Physica Medica: European Journal of Medical Physics - Volume 32 - Supplement 1 - February 2016

70 / 146

Materials and Methods: This preliminary study was conducted in collab-
oration with four Italian Institutes (Rome, Terni, Rieti and Viterbo). 20
patients with diagnosed prostate cancer were enrolled and 3 OARs (bladder,
rectum and femoral heads) were manually contoured according to pub-
lished guidelines. CT images from the same sample were selected as input
for Prowess 5.2 TPS and OARs contours were obtained by means of the
autosegmentation tool. DICOM RT-structures and CT data were exported
and analyzed using VODCA 5.4 software. A comparison between bladder,
rectum and femoral heads structures manually and automatic contoured
was performed concerning DICE similarity coefficient index.
Results: DICE computation on bladder structure provided better results,
with mean and median DICE coefficients of 0.88 and 0.90, respectively. Mean
and median DICE values of 0.35 and 0.42 were obtained for rectum, while
the agreement between manually and autosegmented images for femoral
heads results in a mean and median DICE values of 0.54 and 0.56, respec-
tively. Autocontouring speed is approximately 10 seconds for each patient.
Conclusions: Autocontouring procedures offer considerable time saving in
the segmentation process with acceptable similarity of the OARs con-
tours, especially in bladder case. Additional investigations are needed when
considering other critical structures such as rectum and femoral heads.
As a general rule, an independent check procedure for optimization of the
autocontouring process is still required to increase contour adherence to
referential benchmark gold standard among experts.
A.219
INTRA- OR PERIPROSTATIC CALCIFICATIONS AS NATURAL FIDUCIAL
MARKERS FOR PROSTATE LOCALIZATION IN IMAGE-GUIDED
RADIOTHERAPY
P. Stucchi L. Pozz C. Bianchi C. Mordacchin D.P. Doino S. Gottard
P. Antognoni R. Novario a Ospedale di Circolo Fondazione Macchi,
Department of Medical Physics, Varese, Italy; b Ospedale di Circolo Fondazione
Macchi, Department of Radiotherapy, Varese, Italy; c Ospedale di Circolo
Fondazione Macchi, Department of Biotechnology and Life Science, Varese, Italy
Purpose: To review CBCT images of 35 patients (pts) receiving radical ir-
radiation for prostate cancer, according to an IGRT protocol based on the
use of intraprostatic calcifications (IPC) as natural fiducial markers for the
evaluation of inter-fraction organ motion. The uncertainties involved and
the CTV to PTV margin were evaluated.
Material and Methods: 646 CBCT images of 35 pts were acquired to eval-
uate the role of IPC as natural fiducial markers. All the pts presented at least
3 calcifications, inside or at the borders of the CTV, clearly defined and
>2 mm in each direction. To investigate the internal stability of the IPC and
the feasibility of using them as surrogates for the prostate position, the in-
ternal distances between IPC and the distances between IPC and the center
of mass of CTV were measured for each patient on both CT-sim scans and
each CBCT. Mean and st.dev of the differences are given as an indication
of IPC stability.
A statistical study was performed on 440 CBCT images to analyze system-
atic and random uncertainties to obtain the CTV to PTV margins.
The total positioning error (T) was calculated by registration of the CBCT
with the CT planning scan and matching CTV, comparable with match IPC.
Automatic image registration was performed according to the bony pelvic
anatomy (BM). Prostate bed motion was defined as the difference between
T and BM.
Results: The calcifications were found to be stable during the radiother-
apy treatment.
The mean of differences in distances between IPC and center of mass of
CTV was 0.2 ± 1.6 mm, 0.0 ± 1.6 mm and −0.3 ± 1.4 mm in LR, AP, and CC
direction.
The systematic and random error for the total positioning error results in
a PTV margin of 5, 8, 6 mm in LR, AP, and CC direction.
Using an IGRT protocol based on IPC, PTV margins may be reduced to 2,
6.6, 4 mm. We have not considered the intra-fraction movement, yet.
Conclusion: Our results confirm that IPC can be used as reliable natural
fiducials. An IGRT protocol based on the use of IPC can reduce the PTVmargins.
