<<
>>

Developments in FLUKA

As seen in Chap.2, FLUKA is a general purpose tool for calculations of particle transport and interactions with matter. Despite being born in the field of accelerator shielding, over the various new versions it gained a fundamental role in several applications, including medical physics.

© Springer International Publishing Switzerland 2016

F. Collamati, An Intraoperative Beta-Probe for Cancer Surgery, Springer Theses, DOI 10.1007/978-3-319-33699-2_5

5.1.1 PET Import

As described in Sect.2.2.1, FLUKA is able to import CT scans directly converting the DICOM files in geometrical bodies to be used in the simulation. This is a key feature in medical applications, and FLUKA is indeed widely diffused within dosimetry studies and treatment planning for radiotherapy and hadrontherapy.

However, by default FLUKA does not support PET scans import. Despite being the files used of the same format as CT (DICOM .dcm files, one for each slice of the patient) there is a profound difference between the information stored therein in the two cases. In fact, while in CT scans for each pixel the Hounsfield value is given (see Sect.2.2.1), in PET images the grey-scale value is related (via two conversion factors) to the radioactivity of the voxel volume, and is measured in Bq/mL.

For this reason, being an “activity map” rather than a physical one, there is no usefulness in importing it as geometry voxels.

However, PET provides information that can be considerably important for some applications.

For example, in the case of Radio Guided Surgery, being able to simulate the exact distribution of activity within the patient allows to test a probe prototype in a completely realistic environment, avoiding to take any hypothesis of uniformity of the concentration of radipharmaceuticals in a certain zone of interest.

For these reasons, in my Thesis work I developed a work-around to import PET scans in FLUKA making them available as a proper “activity map”.

This is achieved in two consecutive steps.

First, the DICOM importer is used to generate starting from the .dcm files a USRBIN (.bnn). This is one of the available scoring detectors (see Sect.2.4), and is basically a sort of three-dimensional histogram, in which at each spatial bin is assigned a value. This value is usually the energy deposited, a particle fluence, the equivalent dose and so on. In such a way, the PET scan is converted into a mesh in a standard FLUKA format.

The second step makes use of a user routine (see Sect.2.3), which is a Fortran pro­gram that reads this USRBIN file interpreting it as a “probability map” according to which primary particles’ positions are generated. The other particles’ characteristics (kind, energy...) are then provided via the BEAM card. It is for example possible in this way, by selecting the apposite radioactive nuclide as particle, to obtain within the simulation the same exact situation of a patient injected with a radiopharmaceutical.

The result of this import process is showed in Fig.5.1.

Despite appearing a quite straightforward approach, there are some caveat to take into account.

First, the isotopes used for PET exams are different from the ones of interest in case of RGS (see Sect.1.3.2). This implies that all this process relies on the assumption (normally accepted by nuclear physicians [1]) that the uptake of a certain radiopharmaceutical depends only on the carrier molecule, and not on the radioactive isotope.

It is then reasonable to use PET images acquired with 68Ga-DOTATOC to know the distribution of an eventual injection of 90Y-DOTATOC. Moreover, the above

Fig. 5.1 Result of the importation of PET (18F-FDG) dicom files in FLUKA, superimposed with the aforesaid CT scan. To highlight internal structures the patient’s volume has been clipped by a dummy volume. Here are clearly showed both healthy organs and tumor uptake, in this case meningioma visible on the head

described method used in FLUKA, using an “activity map”, makes it easy to change the emitted particle.

Secondly, in the PET DICOM files the activity is assumed to be uniform within the voxel, due to the scanning apparatus sensitivity. This means that the effective resolution is limited to the voxel one, usually of the order of 3 x 3 x 3 mm3. Hence, when superimposing PET and CT in FLUKA this difference in spatial resolution must be considered.

Lastly, as mentioned above, PET information is given in Bq/mL, that means that the total activity is divided by the considered volume. For this reason, to obtain the total number of decays in a second in a certain voxel it is necessary to multiply the value given from the PET (and hence imported into the .bnn file) for the voxel volume. On the contrary, the routine that generates particles from the activity maps reads the voxel value as a probability with no volume considered: it is thus necessary a downstream normalization.

For this purpose, I used the routine to calculate the sum of each voxel specific activity (usually of the order of 109Bq/mL). By multiplying this number for the voxel volume (~0.024 mL) the total activity is obtained. This value (~ 10 MBq) represents the number of radioactive decays occurring in each second in the whole patient body.

Therefore, using it as number of primary particles to be generated makes the simulation compatible with an acquisition time of 1 s.

5.1.1.1 Dosimetry Simulation

Among the possible applications of importing within FLUKA PET scan images, one of the most important is the evaluation of the dose given to the medical personnel during a RGS procedure.

