and direct DR   Measure of resolution or sharpness   Not as simple as pixel size! Detectability depends on both contrast resolution and noise Signal to noise ratio (SNR) What we want (signal) relative to what we don't like (noise) DQE is the ratio of the squared SNR of output (image detected) vs. input (x-ray info)               DQE =  SNR2out              SNR2in Why talk about DQE? Best measure of imaging detector performance/image quality DQE can tell you: Efficiency of a detector Tradeoff between signal and noise Detectable minimum signal difference Detectable minimum contrast Maximum number of resolvable grey levels DQE: SRM, indirect DR, and direct DR   DQE: indirect vs. direct DR Direct > indirect   NW Marshall, Detective quantum efficiency measured as a function of energy for two full-field digital mammography systems, Phys. Med. Biol. 54 (2009) 2845–2861 Why shoot digital differently? Independent optimization of technique No longer limited to Mo spectrum Higher quality spectrum can mean: Better image quality Reduced dose Harder beam quality SFM required lower energy for better absorption in screen Digital detectors have better DQE even at higher energies Sharpness Thicker phosphors would reduce dose, but mammo screens must stay thin to be sharp FFDM can have thick detectors without light spread problem  5cm 50/50 breast W/Rh 0.9mGy, Mo/Mo 1.7mGy, 47% dose saved Tungsten has lower contrast but image processing can equalize the contrast  Using W/Rh instead of Mo/Mo At same image quality, dose savings 9% - 63% Maintain quality, reduce dose 6 cm fatty breast Dose savings: 40% 2cm dense breast W/Rh offers least dose savings (9%) 3 out of 3 radiologists preferred Mo/Mo