A GUI of Automatic Image Registration Software
Suitable for Radiotherapy Treatment Planning
D Rajasekar, Niloy R Datta, Rakesh K Gupta* and Sajja B Rao¶
Departments of Radiotherapy, Radiodiagnosis and Imaging *, Sanjay Gandhi Postgraduate Institute of Medical Sciences,
Lucknow, India ¶Department of Mathematics, Indian Institute of Technology, Kanpur, India.
Abstract
Medical imaging forms a vital component of radiotherapy treatment planning and its evaluation. The integration of
the useful data obtained from multiple imaging modalities for radiotherapy planning can be achieved by image
registration softwares. In radiotherapy planning systems, normally the CT slices are kept as a standard upon which
other modality images (viz. MRI, SPECT, PET etc) are aligned - automatically or interactively. Following validation
of successful registration, they are resampled and reformatted, as per the requirements. This paper defines the
minimum requirements of automatic image registration software for 3D radiotherapy planning and describes the
implementation of a suitable graphical user interface developed in Visual Basic (version 5). The automatic image
registration (AIR) routines freely available from Dr. Roger P Woods, UCLA, (USA) were used in this software. This
software could be easily implemented and found to be easy to use for image processing suitable for radiotherapy
planning systems.
Keywords
Automatic image registration, Image fusion, Multimodality imaging, Radiotherapy planning
Introduction
The imaging modalities employed for effective three
dimensional radiation therapy planning (3D RTP) can
be divided into two global categories: anatomical and
functional. Since information gained from any two
type of images are usually of complementary nature,
proper integration of useful data by image registration
technique is a necessary pre-processing step. Image
based registration can be divided into extrinsic, i.e.,
based on foreign objects introduced into the imaged
space and intrinsic methods, i.e., based on the image
information that are generated by the patient itself.
Extrinsic methods rely on artificial objects attached to
the patient, which are designed to be well visible and
accurately detectable in all of the imaging modalities
under study.1-4 As such; this kind of image registration
is comparatively easy, fast and can be automated.
Since the registration parameters can be computed
explicitly, these methods don’t require complex
optimization algorithms. The main drawbacks of
extrinsic registration are that they are prospective in
character, i.e. provisions must be made in the preacquisition phase, and secondly the invasive nature of
the marker objects used.5-7
Intrinsic methods rely on patient generated image
content only. Registration can be based on a limited set
of identified salient points (landmarks based)8, on the
alignment of segmented binary structures - most
commonly object surfaces (segmentation based)7,9-12,
or directly onto measures computed from the image
intensities / gray values (i.e. voxel property based).1, 13
The voxel property method is the most interesting
method currently being explored as they operate
directly on the image intensities, without prior data
reduction by the user or segmentation.
Any comprehensive automatic image registration
(AIR) program for radiotherapy planning should have
the following basic features:
 Convert the 2D images from various imaging
equipments into a 3D volume file, suitable for
registration analysis
 Align any two volume files using specific
registration models with provision for manual
registration
 Sample the aligned volume into the reference
locations
 User-friendly
graphical
qualitative
and
quantitative validation tools
 Convert the aligned images into an format
suitable for radiotherapy planning systems
There are many free and commercial image
registration softwares available in the market.
However they are either too costly or not suitably
designed to be used in radiotherapy treatment planning
systems. This paper describes the implementation of
windows based graphical user interface for the AIR
routines freely available from Dr. Roger P Woods,
UCLA, USA, which has been suitably modified to be
used for automatic registration of images for
radiotherapy treatment planning.
Materials and Methods
The AIR routines freely available from Dr. Roger P
Woods, UCLA, (USA) is a widely used software in
image processing field. These software routines can be
downloaded freely from the website: (http://bishopw.
loni.ucla.edu/AIR5/index.html).
However,
these
software routines are for general-purpose image
registration purpose, not optimally designed for
radiotherapy planning. It is too cumbersome and timeconsuming to use these AIR programs in routine
clinical applications. To facilitate easy image
registration process with the mandatory requirements
stated above, an in-house automatic image registration
software (AIRwin) with user-friendly graphical user
interface (GUI) have been developed using Visual
Basic (version 5) compiler.
The AIR programs can handle only images of a
specific format (i.e. 8 or 16 bit binary matrix of
specified size with the corresponding header
information in a separate file). Since it is time
consuming and error prone to manually convert each
image into format suitable for AIR module or into TPS
compatible formats, a simple software command
should be added in the GUI. The registration process
should be able to align images of 2D or 3D types. A
2D-linked cursor tool which is the simplest of all and
can be used for qualitative and quantitative validation
of registration, has been proposed included the GUI.
The header intensities would adversely affect the
registration accuracy from image obtained from hard
copy films, a simple method should be designed to
properly remove them prior to image registration.
Appropriate graphic controls were proposed for easy
conversion of images suitable for radiotherapy
planning systems.
