Low contrast detail detectability measurements on multi-slice CT scanners Nicholas Keat, Sue Edyvean ImPACT Group London, UK www.impactscan.org RSNA 2003
Clinical importance of low contrast detectability • Studies where soft tissue differentiation is important are common in CT •Abdomen, Pelvis 26 % •Cerebrum 22 % •Spine 20 % Contrast resolution more •Mediastinum 7 % important in ~90 % •Lung parenchyma 6 % •Lung parenchyma 6 % •Trauma 5 % •Interventions 4 % Spatial resolution more •Base of skull 3 % important in ~10 % •Pediatrics 3 % •Orthopedics 3 % •Orthopedics 3 % •Inner ear 1 % •Inner ear 1 % Typical case breakdown for a UK general hospital RSNA 2003
Assessment of LCD • Usually use uniform phantoms with variable size low contrast inserts • Catphan was used in this study • All vendors quote scanner performance on this phantom 0.3% (3 HU) contrast 0.5% (5 HU) 1.0% (10 HU ) contrast contrast Details 2-15 mm diameter Catphan 500 RSNA 2003
Scanners’ stated performance GE Philips Siemens Toshiba Scanner LightSpeed + Mx8000 Volume Zoom Aquilion Multi Phantom Catphan Catphan Catphan Catphan Contrast 0.3% 0.3% 0.3% 0.3% Slice width 2 x 10 mm 10 mm 1 x 10 mm 10 mm 120 kV, 150 mAs* Surface Dose 18 mGy 27 mGy 21 mGy Detail Size 5 mm 4 mm 5 mm 4 mm Detail visibility ? ? ? ? criteria *ImPACT estimated CTDI: 24 mGy • Data not directly comparable Source: ImPACT Four Slice CT Scanner Comparison Report, V5 RSNA 2003
Standard LCD assessment conditions • In order to provide more comparable results, standard exposure and reconstruction parameters were used – 120kV, 10 mm image*, 20 mm collimation*, 25 mGy surface dose, 20 images – Standard kernel, 25 cm FOV, no bone correction where possible • Images scored by four observers under standard conditions with written visibility criteria – All images viewed in a single session in random order – 0.3 % contrast (3 HU) details scored * Closest available setting used, corrections made where necessary RSNA 2003
Image scoring • Images scored for smallest visible detail using custom written IDL program RSNA 2003
Result presentation • Percentage of images at each detail size that is visible is plotted (20 images) – e.g. 15 mm detail visible in 18 images: 90 % visibility – 7 mm detail visible in 10 images: 50 % visibility 100% 100% 90% 90% 80% 80% 70% 70% Visibility (%) 60% 60% Visibility (%) 50% 50% 40% 40% Better LCD up and 30% 30% towards left 20% 20% 10% 10% 0% 0% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Detail diameter (mm) Detail diameter (mm) RSNA 2003
Results: Inter-viewer variability • Four viewers for single group of 20 images – e.g. for > 50% visibility, results vary between 5 and 7 mm 100% 100% 100% 90% 90% 90% 80% 80% 80% 70% 70% 70% Visibility (%) Visibility (%) Visibility (%) 60% 60% 60% 50% 50% 50% 40% 40% 40% 30% 30% 30% 20% 20% 20% 10% 10% 10% 0% 0% 0% 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13 14 14 15 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Detail Diameter (mm) Detail Diameter (mm) Detail Diameter (mm) RSNA 2003
Results for 16 slice scanners 100% 100% 90% 90% 80% 80% 70% 70% Visibility (%) Visibility (%) 60% 60% 50% 50% GE LightSpeed16 GE LightSpeed16 40% 40% Philips Mx8000IDT Philips Mx8000IDT 30% 30% Siemens Sensation 16 Siemens Sensation 16 20% 20% Toshiba Aquilion 16 Toshiba Aquilion 16 10% 10% 0% 0% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Detail Diameter (mm) Detail Diameter (mm) Bars show range of results from four assessors RSNA 2003
Results for 4 slice scanners 100% 100% 90% 90% 80% 80% 70% 70% Visibility (%) Visibility (%) 60% 60% 50% 50% GE LightSpeed Plus GE LightSpeed Plus 40% 40% Philips Mx8000 Philips Mx8000 30% 30% Siemens Sensation 4 Siemens Sensation 4 20% 20% Toshiba Aquilion Multi Toshiba Aquilion Multi 10% 10% 0% 0% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Detail Diameter (mm) Detail Diameter (mm) Bars show range of results from four assessors RSNA 2003
Result variability: 4 slice scanners • Four viewers assessing 80 images (20 from 4 scanners) – Complete agreement of all four viewers for 6 images (7.5%) – Standard deviation from mean score for each image was 1.1 details RSNA 2003
Results: Intra-viewer variability • Single viewer, assessing same group of 20 images on 5 occasions (> 1 month apart) – For > 50% visibility, results vary between 5 and 8 mm 100% 90% 80% 70% Visibility (%) 60% 50% 1st Reading 2nd Reading 40% 3rd Reading 30% 4th Reading 20% 5th reading 10% 0% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Detail Diameter (mm) RSNA 2003
Result variability: single set of images • Single viewer assessing 20 images viewed 5 times – Complete agreement for 0 images – Standard deviation from mean score for each image was 1.4 details RSNA 2003
LCD and dose • Single viewer, looking at images acquired at different dose (mAs) levels at the phantom surface – Expected improvement in visibility is seen at higher dose 100% 10 mGy 90% 15 mGy 80% 20 mGy 70% 25 mGy Visibility (%) 60% 30 mGy 35 mGy 50% 40% 30% 20% 10% 0% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Detail Diameter (mm) RSNA 2003
Conclusions • Definitive assessment of LCD made difficult by inherent subjectivity and viewer variability • Comparisons of results from separate image viewing sessions will lead to inconsistency • Within a single viewing session, results can be compared – Surface dose differences of 5 mGy were differentiated • Differences in Catphan LCD performance of 4 and 16 slice scanners under these conditions are small, and the range of results for scanners overlap • There is a difference between the clinical tasks of diagnosis in CT and the assessment of circular, well defined objects with a priori knowledge of their position and size Slides available at www.impactscan.org RSNA 2003
Recommend
More recommend
