Information Technology Reference
In-Depth Information
which the tumour must receive. To deliver it, the medical physicists (the technicians
who calculate the final treatment) follow a strict planning protocol which includes
a simulation to determine the final doses in order to ensure its quality and effec-
tiveness, using the in-house installed TPS. The plan must be available quickly to
allow treatment to commence as soon as possible and frequently requires a second
calculation for quality control or, in complex cases, the experimental verification,
which is costly in time and money. Reducing the full time required to perform, and
to check the calculations or improving its accuracy will improve quality, efficiency
and satisfaction in the hospital procedure.
The Business Experiment BEinEIMRT includes a new set of remote tools to
help the medical physicists to define these plans: an optimizer, and a virtual veri-
fier. These tools were developed in the framework of a previous national Spanish
research project named e-IMRT (http://eimrt.cesga.es), funded by the regional
government of Galicia (Xunta de Galicia). The optimization tool provides them
with a set of suitable plans which fulfil the prescriptions. The plans can be compared
between them and analyzed by the technicians. In the case that one of the plans is
considered valid, the medical physicist can download the plan in DICOM-RTPLAN
format to be recalculated with their internal TPS. The advantage of this tool, addi-
tional to the usage of updated and accurate algorithms and its extensibility to several
optimization models, is that it examines several treatment modalities simultane-
ously. It produces results for CRT, IMRT and few-levels radiotherapy techniques.
So, the technician can compare among them and selects the most effective or that
which may not be the most effective but the less invasive to deliver (and in some
cases, cheaper than another plan).
The verification tool allows the medical physicists to virtually check the treat-
ment. Usually, the internal protocols of the hospitals include a cross-checking of
the treatment plan with a simpler dose calculation method. This cross-check tries
to avoid errors in the treatment planning which can be dangerous for the patient.
Frequently, it could be done experimentally. To do it, a phantom which emulates
the patient is instrumented to record the doses in certain control points and the
planned treatment is fully delivered. The recorded doses are analyzed and compared
with those calculated by the TPS. Only when both of them are in agreement, the
treatment plan is considered valid. However, this experimental verification is costly
in time and money, and, what is worse, the Linac cannot be used to deliver treat-
ments to the real patients during the data acquisition time. So, the medical physicists
have demanded new software tools for accurate verification of the plans, and Monte
Carlo simulation methods are considered the best solution. But, additionally to the
complex technical details of such simulations, these methods need a large amount of
CPU cycles which make them unpractical, and almost impossible with the current
computing infrastructure of the hospitals.
The e-IMRT (Mouriño Gallego et al. 2007) solution has been designed to provide
these services following a Software-as-a-Service (SaaS) paradigm. The application
provider must first solve the problem of the computing resources that they require. It
can use its own local resources, but this solution limits the scalability of the service
and increases enormously the initial investments. To solve those constraints, the
