"PHOTON RADIATION THERAPY MONITORING APPARATUS"

Technical Field

The present invention relates to an apparatus designed for dose verification or monitoring photon radiation therapy.

Summary

The present invention describes a photon radiation therapy monitoring apparatus (1 ) comprising a detector-head, comprising a collimator (7), a detector (6) and readout electronics (16), arranged for collimating and detecting photons scattered or emitted by the radiation target (4) at an angle with the central axis of the beam (2) of said radiation therapy.

A preferred embodiment further comprises a data acquisition system (5) arranged to calculate accumulated photon intensities in the target image space.

In a preferred embodiment said data acquisition system (5) is further arranged to obtain a time correlation between detected events (r _y ) and the respective beam (2) incoming positions (rt,), and arranged to calculate by spatial correlation from the 2D information of each detector head of the time-correlated detected events (r _Y ) a 1 D, 2D, 3D, or 4D image with accumulated photon intensities in the target image space.

A preferred embodiment further comprises a plurality of said detector-heads.

In a preferred embodiment said collimator (7) has its septa placed exteriorly to the detector (6).

In a preferred embodiment said collimator (7) has its septa prolonged into the detector (6).

In a preferred embodiment said collimator (7) septa are thinner on the inside of the detector (6).

In a preferred embodiment said detector (6) comprises any one or any combination of either one or multiple: gas-filled detectors, liquid-filled detectors, scintillator detectors, or semiconductor detectors.

In a preferred embodiment said apparatus (1 ) is physically supported by an arm (8) attached to the support structure (1 1 ) of the LINAC of said radiation therapy.

In a preferred embodiment said apparatus (1 ) is physically supported by a telescopic arm (12) sliding in an extensible runway (13) attached to the support structure (1 1 ) ) of the LINAC of said radiation therapy. In a preferred embodiment said apparatus (1 ) is attached to a support (14) which is independent of the support structure (1 1 ) of the LINAC of said radiation therapy.

In a preferred embodiment said device (1 ) is positioned for detecting photons escaping the target (4), at an angle, to the central axis of the beam (2) of said radiation therapy, of 45° to 135°, or 60° to 120°, or 85° to 95°.

In a preferred embodiment said device (1 ) is positioned for detecting photons escaping the target (4), substantially perpendicular to the central axis of the beam (2) of said radiation therapy.

A preferred embodiment further comprises a noise photon absorber (17) between the detector (6) and the target (4).

A preferred embodiment further comprises shielding material (18), in particular between the LINAC head (9) of said radiation therapy and the detector (6) and/or readout electronics (16).

In a preferred embodiment the data acquisition system (5) is arranged to time- correlate and differentiate between the noise signal received directly from the LINAC head (9) of said radiation therapy and the signal received from the target In a preferred embodiment the data acquisition system (5) is arranged to time- correlate and differentiate the spatial location of each detected photon between the timings of the signals received according to the moment of detection of each photon.

Background Art

The handling of both single-fraction and multiple-fraction radiation therapy (RT) requires extreme care in order to guarantee maximum dose exposure to the tumour, while maintaining minimum healthy-tissue dosage so to limit undesirable side-effects. Nevertheless, there is growing, documented knowledge stating that several physical and biological mechanisms disturb that precision requirement to an unknown and unpredictable extent. Examples of such RT-disturbing imprecisions range from patient mispositioning, or beam range uncertainties, to anatomical morphological changes occurring during the course of the fractionated treatment. The latter include e.g. biological responses such as tissue swelling due to oedema and inflammation, tumour shrinkage, and/or filling of initially air-filled body cavities with unaccounted mucus or oedematous tissue. These and other morphology-changing factors, such as respiratory or bowel motion, constitute the main challenge to the development of modern RT techniques, and are the main limitation to the improvement of the clinical outcomes associated with present-day and next-generation RT.

Consequentially, it is the desire of any radiotherapist to be able to detect and even quantify the dose-changing mechanisms just stated. Such knowledge would allow the improvement of RT clinical outcomes by continuously supporting the medical doctor in its decision to: 1 ) stop and re-plan a given treatment that has evolved to a situation that does not fulfil the initial dose assumptions, or 2) proceed the treatment with reassured quality (an important variable both for the RT team and to the patient). State-of-the-art technology aiming at this task is named image-guided radiation therapy (IGRT), comprising so-called in-beam and off-beam positron emission tomography (PET). IGRT has its own potentialities, limitations, and even side-effects due to extra-dosage in the case of repeated mega- and kilovoltage computed tomography (CT) in some IGRT systems. Finally, none of these techniques is able to provide a complete answer to the quest of detecting the dose-changing mechanisms mentioned above.

