1 Introduction

Although ultra-high dose rate (UHDR) fields were already used in the 1970s for in vitro radio-oncology experiments [1,2,3,4], it was only in 2014 that what is presently known as “FLASH effect” was demonstrated: as compared to a traditional fractionated radiotherapy, an UHDR radiotherapy treatment with the same tumour control probability can produce less severe normal tissue complications [5]. This triggered intense research efforts producing hundreds related papers [6,7,8,9,10,11,12,13], leading to first trials on humans in 2019 [14] and to first follow-up studies, showing no evidence of delayed effects [15]. To date, the FLASH effect was observed not only with multi-MeV electron beams, but also with gammas [16], protons [17] and ions [18].

The radiobiology of the FLASH effect is not fully understood [19, 20], thus more research is needed before the FRT can be made available to humans. Meanwhile, suitable radiation sources and dosimetric tools are gradually becoming available. Novel electron accelerators with UHDR capabilities are already on the market, such as Oriatron eRT6 [21], Mobetron [22] and ElectronFlash [23, 24]. These machines can deliver radiation pulses with duration 1–10 µs, repetition rate up to hundreds Hz and dose per pulse up to tens of Gy. The instantaneous dose rate may reach MGys/s. Performing accurate dosimetry in UHDR is highly challenging, as conventional radiotherapy dosimeters such as ionization chambers, semiconductors and scintillators tend to lose their linearity at these dose rates [23]. Although passive dosimeters such as radiochromic films and alanine were successfully used [25, 26], efforts to develop suitable active dosimeters are highly needed for the rapid, accurate and effective dosimetry of FRT [27]. Results were obtained by adapting conventional dosimeters to UHDR, as illustrated below:

  • In ionisation chambers, the main disrupting effect of UHDR is the charge loss by recombination. This was limited by using a free-in-air parallel plate chamber with 0.25 mm electrode distance [28]. Such a device can linearly operate up to few Gy/pulse with 9 MeV electrons. Another solution was hypothesised by using a noble gas at appropriate pressure [29].

  • Specifically designed, radiation hard, semiconductors were used in UHDR, such as the flashDiamond Detector T60025, available for research purposes from PTW [30]. This was proved to linearly operate up to 20 Gy/pulse with 9 MeV electrons. Silicon Carbides are also under investigation and their operation is currently limited to about 2 Gy/pulse [31].

All mentioned devices operate in time-integrated mode, providing the delivered dose at the end of an exposure without providing the time structure of a single radiation pulse.

A different approach was followed at INFN-LNF with UCD (Unbalanced Core Detector). UCD is an electric sensor whose pins are isolated in absence of radiation. The exposure to intense radiation field causes an electric current that is measured with an especially designed variable-gain front-end electronics. Due to pending initiatives to protect IP, no internal details of UCD are disclosed here. However, the following Sections full describe the results of the characterisation tests. These were performed using ElectronFlash electron machines at SIT Sordina IORT Technologies S.P.A. (Aprilia, Italy), hereafter called SIT. Importantly, UCD can provide time-resolved information on the UHDR pulse, thus allowing to: (1) check the pulse-to-pulse reproducibility in an irradiation session, and (2) detect pulse delivery errors and machine instabilities. These properties are especially important in view of the clinical transition of FRT.

2 Materials and methods

The Unbalanced Core Detector (UCD) (Fig. 1) is formed by:

  • A 6 mm diameter spherical sensitive nucleus (patent in progress). In principle, the nucleus can be manufactured virtually in any geometry and size, the spherical geometry being the most suited for clinical dosimetry because of its intrinsic isotropic response. According to experimental tests performed at SIT, the dosimeter sensitivity roughly linearly increases with the radius of the nucleus. Another aspect is the manufacturing process: to date the UCD is manually cabled, so nucleus sizes in the order of few mm are better handled. On the other hand the nucleus should be as small as practically achievable, in order to minimally perturb the radiation field. 6 mm diameter is the result of an optimisation process involving these three aspects. Due to its size, the UCD is characterised by an inherent "build up" layer estimated in 4 mm of water equivalent. The effective measurement point coincides with the geometric centre of the spherical nucleus.

  • A water-proof 3D-printed cap with 41 mm height, 8.2 mm inner diameter and 0.5 mm thickness.

