Congress:
EuroSafe Imaging 2017
Keywords:
Radiation safety, Physics, Diagnostic procedure, Mammography, Breast, Action 3 - Image quality assessment based on clinical indications, Action 2 - Clinical diagnostic reference levels (DRLs), Action 4 - Quality of radiological equipment, Action 3 - Optimisation, diagnostic reference levels, image quality, Image verification, Quality assurance
Authors:
I. Fitton, A. Noël
DOI:
10.1594/esi2017/ESI-0019
Description of activity and work performed
Materials and methods
The DBT system and images acquisition:
Hologic Selenia Dimensions®.
FFDM and DBT acquisitions: COMBO-HD mode.
The COMBO-HD mode acquires in the same compression both DBT and FFDM images and generates the synthesized C-View 2D image.
Cranio-Caudal incidence.
Clinically used AEC mode.
Breast phantom:
Tomomam®3D (ORION,
France,
serial number 1257-00-00A E2591A) (7).
Image parameters:
Four parameters were tracked from February until June 2016 at the European Georges Pompidou Hospital (Paris,
France) to compare image quality between DBT and C-View 2D images versus 2D FFDM.
-
Image scoring: we scored detectability of micro-calcifications,
spheroidal masses and fibers separately based on partial or total visibility of micro-calcifications,
masses and fibers in clinical viewing conditions.
A score equal to 1 or 0.5 was attributed for each insert according to its total or partial visibility respectively.
The maximum scores for micro-calcifications,
spheroidal masses and fibers were 6.
All readings and image analysis involved the same experienced physicist in image processing.
- Signal Difference-to-Noise Ratio (SDNR): it was evaluated from four contrast spheres according to the glandular/adipose ratios: 100% vs 80% glandular,
and 80% vs 100% adipose.
MVV= Average voxel value,
SD= Standard Deviation.
-
Geometric distortion: the distortion of a 3D matrix consisting in 1mm diameter Aluminium balls was calculated according to the deviation of superimposed balls.
Al balls are located on three different levels inside the Tomomam® phantom (z=+10mm,
z=+30mm,
z=+50mm) Fig. 1
-
Homogeneïty in the plane: Four regions of interest were defined at +10 mm z-level,
which was the level inside the phantom free of inserts and artefacts: one reference ROI (in the center of the plane) and three others ROIs located close to the nipple (ROI 2) and on the laterals sides (ROI 1 and ROI 3) of the plane respectively.
The average pixel value in each ROI was calculated and divided by the average pixel value of the reference ROI.
Results
- Comparison of image scoring between the three types of images: FFDM,
DBT and C-View.
Mean and 95% confidence interval of image scoring is indicated.
Type of mammography images |
Masses |
Microcalcifications |
Fibers |
FFDM |
4.90±0.22 |
3.40±0.55 |
4.00±0.35 |
DBT |
5.10±0.22 |
3.80±0.91 |
4.90±0.65 |
C-View |
3.60±0.89 |
2.80±0.27 |
2.90±0.42 |
- Comparison of SDNR between FFDM,
DBT and C-View 2D images according to adipose and glandular tissues.
- Geometric distortion.
Mean and 95% confidence interval of image distortion is indicated.
Al balls deviation (in %) |
FFDM |
DBT |
C-View |
Al ball 1 x-axis |
3.7±0.7 |
3.5±0.7 |
3.6±0.7 |
Al ball 1 y-axis |
1.1±0.1 |
1.2±0.2 |
1.3±0.3 |
Al ball 2 x-axis |
0.5±0.1 |
0.5±0.1 |
0.6±0.2 |
Al ball 2 y-axis |
1.4±0.2 |
1.4±0.2 |
1.4±0.3 |
Al ball 3 x-axis |
3.6±0.7 |
3.6±0.8 |
3.6±0.8 |
Al ball 3 y-axis |
2.1±0.4 |
2.0±0.2 |
2.1±0.3 |
Al ball 4 x-axis |
0.6±0.1 |
0.6±0.1 |
0.7±0.1 |
Al ball 4 y-axis |
1.8±0.3 |
1.6±0.2 |
1.8±0.3 |
- Homogeneity in the plane.
Mean and 95% confidence interval of image distortion is indicated.
|
FFDM |
DBT |
C-View |
ROI 1 vs ROIref |
1.02±0.01 |
1.05±0.01 |
1.05±0.01 |
ROI 2 vs ROIref |
1.02±0.01 |
1.02±0.01 |
1.01±0.01 |
ROI 3 vs ROIref |
1.03±0.01 |
1.05±0.01 |
1.04±0.01 |