The rationale for RT monitoring arises from several facts. In a fractionated treatment-course, patient misalignments and changing internal anatomy are becoming more critical since higher conformality holds a higher risk of target underdosage or organ at risk (OAR) overdosage. Even with rigid fixation devices, maximum positioning errors higher than 1 cm are observable (Thieke et al. 2006). One example of changing internal anatomy was already described elsewhere and it was reported the existence of oedematous swelling of the tumour occurring during the first 3 days of treatment to the point of affecting the ventilatory function of the patient. If such swelling is located proximal to the target volume (before it), tumour underdosage most probably will occur.

An opposite example, resulting in dangerous overdosage of a distally positioned OAR (brainstem) due to tumour regression was also described elsewhere. In addition to the macroscopic critical issues mentioned, microscopic dynamics in the irradiated tissues also occur. Indeed, radiation induces a pro-coagulant and pro-inflammatory environment. Consequently, the normal tissue that is included in the radiation field changes dramatically from the time of delivery of the first fraction to the delivery of the last fraction. In other words, the normal tissue that is irradiated at the beginning of the radiotherapy course may be very different from the normal tissue that is irradiated towards the end.

As a response to the consecutive, fractionated irradiation burden the pro- coagulant and pro-inflammatory environment results in increased vascular permeability. Consequently, higher blood presence in irradiated tissues may be expected progressively with fraction number. In the brain, this may represent a density increase of up to 2.7 % along the beam path (specific gravity of hemispheres and whole blood is 1.0335 and 1.0621 , respectively). This density increase may rise further if red blood cells (specific gravity between 1.093 and 1.096) are mobilized to the inflamed region. Another inflammation-related response is oedema. This represents a significant increase in density for irradiation of the lung (typical oedema versus lung specific gravity are 1.0 and approximately 0.3, respectively).

For brain treatments, however, two scenarios are possible. If the oedematous region(s) involve(s) empty cavities like e.g. the paranasal sinuses, then the beam must traverse higher densities than predicted in the treatment planning. Tumour underdosage most probably will occur. On the contrary, if the oedematous region(s) involve(s) brain tissue, then the beam must traverse material with lower density than predicted (approximately -3 %) yielding, consequently, probable overdosage to distal organs-at-risk. Consequences of tumour underdosage are potentially decreased tumour control rates.

IGRT became a fundamental requirement in any radiation therapy treatment (Evans 2008). During delivery any type of intra or inter-fraction deviations relatively to the planning computed tomogram (CT) will reduce the effectiveness of the treatment and may cause severe patient complications. Different methods have been developed to assess such patient variations. Fast 2D image guided techniques are available since the 80's with the integration of portal imaging devices in linear accelerators (LINAC, van Elmpt et al. 2008). However, these provide only bony structures deviations. Any anatomical deformations, organs movement or tumour response during treatment cannot be accounted for.

More sophisticated imaging techniques are then required. 3D IGRT, with or without functional tumour information, becomes then an essential step in the treatment process and even more when new treatment techniques are delivered. With the advent of intensity modulated radiation therapy (IMRT) highly irregularly-shaped dose distributions with steep dose gradients in the tumour borders can be produced. Cone-Beam CT tools, using either the MV treatment beam or perpendicularly positioned kV X-rays sources, produce a 3D online patient representation (Jaffray 2005). Using this technology, which requires extra dosage in respect to the treatment planning, daily 3D in vivo dosimetry has been investigated and it is becoming standard practice in some clinical centres (McDermott et al. 2008).

Alternative solutions include CT or magnetic resonance imaging (MRI) scanners inside the treatment rooms or integrated in the LINAC. Each of these techniques has its imaging potentialities and limitations, presenting even side effects due to extra-dosage in the case of both MV and kV CT (discussed in the next paragraph). Although ionizing dose is not an issue with MRI , repeated MRI scanning for IGRT does represent an extra patient and hospital burden. An MRI integrated in the LINAC does not have any of these drawbacks, but it is not implemented yet at the clinical level and, additionally, the associated costs may be expected to be high.

Furthermore, new European directives (2004/20/CE) restricting exposure to electromagnetic fields may pose working limitations affecting the usage of high magnetic fields present in MRI devices and its neighbouring areas. A completely different concept of treatment machines was implemented with tomotherapy which is based on rotation fan-beam delivery. This concept allows the inherent acquisition of CT images for image guidance (Hong et al. 2007). Nevertheless, a non-negligible extra dose exists. This dose is similar to that deposited with low-energy (kV) CT, which is further accumulated during fractionated treatments if repeated, daily scans are necessary for IGRT. For the particular case of moving targets, like lung tumours, 4D gating and tracking solutions are under development through increasingly sophisticated technological approaches. Several systems based on feedback mechanisms that enable online adaptation of the radiation delivery are becoming available.

Also new methods for in vivo dosimetry based on the physical interactions occurring during irradiation are being tested like the in-beam PET applied to photon-RT. This is based on the emission of positrons produced by photoneutron reactions occurring in the nuclei of C-12, 0-16 and N-14 of the irradiated tissues by high energy photons.

