8/1/2013
Disclosures (required by UC Davis):
Advanced Imaging for Breast Cancer: Screening, Diagnosis, and Assessing Response to Therapy
Breast CT
• Varian Imaging Systems, Consultant
• CT Imaging, Consultant
• Stanford Research Institute, Consultant
• DxRay, Inc, Consultant
• Cedars Sinai Medical Center, Expert Witness
• Alston & Bird LLC, Expert Witness
• Varian Imaging Systems, Research Funding
• Hologic Corporation, Research Funding
• Fuji Medical Systems, Research Funding
• Stanford Research Institute, Research Funding (R21 subcontract)
John M. Boone, Ph.D., FAAPM, FSBI, FACR
Departments of Radiology and Biomedical Engineering
University of California Davis Medical Center
Breast CT
• Siemens Medical Systems, Research Funding
Dedicated Breast CT
Breast CT Technology
Preliminary Clinical Assessments
Mathematical Observer Metrics
Lesion Insertion / t3D versus 2D
Anatomical noise / M↔T↔bCT
CE‐bCT versus Mammo and Tomo
Summary
1
8/1/2013
2001
2004
2005
5
UC Davis Medical Center
Albion
Bodega 2007
Bodega
Albion 2004
Cambria
Cambria 2012
Doheny 2013
Doheny
8
2
8/1/2013
Spatial Resolution: Modeled & Measured MTF’s Breast CT
gt
( ) f  A a1(1exp(t/T1))a2 (1exp(t/T2))a3(1exp(t/T3))
Breast CT Technology
Preliminary Clinical Assessments
Albion & Bodega
Mathematical Observer Metrics
Lesion Insertion / t3D versus 2D
Cambria (2 x 2) [388 m pixels]
Cambria (1 x 1) [194 m pixels]
Anatomical noise / M↔T↔bCT
Doheny
CE‐bCT versus Mammo and Tomo
3X Spatial Resolution
Summary
9
296
second volunteer imaged: January 2005
11
first breast cancer imaged: January 2005
12
3
8/1/2013
bCT (no injected contrast)
True 3D
Display !
14
13
2008
PRE CONTRAST
POST CONTRAST
bCT (with contrast)
PRE CONTRAST
POST CONTRAST
16
4
8/1/2013
Breast CT
2010
Breast CT Technology
Preliminary Clinical Assessments
HU
AUC = 0.87
Mathematical Observer Metrics
Lesion Insertion / t3D versus 2D
Anatomical noise / M↔T↔bCT
CE‐bCT versus Mammo and Tomo
Summary
∑(fi × Ii) = dꞌ
filter
# observations
pre‐whitened matched filter
“computer observer”
lesion absent
lesion present
dꞌ
inserted lesion
inserted non‐lesion
breast CT image (actual images are used)
19
5
8/1/2013
Breast CT
“mammo”
46%
mass lesions only / results do not reflect microcalcifications
Breast CT
Background Noise
Breast CT Technology
Anatomical Noise
Preliminary Clinical Assessments
Mathematical Observer Metrics
low
Lesion Insertion / t3D versus 2D
Anatomical noise / M↔T↔bCT
CE‐bCT versus Mammo and Tomo
Summary
high
6
8/1/2013
Digital Subtraction Angiography
(Temporal Subtraction)
Mammo
Tomo
Breast CT
Reduces Anatomical Noise
Dual Energy Chest Radiography
(Energy Subtraction)
Reduces Anatomical Noise
23 patients were imaged using all 3 modalities
Sylvia Sorkin
Greenfield Award 2012
2D Fourier analysis
10
20
30
40
50
60
10
20
30
40
50
60
radial averaging

linear regression

27
7
8/1/2013
bCT, Tomo, and Mammo Comparisons
N = 23 pts
1000 ROIs per
breast CT
Adipose Breasts
tomosynthesis
mammo
breast CT
mammography
Dense Breasts
Tomo
~10 mm
axial view (~cc)


slice thickness (mm)
Breast CT images
8
8/1/2013
measured data on the breast CT system
disk diameter (mm)
tomographic angle
Mammo
Breast CT Images
Breast CT
Breast CT Technology
Preliminary Clinical Assessments
Mathematical Observer Metrics
0 mm
Lesion Insertion / t3D versus 2D
Anatomical noise / M↔T↔bCT
55 mm
CE‐bCT versus Mammo and Tomo
Tomosynthesis Images
Summary
9
8/1/2013
Prospective Clinical Trial
105 patients /103 lesions (BIRADS 4 or 5)
imaged on VCO mammo / tomo / CE‐bCT
all biopsied
microcalcifications
masses
malignant
31
27
benign
27
18
total
58
45
2 Radiologists Rated Lesions using a 0 to 10 Conspicuity Score
Shadi Shakeri, M.D.
Embargoed Data until Published
(so not in printed notes)
one‐way ANOVA with correction for multiple comparisons 0 = not seen 10 = excellent
Breast CT
Breast CT (Summary)
Breast CT Technology
Patients find bCT more comfortable
Preliminary Clinical Assessments
Radiation dose is the same as 2V mammography
Mathematical Observer Metrics
Early trials and computer sims show better mass lesion detection performance than mammography
Lesion Insertion / t3D versus 2D
Anatomical noise / M↔T↔bCT
CE‐bCT versus Mammo and Tomo
Summary
Computer simulations show bCT reduces anatomical noise / Mammo or Tomo / reasons understood
CE‐bCT has better sensitivity and specificity than mammo or tomo
10
8/1/2013
Acknowledgements:
Breast Tomography Project
University of California Davis
California BCRP 7EB‐0075
California BCRP 11I‐0114 R01 CA•89260
R01 EB•002138‐10 (BRP)
R01 CA•129561 (RDB)
P30 CA•093373 (CCSG)
Susan G. Komen Foundation
University of Pittsburgh 11