The BT and DM images were acquired using the homogeneous background part of the TOR MAM mammographic phantom (Leeds Test Objects Ltd,
North Yorkshire,
United Kingdom).
Phantom details of 3 mm diameter simulate pathological breast features like masses of different nominal contrast values (4%,
3%,
2%,
1.5%,
1% and 0.5%),
organized in six (6) groups of three (3) different mass detail contrasts,
with each contrast appearing up to three (3) times.
MC clusters are simulated by groups of particles of varying number,
size (upper and lower particle size limit of each cluster: 354-224 μm,
283-180 μm,
226-150 μm,
177-106 μm,
141-90 μm and 106-93 μm),
shape and spatial distribution.
The system used to acquire DM and BT acquisitions of TOR MAM phantom is a commercially available mammography system (Hologic,
Selenia Dimensions,
Bedford,
USA),
located at University Hospital of Patras.
The acquisition exposure conditions (tube voltage and load) were selected automatically by the system.
The analysis focuses on the five (5) highest nominal contrast values of the masses,
as well as on the four (4) largest microcalcification cluster sizes,
discernible at clinical conditions (Table 1).
The SDNR index was calculated for these lesions on both DM and focal plane BT reconstructed images.
The phantom thickness considered is the standard breast thickness (45mm PMMA) simulating a medium compressed breast thickness.
Table 1: Nominal contrast values of masses (M) and sizes of microcalcification (MC) clusters in TOR MAM mammographic phantom analyzed in this study.
Two (2) different approaches for mass SDNR calculation and two (2) for MC cluster SDNR calculation were implemented,
that assume different definitions of target (Full-Width-at-Half-Maximum or whole lesion) and/or background regions (distant or surrounding).
Mass SDNR:
- First mass SDNR definition: One (1) circular ROI was utilized manually to define the central part of the lesion target signal (Full-Width-at-Half-Maximum,
FWHM) and three (3) circular ROIs of the same size for defining the average background signal,
located in the mass group background region (Fig.
2).
Mass SDNR was calculated per nominal contrast (M1-M5) by the average value of SDNR of homologous nominal contrast details:
Fig. 1: The equation used for calculating the average SDNR value for the first mass SDNR definition.
Fig. 2: Example of first Mass SDNR ROI definition, with target and background ROIs for M1 (corresponding to nominal contrast 4%) at each of the 3 positions located, in DM mode.
- Second mass SDNR definition: One (1) circular ROI covering the entire mass region was utilized to define the lesion target signal and one (1) ROI of annular shape was used for defining the background signal (Fig.
4).
Mass SDNR was calculated per nominal contrast (M1-M5) by the average value of SDNR of homologous nominal contrast details:
Fig. 3: The equation used for calculating the average SDNR value for the second mass SDNR definition.
Fig. 4: Example of second Mass SDNR ROI definition, with target and background ROIs for M1 (corresponding to nominal contrast 4%) at each of the 3 positions located, in DM mode.
MC cluster SDNR:
- First MC cluster SDNR definition: Ten (10) most visible individual MC particles were manually outlined to define the region of average target signal of the MC cluster and four (4) circular ROIs neighboring to each MC cluster were used to estimate the average background signal (Fig.
6).
The average value of MC cluster SDNR was calculated per cluster size (MC1-MC4):
Fig. 5: The equation used for calculating the average SDNR value for the first MC cluster SDNR definition.
Fig. 6: Example of the first MC cluster SDNR definition for the 4 microcalcification clusters analyzed, corresponding to upper and lower particle size limit of each cluster: MC1: 354-224 μm, MC2: 283-180 μm, MC3: 226-150 μm, MC4: 177-106 μm and target (segmented) and circular background ROI arrangement in DM mode.
- Second MC cluster SDNR definition: SDNR was calculated for each of the ten (10) most visible MC particles,
with the target signal estimated by manually outlined target regions and background signal was estimated from individual particle surrounding regions,
by means of morphological dilation (Fig.
8).
The average value of MC cluster SDNR was calculated per cluster size (MC1-MC4):
Fig. 7: The equation used for calculating the average SDNR value for the second MC cluster SDNR definition.
Fig. 8: Example of the second MC cluster SDNR definition: target (segmented MC) and background (surrounding) ROIs for MC1 (upper and lower particle size limit: MC1: 354-224 μm) in DM mode.