Instrumentation & Measurement Magazine 24-9 - 67

Fig. 5. RIDOS components. (1) NI-PXI crate with FPGA interfaced with the CNAO dose delivery system and with the ANZAI analog signal; (2) Patient respiratory
phase, as measured with the ANZAI system; and (3) The workstation equipped with a GPU NVIDIA TESLA K40 for the online dose reconstruction and comparison
displayed on the RIDOS GUI.
a system for comparing the delivered and planned dose distributions
during the treatment delivery at CNAO. Therefore, the
system was designed for a synchrotron-based facility, and specifically
to give feedback on the accuracy of the partial doses
delivered during each accelerator spill cycle. This strategy exploits
the beam delivery time structure characterized by 2 to 5 s
of inter-spill time, to perform these comparisons.
The core of RIDOS is the rapid dose reconstruction tool,
based on NVIDIA GPU and running on a dedicated, highperformance
workstation (WS). The WS calculates the dose
delivered to the patient by thousands of spots within the available
inter-spill time of a few seconds. To be tested in a clinical
environment, RIDOS was designed as an extension of the
CNAO DDS, and an NI PXIe crate, equipped with a controller
and a NI FPGA-7831R module, was used to interface the
WS with the FPGAs of the DDS (see above). For the dose reconstruction,
this means that RIDOS receives measured data
on the position, timing, and delivered dose for each beam spot.
To also consider the interplay effects of anatomic motion, an
interface with the commercial Anzai AZ-733 V respiratory gating
system (Anzai Medical, Tokyo, Japan) was implemented.
This allows for synchronizing spot data with the patient motion
phase, as used for the gating delivery technique currently
used at CNAO [35]. The latter allows for selecting the proper
3D-CT image among the sequence of CT images available
through a 4D-CT.
Additionally, a fast image deformation algorithm has been
developed and tested independently [20] to deform the dose of
December 2021
each spill on points of the reference CT used by the TPS. The cumulative
dose distributions can be computed to be compared
with the planned one. The use of synchronized CT images has
yet to be integrated into the RIDOS system and tested.
By accounting for beam deviations, interplay effects and
anatomic deformations, RIDOS will provide the CNAO DDS
with the capability of comparing the planned and delivered
dose distributions for each spill with two GPU-based tools: the
fast dose reconstruction and the fast 3D-gamma index algorithm
[36]. A Matlab-based Graphic User Interface (GUI) was
developed to display a summary of the computed doses and
comparisons in the RIDOS monitor (placed in the CNAO local
control room), as shown in Fig. 5. Controls on the GUI allow
the user to set the computing grid dimensions, the dose-difference
and distance-to-agreement criteria for the gamma-index
computation. Spill-by-spill, the gamma-index pass rate updates
a histogram to provide a trend of the delivered dose
accuracy.
The RIDOS system currently provides no direct feedback
to the DDS, that is, the available data on the actual delivery is
not exploited to modify the ongoing delivery of the remaining
plan. While this remains a highly interesting research project,
the ramifications on quality assurance and patient safety are
complex and currently prohibit a direct adaptation of the plan.
Conclusions
The design of the CNAO DDS began over 15 years ago, and
the developed instrumentation has been used for treating
IEEE Instrumentation & Measurement Magazine
67

Instrumentation & Measurement Magazine 24-9

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 24-9