J E de Groot1, I G M Hopman2, M G J T B van Lier3, W Branderhorst4, C A Grimbergen5, G J den Heeten6. 1. Sigmascreening, Meibergdreef 45, 1105 BA, Amsterdam, The Netherlands. Electronic address: jerry.degroot@sigmascreening.com. 2. University of Amsterdam, The Netherlands. Electronic address: ingrid-hopman@hetnet.nl. 3. Sigmascreening, Meibergdreef 45, 1105 BA, Amsterdam, The Netherlands. Electronic address: monique.vanlier@sigmascreening.com. 4. Sigmascreening, Meibergdreef 45, 1105 BA, Amsterdam, The Netherlands. Electronic address: w.branderhorst@amc.uva.nl. 5. Academic Medical Center, Meibergdreef 9, 1105 AZ, Amsterdam, The Netherlands. Electronic address: c.a.grimbergen@amc.nl. 6. Academic Medical Center, Meibergdreef 9, 1105 AZ, Amsterdam, The Netherlands. Electronic address: denheeten@gmail.com.
Abstract
INTRODUCTION: A recent technological development allows pressure-standardised mammography by personalizing the compression force to the breast size and firmness. The technique has been shown to reduce pain and compression variability between consecutive exams, but also results in a slightly thicker compressed breast during exposure. This raises the question whether visibility, contrast and sharpness of lesions are affected? METHODS: Four experienced radiologists compared 188 stable lesions and structures including (clusters of) calcifications, (oil) cysts and lymph nodes that were visible in mammograms obtained in 2009 with a pain-tolerance limited 18 daN target force compression protocol, and in 2014/2015 obtained with a 10kPa (75mmHg) pressure-standardised compression protocol. Observers were blinded for all DICOM metadata and rated which of the randomly ordered, side by side presented images had better lesion visibility, contrast and sharpness, or whether they saw no difference. They also indicated which overall image they preferred, if any, and whether the non-preferred image was still adequate. Statistical non-inferiority is concluded when the lower limit of the 95% confidence interval of the 4-rater averaged 'new protocol better' proportions exceed the non-inferiority limit of 0.463. RESULTS: In 2014/2015, the compressions were significantly milder, with on average 17% (mediolateral oblique) to 29% (craniocaudal) lower forces. Breasts remained on average 2.4% (1.4mm) thicker. Dose was significantly lower (6.5%), which is explained by glandular atrophy. The 95% confidence interval lower limits are 0.479 for visibility, 0.473 for contrast, 0.488 for sharpness and 0.486 for preference, all exceeding the non-inferiority limit. Of the 60 non-preferred mammograms, multiple observers found only five to be inadequate: 4 obtained with the force protocol and 1 with the pressure protocol. CONCLUSION: Pain-reduced mammography with 10kPa pressure-standardised compression has non-inferior visibility, contrast and sharpness for stable lesions compared to pain-tolerance limited 18daN target force compression.
INTRODUCTION: A recent technological development allows pressure-standardised mammography by personalizing the compression force to the breast size and firmness. The technique has been shown to reduce pain and compression variability between consecutive exams, but also results in a slightly thicker compressed breast during exposure. This raises the question whether visibility, contrast and sharpness of lesions are affected? METHODS: Four experienced radiologists compared 188 stable lesions and structures including (clusters of) calcifications, (oil) cysts and lymph nodes that were visible in mammograms obtained in 2009 with a pain-tolerance limited 18 daN target force compression protocol, and in 2014/2015 obtained with a 10kPa (75mmHg) pressure-standardised compression protocol. Observers were blinded for all DICOM metadata and rated which of the randomly ordered, side by side presented images had better lesion visibility, contrast and sharpness, or whether they saw no difference. They also indicated which overall image they preferred, if any, and whether the non-preferred image was still adequate. Statistical non-inferiority is concluded when the lower limit of the 95% confidence interval of the 4-rater averaged 'new protocol better' proportions exceed the non-inferiority limit of 0.463. RESULTS: In 2014/2015, the compressions were significantly milder, with on average 17% (mediolateral oblique) to 29% (craniocaudal) lower forces. Breasts remained on average 2.4% (1.4mm) thicker. Dose was significantly lower (6.5%), which is explained by glandular atrophy. The 95% confidence interval lower limits are 0.479 for visibility, 0.473 for contrast, 0.488 for sharpness and 0.486 for preference, all exceeding the non-inferiority limit. Of the 60 non-preferred mammograms, multiple observers found only five to be inadequate: 4 obtained with the force protocol and 1 with the pressure protocol. CONCLUSION:Pain-reduced mammography with 10kPa pressure-standardised compression has non-inferior visibility, contrast and sharpness for stable lesions compared to pain-tolerance limited 18daN target force compression.