In fact, despite the existence of wide literature about the exposition of medical staff to radioactive sources during several kind of procedures common today in nuclear medicine, there is a quiet wide variation among the estimated values. Moreover, this literature is strongly focused on low energy photons applications, such as 99Tc, that, as widely discussed in this Thesis, present a deeply different behavior in term of penetration and dose release in human tissue.

Actually, there is practically no literature about the exposition of medical person­nel after the patient injection of pure â- emitters, being this kind of radiopharma­ceutical used today only for radio metabolic treatments, in which the physician does not need to spend a significant time near the patient.

For these reasons, a contribution from the MC is valuable in order to give an upper limit to the radio exposition of the staff involved in the RGS technique here proposed, in particular with respect to standard 7-RGS.

To this aim, both CT and PET scans from a patient affected by hepatic Neuro Endocrine Tumor were imported into FLUKA by means of the procedure described in Sect.5.1.1. To overcome computational problems in FLUKA, correlated with the huge number of different voxels of a whole body CT scan, the simulation was split in three, each one simulating a part of the patient body. PET and CT scans were precisely superimposed in order to avoid primary particles coming from outside the patient.

Beside the patient body, two phantoms were created to simulate the surgeon’s torso and hand. In particular, the torso was modeled with a water cylinder of 30 cm diameter and 60 cm height placed at about 50 cm from the patient. The hand was modeled with a water parallelepiped (10 x 10 x 2 cm3) placed about 5 cm above the patient body. To simulate the presence of the glove, which plays a fundamental role in case of â- emission, a layer of 1 mm of plastic was placed outside the hand phantom. The setup of the simulation is visible in Fig.5.2.

The PET data were used as an activity map for primary particles, normalizing the number of generated particles as described in Sect.5.1.1 to simulate a 1 sexposition. Both 90Y and 99Tc were simulated. In order to compare these two approaches. For each configuration, the Equivalent Dose (DOSE-EQ) was scored in the torso and hand regions. The results are show in Table5.1.

The equivalent dose given to both surgeon’s torso and hand are found to be signif­icantly lower in the case of â--RGS than in 7-RGS, because of the greatly reduced range of electrons with respect to photons. Such a low value of radio exposition would result in practically no effect on the surgeon, not even from a legal point of view, avoiding the physician to be classified as radio protected worker. The thus obtained values for 99Tc are in reasonable agreement with the present literature [3].

Fig. 5.2 Setup of the simulation used to evaluate the radio exposition of the surgeon during a â-RGS procedure, as described in Sect.5.1.1.1.Both CT and PET scans of a patient are imported; the white cylinder represents the surgeon’s torso, while the parallelepiped represent his hand

Table 5.1 Equivalent dose to the surgeon’s hand and torso during a â--RGS procedure with respect to a 7-RGS [2]

99Tc 90y
Hand 13.8 ^Sv/'h 0.35 ^Sv/'h
Torso 1.39 ^Sv/h 0.04 ^Sv/'h

In the end, it has to be remarked that such an evaluation of the radio exposition provides an order of magnitude rather than a precise value, due to the possible variability among patient and imaging data, the distance of the surgeon and so on.

However, this values are good enough to demonstrate one of the key feature of the RGS technique proposed in this Thesis, i.e. the great reduction in the dose given to the medical personnel, that would result in a much simple procedure to be performed even from a technical and legal point of view.

5.1.2 Scoring Routines

As mentioned in Sect.2.4, FLUKA offers several estimators to evaluate certain quan­tities as deposited energy, energy spectrum, particle fluence and so on. However, for more advanced requirements and analysis custom user routines are more advisable.

For this reason, in my Thesis work I used a tool to convert FLUKA outputs in Root files. This is a set of Fortran routines that act on a step by step basis during

Fig. 5.3 Typical structure of the file obtained via the scoring routines described in Sect.5.1.2

the simulation, storing the information in Root files according to specifically build classes. This crossing between languages (Fortran77 and C++) is managed by means of intercommunicating functions and classes within the routines.

The thus obtained Root file consist of a TTree that allows many analysis, from energy spectrum in the detector to correlation between initial particle momentum and track length therein, and so on. The typical file structure is showed in Fig.5.3.

This approach was the most used within this Thesis work. However, due to the huge amount of information stored, it implies greater simulation times, and the produced files are often quite large (of the order of several GB). Hence, other simpler scoring techniques were also used in some cases.

5.2

<< | >>
Scientific source Francesco Collamati. An Intraoperative Beta-Probe for Cancer Surgery. Springer Theses Recognizing Outstanding Ph.D. Research. 2016

Other medical related information Developments in FLUKA:

  1. Supervisor’s Foreword
  2. Preface
  3. Contents
  4. Chapter 2 The FLUKA Monte Carlo Code
  5. 3.2 First Probe Prototypes
  6. Probes Optimization and Development
  7. Developments in FLUKA
  8. Chapter 6 Conclusions