Results
The ‘AIRwin’ software was developed with userfriendly GUI with modules for 'Automatic Image
Registration' (Fig. 1) and 'Validate Registration' (Fig.
2). Various steps in registration of medical images as
discussed in the material methods section were
grouped in these modules for easy of use and
workflow. All the executable AIR functions were kept
in a separate directory, which could be modified by the
user. The mandatory and optional parameters
necessary for various functions and generic error
messages could be downloaded from the AIR5
website.
Using the ‘Validate Registration’ module, one can
covert the images files from popular formats into AIR
format. Currently the ACER-NEMA (*.ima), DICOM
(*.dcm), windows gray scale bitmap (*.bmp), float
numbers (*.flt) images can be converted into the AIR
format. These image files can be selected from the file
list control (single or multiple files) and the
corresponding (*.img) files can be created by ‘Save
AIR’ button. The header files (*.hdr), corresponding to
these ‘*.img’ files can be created in ‘Automatic Image
Registration’ module using ‘Make Header’ button,
after entering appropriate header data. The ‘Create
Volume’ button combines selected 2D image files into
a single 3D-volume file with the name entered in the
text box. The header information can be verification
through the ‘View header’ command button. The
default global maximum value can be changed using
'Set Max' command.
Fig 1. Automatic Image Registration module of
‘AIRwin’ software.
Automatic Image Registration
The most important function in this software is 'Auto
Register' -- the registration function. Any two image or
volume files can be used for 2D or 3D automatic
registration process respectively. One among them can
be kept as a reference image / volume and the other
image / volume (test / register volume) is transformed
and made to align with the reference image / volume in
this process. The resulting transformation parameters
in the output file depend on the registration model
selected and other optional parameters that can be
entered in the ‘Auto Register’ section. The output from
the 'Auto Register' process, i.e. information regarding
input and output volumes / images and optional
parameters as well as resultant transformation
parameters, etc., are stored in the output file (*.air)
entered in the corresponding text box.
The function 'Manual Register' is the only interactive
program in this module, which requires the operator to
specify values for x_shift, y_shift, and z_shift,
x_rotation, y_rotation, z_rotation, x-axis_scaling, yaxis_scaling and z-axis_scaling. These values can then
be stored in an initialization file for use with the 'Auto
Register' function. Alternatively, these values can be
used to manually generate an 'air file' or to manually
'Reslice' a volume file according to these parameters
without generating an "air file".
Validation of automatic registration
In 'Validate Registration' module a set of 'reference'
and 'registered' 2D slices can be selected from the
respective list boxes and they are displayed in the
picture boxes and are verified graphically using a
linked cursor display (Fig. 2).
Fig 2. Validation module of AIRwin software showing
the reference (pre-RT T2-wt MRI) and registered
(post-RT T2-wt MRI) axial slices with the linked
cursor validation tool.
A simple software tool ‘BMPview’ was designed to
delete the header information from the scanned images
(Fig. 3). Using this software tool a polygon can be
drawn interactively using the mouse covering only the
patient generated intensities and filling the image area
outside the polygon with background intensity leaving
patient generated intensities intact.
accurately aligning the information in the different
images, and providing tools for visualizing the
combined images.
The process of integrating information from different
imaging modalities can be divided into two main tasks.
The first task is to estimate the transformation
parameters (rotation, translation and scaling vectors)
that relate the coordinates of any two imaging studies.
The second task is to apply the resultant
transformation to map structures or features of interest
from one imaging study to another or to reformat the
images from one study to match the orientation and
scale of the images of the other. The set of nine
parameters can be estimated to for the necessary
transformations are: three rotation angles (Ax, Ay,
Az), three translation values (Tx, Ty, Tz) and three
scaling factors (Sx, Sy, Sz). The rotation and
translation parameters account for differences in
orientations and location of the patient with the
different imaging devices.5, 10 The scaling parameters
are included to account for possible mis-calibrations of
the imaging devices. In theory, all machine calibration
parameters should be determined before hand and used
to correct the imaging data before integration or be
made available as known parameters to incorporate
them into the integration process.
The current version of the ‘AIRwin’ software used the
following AIR routines: makeaheader, viewheader,
setmax, alignlinear, scanair, reunite, reslice and
separate. For each process, it created a batch file with
appropriate AIR command with the mandatory factors
and optional parameters as per the selection of the
user. After successful creation, these batch files were
executed. The description of the basic command line
and the various mandatory / optional parameters could
be downloaded from the AIR website (http://bishopw.
loni.ucla.edu/AIR5/index.html).
Discussions
The validation registration of images is absolutely
indispensable in any automatic registration process.
