ESTRO 36 Abstract Book

S159

ESTRO 36 2017

_______________________________________________________________________________________________

Symposium: Registration and fusion techniques

SP-0310 Rigid registration techniques for different

imaging modalities

N. Nesvacil

Medical University of Vienna, Department of

Radiotherapy- CCC- Christian Doppler Laboratory for

Medical Radation Research for Radiation Oncology,

Vienna, Austria

Image registration has become an important part of

treatment planning and execution in 3D image guided

external beam radiotherapy (EBRT) and brachytherapy

(BT) during the last decades. In principle, the same

algorithms for rigid or non-rigid image registration can be

applied to either type of radiotherapy. However, for their

application in brachytherapy the presence of the

brachytherapy delivery device, i.e. the applicator, plays

an essential role. This presentation will provide an

overview of rigid registration techniques in

brachytherapy, compared to external beam radiotherapy.

In gynaecological brachytherapy, where the applicator

and CTV might move in relation to the bony anatomy

during the course of a (multi-fractionated) treatment,

applicator-based rigid registration can be used to combine

images for treatment planning acquired with the same or

different modalities at different time points. One of the

most useful applications of this technique is to transfer

target contours defined on a reference image, e.g. an MRI

at the time of the first BT, to subsequent CT image

volumes for planning of further BT fractions, if MRI is not

always available for dose plan adaptation to the anatomy

of the day. Requirements and pitfalls for clinical

applications of this technique will be discussed. In order

to analyse interfraction variations for target and organs at

risk (OARs) based on image volumes acquired at different

time points, rigid image registration can provide a good

estimate of the dosimetric impact of anatomical changes

between applicator, CTV and OARs. Clinical examples will

be discussed for different treatment sites. Limitations of

the technique will be summarized and special focus will

be given to prostate and gynaecological BT treatment

planning.

Multi-modal rigid image registration for brachytherapy is

also used to improve target delineation and dose plan

optimization for a single fraction. MRI acquired before

brachytherapy can be combined with CT images for

treatment planning of prostate cancer, in order to

visualize and delineate intraprostatic lesions as boost

target volumes in HDR or LDR brachytherapy.

In the setting of gynaecological cancer brachytherapy

where only CT is available for visualization of the

applicator and surrounding organs after implantation,

rigid registration of ultrasound images obtained with

applicator in situ could be used in the future for dual

modality dose planning of a single fraction. Challenges and

solutions for the registration of ultrasound and CT images,

such as determining the applicator position in the

ultrasound image volume, will be explored to conclude

this presentation.

SP-0311 Deformable registration for dose summation

K. Tanderup

Aarhus University Hospital, Department of Oncology,

Aarhus C, Denmark

Dose summation across brachytherapy fractions and

accumulation with external beam radiotherapy (EBRT)

dose is essential for assessment of total dose to both

targets and organs at risk in treatments with fractionated

brachytherapy and in combinations with EBRT.

Brachytherapy

dose

distributions

are

highly

heterogeneous, and the EBRT dose distribution may also

include significant dose gradients in the vicinity of the

brachytherapy high dose region, e.g. in case of lymph node

boosts. Organ motion between fractions as well as change

in anatomy between brachytherapy and EBRT constitute

specific challenges for dose summation. Deformable

image registration (DIR) aims to match each tissue voxel

irradiated by each fraction of external-beam radiation

with the corresponding voxel irradiated by each fraction

of brachytherapy. However, DIR is related with specific

uncertainties and does not necessarily provide added

value for dose accumulation. The alternative of

accumulating dose through DIR is to perform direct

addition of DVH parameters as recommended in the ICRU

report 89. Direct addition assumes inherently that hot

spots and cold spots remain in the same spatial region

across suceeding fractions and is therefore also related

with uncertainties or even not appropriate for certain

scenarios. DIR can be carried out with deformation models

based on image intensity, biomechanical models, or

combinations of these. Biomechanical models take into

account organ shapes and potentially biomechanical

properties of organs or organ walls/surfaces, such as

elasticity. Biomechanical models based on contours need

to have these defined in both source and target images,

and the correspondence of contours becomes part of the

objective function which drives the optimisation. DIR

models based entirely on image intensity do not take into

account contours, and the objective function is based on

correspondence in image intensity. The major questions

with regard to DIR and dose accumulation in

brachytherapy are: 1) Is it problematic to accumulate dose

without DIR? I.e. what is the accuracy and limitations of

dose accumulation with DVH addition? 2) Can DIR solve the

problem? I.e. what is the accuracy of DIR-based dose

accumulation? For summation of EBRT and brachytherapy,

direct DVH addition is accurate if the EBRT dose

distribution is homogenous in the region where the BT

boost is going to be delivered. In case of a homogeneous

EBRT dose, the EBRT dose contribution to the primary

target D

90%

and D

98%

as well as D

2cm3

for organs at risk will

be equal to the prescribed EBRT dose. Dose distributions

from four-field -box techniques are normally

homogeneous . Furthermore, it is also possible to control

the homogeneity of IMRT and VMAT in the region the the

BT boost through dose optimisation, e.g. by introducing

help structures in the region of the primary target/GTV

with specific constraints on homogeneity. However, in the

case of lymph node boosts which are in close relation to

the primary target and the BT boosted region, direct

addition of EBRT and BT DVH parameters may not reflect

the true accumulated dose. DIR has not been investigated

for this purpose, but may provide an added value. For

summation of dose from succeeding BT fractions, there

are indications that DVH addition has an accuracy better

than 5% for organs such as bladder and rectum, as hotspots

are quite stable across brachytherapy fractions. For target

structures, there have not been any systematic

evaluations, but often cold spots are located in the same

region of the target, and DVH addition is assumed to work

well. For highly mobile organs such as sigmoid colon or

bowel loops, it is well know that hotspots may end up in

very different parts of the loops and DVH addition is

expected to significantly overestimate the hotspot dose.

DIR algorithms which are based on contours and

biomechanical models have been demonstrated to work

well for bladderand improves dose assessment although

the improvement with DIR as compared to direct DVH

addition is normally less than 5%. However, some DIR

algorithms based on image intensities can be related with

significant uncertainties and may provide dose

assessments which are less accurate than with DVH

addition, and DIR should therefore be used with great

caution. Sigmoid and bowel are highly deformable organs

and represent a significant challenge for DIR. There are