A.220
IN VIVO DOSIMETRY WITH GAFCHROMIC FILMS DURING
INTRAOPERATIVE RADIATION THERAPY (IORT): A GOOD PRACTICE FOR
QUALITY ASSURANCE
P. Tabarelli De Fatis *, M. Liotta, S. Fissi, A. De Simone, L. Regolo, E. Scoccia,
G.B. Ivaldi. Fondazione Salvatore Maugeri, Pavia, Italy
Introduction: The aim was to measure the entrance and exit dose (D1 and
D2) of the surgical bed (SB) and to evaluate the shielding disk (SD) align-
ment during breast IORT as full dose (21 Gy) and boost (12 Gy).
Material and Methods: In vivo dosimetry (DIV) was performed in 63 pa-
tients (pts) with two EBT3TM gafchromic films. The first (D1, 2 × 2 cm2) was
placed at the SB surface, while the second (D2) on the upper side of the
SD (with its same dimension) behind the SB, in order to measure the dose
and to check the SD alignment. The SD was fastened between the gland
and the pectoralis muscle for all the pts (pts 1 + pts 2) and for the last 31
(pts 2) the position of the SD was also externally verified through two sur-
gical threads. The EBT3 were calibrated with 6 MeV electrons by the mobile
accelerator LIACTM (SIT, Aprilia, Italy), and they were read after at least 24
hours with a flatbed scanner (EPSONTM 1000XL) both for calibration and
DIV. For the alignment analysis the irradiation area on the film (up to the
20% prescription dose) was compared with the collimator area at the ir-
radiation depth. An evaluation in the high dose region > 85% prescription
dose was also performed computing the expected area considering only
the 85% of the field dimension at the irradiation depth.
Results: The mean difference between D1 measured and expected was
0.6% ± 5%. For D2, the value was 7.2% ± 7.0%, due also to the backscatter com-
ponent. Both in low and high dose regions, the alignment agreement was
within 90% of the area for 15 of 32 pts 1, between 90% and 66% of the area
for 14, while 3 had a worse result. For pts 2 the agreement was within 90%
of the area for 23 of 31 pts, between 90% and 66% of the area for 7 pts and
1 had a worse result.
Conclusion: DIV has been important in our IORT treatments to exclude ac-
celerator malfunctioning, dose delivery errors and to evaluate the accuracy
of the surgical bed reconstruction. Moreover, film dosimetry has been useful
to verify and to improve disk alignment procedure.
A.221
CHARACTERIZATION OF A BIDIMENSIONAL POLYCRYSTALLINE CVD
DIAMOND DOSIMETER WITH PHOTON BEAMS
C. Talamonti A. Baldi M. Scaringella E. Pace A. Pasquin L. Liv
S. Pallotta M. Zan M. Bruzzi a Istituto Nazionale Fisica Nucleare, Università
degli Studi di Firenze, Firenze, Italy; b Università degli Studi di Firenze, Firenze,
Italy
Introduction: The aim of the study is the characterization of a large area
dosimeter (DIAPIX) used for pretreatment verification of intensity modu-
lated treatment plans.
Materials and methods: The device is made of two detector-grade adja-
cent polycrystalline chemical vapor deposited (pCVD) diamond samples,
covering a 2.5 × 5 cm2 area, organized as a matrix of 12 × 24 pixels (2 mm
pitch).
Elekta Synergy LINAC was used to test DIAPIX with 3DRT, IMRT and VMAT
plans. The correct behavior of the matrix under a flat field was verified and
the flatness correction factor was assessed for each pixel. Short and long
term repeatability, linearity and dose rate (DR) dependence were studied.
Since the area of the prototype is smaller to a typical treatment field, in
patient QA, DIAPIX was first positioned at the isocenter and then shifted
in different positions to cover the whole irradiated zone. Patient QA plans
were computed on the phantom with Monaco and Pinnacle TPS.
Results: The device shows a fast dynamic, a stable response due to a neg-
ligible polarization effect and a high signal-to-noise ratio. The sensitivity
is s = 20 nC/Gy. To study the dose rate dependence the current vs DR was
fitted with the semi-empirical expression: I(DR) = I0 + R at the power of D,
I0 is the dark current, R is the fitting parameter and D accounts for the DR
dependence of diamond. Diapix D was 0.99 ± 0.01. A comparison between
the measured maps and the ones predicted by the TPS was performed. Dose
differences with TPS are within 5%, apart in the penumbra region where
the dose gradient is high and the distance-to-agreement is within 3 mm.
e65 Abstracts/Physica Medica 32 (2016) e1–e70