  • A one-meter long coaxial cable connecting the UCD to the front-end electronics though an SMA straight male connector.

Fig. 1
figure 1

The unbalanced core detector

Full size image

The front-end electronics is based on an ultra-low background, double stage, buffered amplifier with adjustable gain. Importantly, the gain is adjusted from the control PC in order to minimise the number of times the user needs to accede the irradiation room. The analogue board output is sent to a commercial digitizer located in the control room. The cable connecting the board to the digitizer can be as long as about 100 m. From the circuital point of view, UCD can be schematized as a current generator, where the current is proportional to the incident dose rate. The front-end converts this current into a voltage drop. Therefore the time-integrated voltage, usually measured in V × µs, is assumed to be proportional to the dose.

The characterization of UCD was performed using an ElectronFlash LINAC (https://www.soiort.com/flash-rt-technology) with the following operation parameters:

  • Horizontal beam;

  • Electron energy 7 and 9 MeV;

  • Pulse duration 5 µs and repetition rate 5 Hz;

  • An IORT-type PMMA cylindrical applicator with 100 mm diameter was used to homogenise the spatial distribution of the beam;

  • The detector was exposed in a 29.5 cm × 29.5 cm × 40 cm water phantom through a holder which can be remotely moved along the three axes (See Fig. 2). The water phantom has 2 cm thick PMMA walls excepted in the beam entrance area, where the wall is minimised through a carbon fiber. By multiplying this number by the quotient of the electron stopping power in carbon and water, it can be obtained that the corresponding thickness of water is 1.3 mm.

  • Each measurement in the water phantom was obtained by delivering 30 pulses.

Fig. 2
figure 2

UCD installed in the water phantom

Full size image

Figure 3 presents a typical electron pulse as acquired by a 5 MS/s digitiser connected to the output of the analogue board. The oscillations are due to the large intrinsic capacity of the detector. As they have null integral, they do not impact the measurement linearity. Nevertheless, efforts are in progress to compensate them with an adequate filter on the analogue board.

Fig. 3
figure 3

A typical electron pulse as acquired by a 5 MS/s digitiser connected to the output of the analogue board

Full size image

All measurements were simultaneously taken with UCD and a PTW flash Diamond detector T60025 [15].

The following measurements were performed:

  1. 1.

    Percentage Depth Dose (PDD) curves;

  2. 2.

    Lateral beam profiles;

  3. 3.

    Calibration;

  4. 4.

    Linearity;

  5. 5.

    Isotropy;

  6. 6.

    Radiation resistance.

3 Results

Figures 4 and 5 report the percentage depth dose (PDD) curves for 7 and 9 MeV electron beams recorded with both UCD (diameter 6 mm) and the flashDiamond. For both dosimeters the measurement reproducibility was better than 1%, and the main source of uncertainty was the positioning. The positioning uncertainty was transferred to the dose profile by applying Eq. (1), where the positioning uncertainty was assumed to be Δx = 1 mm. Both UCD and flashDiamond profiles are affected by the same uncertainties ranging from less than 1%, below 20 mm, to about 10% in the falling edge after the maximum.

Fig. 4
figure 4

PDD curves with 6-mm spherical UCD and flashDiamond at 7 MeV. The associated bars are only shown on UCD for a better inspection of the graphs

Full size image
Fig. 5
figure 5

PDD curves with 6-mm spherical UCD and flashDiamond at 9 MeVThe associated bars are only shown on UCD for a better inspection of the graphs

Full size image
$$\Delta D=\frac{dD}{dx} \times \Delta x$$
(1)

It should be noted that the smallest measured depth for the flashDiamond is about 4 mm lower than UCD. This is due to the UCD physical size, causing an intrinsic build-up of about 4 mm, as mentioned in Sect. 3. Apart this effect, the PDDs of UCD and flashDiamond are indistinguishable.

The lateral beam profiles obtained with both UCD (6 mm diameter) and the flashDiamond at 7 MeV (depth in water 12.5 mm) and 9 MeV (depth in water 17.5 mm) are reported in Figs. 6 and 7. Again, the uncertainty bars are obtained according to Eq. (1) and are associated to UCD for a better inspection of the graphs. The structure of the profile, clearly reflecting the shape of the 100 mm diameter PMMA applicator, is equally described by both dosimeters.