Currently research is going on in area with user
friendly modules to validate the results such as
orthogonal view with curtain / eraser, spy glass tool,
checker board view, 3D linked cursors etc. Validation
of a registration includes more than just the accuracy
verification. The list of items includes must be:
precision, accuracy, robustness / stability, reliability,
resource
requirements,
algorithm
complexity,
assumption verification and clinical use.14
Medical images are increasingly being used within
healthcare for diagnosis, planning treatment, guiding
treatment and monitoring disease progression. In many
of these studies, multiple images are acquired from
subjects at different times, and often with different
imaging modalities. Computerized integration of these
images has great potential benefits, particularly by
Although, a number of similar softwares are routinely
available as a part of the commercial 3D treatment
planning systems, the user may not have much choice
of individualizing the functions of the software to his /
her requirements. Moreover, these are available always
at a price, which could be difficult to procure for a
Fig 3. The ‘BMPview’ software tool used for preprocessing hard copy films scanned from a film
scanner.
routine department especially of a developing country.
The effort made in this study demonstrates that one
can successfully use the benefit of certain freely
available image registration software available on the
Internet to work on various aspects of image
registration to tailor make to meet the requirements of
a department. (This software along with the source
code can be freely supplied upon request to: drsekar@
sgpgi.ac.in).
Conclusions
The major application areas of automatic registration
currently are in radiation therapy and neurosurgery.
The modifications made in the AIR programs as
discussed in this study is currently being used in the
department and has been found to be very much useful
for three-dimensional radiotherapy planning. This
software contains self-explanatory controls that can be
operated by any person with a reasonable image
processing experience.
Acknowledgements:
The Authors would like to thank Dr. Roger P Woods,
UCLA, (USA) and his team for making the AIR
programs freely available though their website and
allowing us to use part of the AIR user manual in this
manuscript.
References
[1] Ardekani BA, Braun M, Hutton E, Mayes A,
Sagar HJ. 1995. A fully automatic multimodality
image registration algorithm. J Comput Assist
Tomogr 19: 615 - 23.
[2] Kooy HM, Herk MV, Barnes PD. 1994. Image
fusion for stereotactic radiotherapy and
radiosurgery treatment planning. Int J Radiat
Oncol Biol Phys 28: 1229 - 34.
[3] Mongioi V, Brusa A, Loi G, Pignoli E, Gramaglia
A, Scorsetti M, Bombardieri E, Marchesini R.
1999. Accuracy evaluation of fusion of CT, MR,
and SPECT images using commercially available
software packages (SRS Plato and IFS). Int J
Radiat Oncol Biol Phys 43: 227 - 34.
[4] Pietrzyk U, Herholz K, Fink G, Jacobs A, Mielke
R, Slansky I, Wuker M, Heis W. 1994. An
interactive technique for three dimensional image
registration: validation for PET, SPECT, MRI and
CT brain studies. J Nucl Med 35: 2011- 18.
[5] Hill DLG, Batchelor PG, Holden M, Hawkes DJ.
2001. Medical Image Registration. Phys Med Biol
46: R1 - R45.
[6] Ten Haken RK, Thornton AF, Sandler HM, La
Vinge ML, Quint DJ, Fraass BA, Kessler ML,
McSchan DL. 1992. A quantitative assessment of
the addition of MRI to CT based 3D treatment
planning of brain tumors. Radiother Oncol 25:
121- 33.
[7] Turking TG, Hoffman JM, Jasczak RJ, MacFall
JR, Harris CC, Kilts CD, Pellizari CA, Coleman
RE. 1995. Accuracy of surface fit registration for
PET and MR brain images using full and
incomplete brain surfaces. J Comput Assist
Tomogr 19: 117 - 124.
[8] Birnbaum BA, Noz ME, Chapnick I. 1991.
Hepatic hemangiomas: diagnosis with fusion of
MR, CT and 99mTc-labeled red blood cell
SPECT images. Radiology 181: 469 - 474.
[9] Herk MV, Kooy HM. 1994. Automatic threedimensional correlation of CT-CT, CT-MRI, CTSPECT using chamfer matching. Med Phys 21:
1163 - 78.
[10] Pelizzari CA, Chen GTY, Spelbring DR,
Weichselbaum RR, Chen CT. 1989. Accurate
three-dimensional registration of CT, PET and /
or MR images of the brain. J Comput Assist
Tomogr 13: 20 - 26.
[11] Kessler ML, Pitluck S, Petti P, Castro JR. 1991.
Integration of multimodality imaging data for
radiotherapy treatment planning. Int J Radiat
Oncol Biol Phys 21: 1653 - 67.
[12] Petti PL, Kessler ML, Fleming T, Pitluck S. 1994.
An automated image-registration technique based
on multiple structure matching. Med Phys 21:
1419 - 26.
[13] Woods RP, Mazziotta JC, Cherry SR. 1993. MRIPET registration with an automated algorithm. J
Comput Assist Tomogr 17: 536 - 46.
[14] Maintz JBA. 1996. Retrospective registration of
tomographic brain images. PhD Thesis, Utrecht
University.
[15] Harrington S. 1987. Computer graphics: A
programming approach. 2 ed. McGraw-Hill Book
Company, New Delhi. pp 70 - 141.