Fig. 6
figure 6

Lateral beam profiles with UCD (6 mm diameter) and flashDiamond at 7 MeVThe associated bars are only shown on UCD for a better inspection of the graphs. Data are normalised to the value at d = 0

Full size image
Fig. 7
figure 7

Lateral beam profiles with UCD (6 mm diameter) and flashDiamond at 7 MeV. The associated bars are only shown on UCD for a better inspection of the graphs. Data are normalised to the value at d = 0

Full size image

The sensitivity of UCD (6 mm diameter) was estimated by comparison with the flashDiamond at 7 and 9 MeV. The devices were exposed at the same time in five positions within the water phantom. As this procedure can be affected by mutual perturbation effects, it cannot be regarded as a true calibration. However, it has the advantage of using the same beam at the same time. By averaging the FlashDiamond / UCD ratio (q) over the five measurement position, the following values were obtained:

$$q_{{{\text{av}}}} \left( {7{\text{ MeV}}} \right) = \left( {1.17 \pm 0.03} \right)\;{\text{Gy}} \times V^{{ - 1}} \times \mu s^{{ - 1}}$$
$$q_{{{\text{av}}}} \left( {9{\text{ MeV}}} \right) = \left( {1.19 \pm 0.02} \right)\;{\text{Gy}} \times V^{{ - 1}} \times \mu s^{{ - 1}}$$

As is known that the energy dependence of flashDiamond is limited to ± 1% over the clinical electron beam qualities from 6 to 18 MeV [32], it can be claimed that UCD has similar energy dependence for 7 and 9 MeV beams.

UCD linearity was verified by exposing it in air, covered by a 15 mm thick build-up cap, at different positions in front of the 9 MeV beams. The positions were previously characterised using a combination of radio-chromic films and the toroidal coil-based monitor system of the ElectronFlash machine. As shown in Fig. 8, UCD behaves linearly up to a delivered dose per pulse of about 25 Gy at least. As the pulse duration was 4 µs, this corresponds to about 6 MGy s−1.

Fig. 8
figure 8

UCD linearity in terms of dose per pulse

Full size image

Preliminary assessment of the UCD (6 mm diameter) angular response was done by exposing the detector in air, as schematised in Fig. 9. As shown in Fig. 10, the response loss due to the incidence angle is only 7% when the incidence angle varies from 0° to 60°. To thoroughly assess this characteristic, an experiment in water is planned for the near future.

Fig. 9
figure 9

Experimental setup for the study of UCD (6 mm diameter) angular response

Full size image
Fig. 10
figure 10

UCD (6 mm diameter) angular response. Values are normalized to their maximum obtained at α = 0°

Full size image

UCD radiation hardness was studied at 9 MeV in air by exposing the device in a fixed position and by gradually increasing the accumulated absorbed dose up to 200 kGy. The delivered dose was estimated using a combination of radio-chromic films and the toroidal coil-based monitor system of the ElectronFlash machine (Fig. 11).

Fig. 11
figure 11

Evolution of the UCD (6 mm diameter) response as a function of the accumulated dose at 9 MeV. In X-axis is reported the absorbed dose determined from the toroidal coil-based monitor system, after calibration against radiochromic films. In Y-axis is reported the UCD reading per unit absorbed dose normalized to the same value for the virgin detector 

Full size image

A decreasing trend of response as a function of the accumulated dose could be suspected, but the response after 200 kGy is still compatible with that of the virgin detector.

4 Conclusions

The Unbalanced Core Detector (UCD) was proposed as a novel type of electron dosimeter for FLASH Radiotherapy. This device was tested using 7 and 9 MeV flash electron beams from an ElectronFlash accelerator. The measured spatial dose distributions in water phantom are equivalent to those obtained with the existing flashDiamond dosimeter. In addition, the UCD demonstrated very high radiation resistance, linear operation up to 6 MGy s−1, and the possibility to acquire the time evolution of a single radiation pulse. The angular response was preliminarily assessed in air, finding a response loss of 7% when the incidence angle varies from 0° to 60°. To thoroughly assess this characteristic, an experiment in water is planned for the near future.

Although more investigation is needed to extend the understanding of the energy dependence and to improve the temporal description of the pulses, UCD can be proposed as a convenient and highly qualified dosimeter in view of the clinical translation of FLASH Radiotherapy.