Nevertheless, the cross-section for photoneutron reactions start at about 20 MeV, rendering photon-RT monitoring by means of in-beam PET useful only with LINACs of higher endpoint energy. Most linear accelerators have two photon endpoint energies of 6 and 15 or 18 MeV, meaning that the energy for the reactions necessary for in-beam PET is mostly inaccessible. Very few newer LINACs with endpoint energies reaching 50 MeV may prove useful for the task of photon-RT in-beam PET but, even so, (1 ) the number of events induced with typical treatment doses seems to be a major challenge, and (2) most treatments worldwide are performed with the above-mentioned lower photon endpoint energies. In addition to the challenges already mentioned for in-beam PET, biological mechanisms, reduced collected statistics from an already-poor induced β+ activity, and image artefacts resulting from limited- angle, in-beam PET geometries yield images of difficult interpretation and consequential limited clinical usefulness unless time-of-flight with ultra-high resolution can be applied.

Bibliography

- Evans PM 2008 Anatomical imaging for radiotherapy Phys. Med. Biol. 53 R151-R191 - Hong TS, Welsh JS, Ritter MA, Harari PM, Jaradat H, Mackie TR, Minesh PM 2007 Megavoltage computed tomography: An emerging tool for image- guided radiotherapy Am. J. Clin. Oncol. 30:6 617-23

- Jaffray DA 2005 Emergent Technologies for 3-Dimensional Image-Guided Radiation Delivery Semin. Radial Oncol. 15:3 208-16

- Leemans WP, Nagler B, Gonsalves AJ, Toth Cs, Nakamura K et al 2006 GeV electron beams from a centimetre-scale accelerator Nature Physics Letters ! 696-699

- McDermott LN, Wendling M, Nijkamp J et al 2008 3D in vivo dose verification of entire hypo-fractionated IMRT treatments using an EPID and cone-beam CT Pad/other. Oncol. 86:1 35-42

- Thieke C, Malsch U, Schlegel W, Debus J, Huber P, Bendl R, Thilmann C 2006 Kilovoltage CT using a linac-CT scanner combination Br. J. Radiol. 79 79-86

- van Elmpt W, McDermott L, Nijsten S et al 2008 A literature review of electronic portal imaging for radiotherapy dosimetry Radiother. Oncol. 88:3 289-309

Disclosure of the Invention

For the reasons already mentioned, a system capable of daily monitoring fractionated RT is highly desirable and put forward here. Such system allows for the improvement of treatment outcomes by being able to detect deviations from the treatment plan with high accuracy and reliability. State-of-the-art technology aiming at this task includes IGRT, and so-called in-beam PET. However, IGRT presents limitations and side-effects, which this system can overcome by selecting photons that are scattered or emitted by the radiation target, and escape with a direction approximately perpendicular to the central axis of the beam direction. Such dose-free monitoring system is proposed here and will be hereafter referred to as apparatus.

Spatial profiles constructed with photons collected with this apparatus allow to monitor either in real-time or not 1 ) the dose being delivered, and 2) undesired, dose-changing anatomical morphological changes occurring statically and/or dynamically.

The principle-of-operation and feasibility details of the apparatus allow for it to monitor each field (static or dynamic) in both real-time and post-treatment during all treatment fractions without whatsoever additional dose imparted onto the patient, a remarkable advantage. In addition, the fact that the events used for image construction are promptly collected renders the method fully insensitive to the dynamical, biological mechanisms that affect in-beam PET. In other words, the diffusion, washout, and/or transport biological mechanisms that disturb the dose-correlated activity with in-beam PET do not play any role with the imaging method provided by this apparatus.

The invention device (1 ) for monitoring photon radiation therapy comprises a radiation detector (6) plus a collimator (7), readout electronics (16) and a data acquisition system (5), supported by a fixed or telescopic arm attached to the LINAC support structure (1 1 ), or by an independent support. Said detector (6) can be a gas-filled detector, a liquid-filled detector, a scintillator material, a semiconductor, or a combination of the mentioned materials. Between the LINAC head (9) and the said collimator (7) a radiation-absorber material (17) can be placed. Also, next to said device (1 ), between it and the LINAC head (9) a shielding material (18) can be used. The collimator (7) allows for photons that escape the target, e.g. the patient (4) on a couch (3), to be collected only if they are emitted approximately perpendicular to the beam direction. Said photons have their origin in well known physics processes such as Compton scattering, pair production, or bremsstrahlung radiation originated from electrons, ejected after single or multiple interactions of the incoming beam with the patient. The photons that reach the detector (6) are considered to fulfil the angular requirement after being collimated by said collimator (7) whatever physical process origin they may have. A time correlation is obtained between each detected event (r _y ) and the beam (2) incoming position (rb), allowing for the 2D information given by each detected event (r _Y ) to be spatially correlated with the image-space of the target in 1 D, 2D, 3D or 4D.

The target is typically a patient, but other bodies may be used, namely for dosimetric purposes comprising LINAC testing and calibration, so therefore for the purpose of this invention patient and target are interchangeable.

The readout electronics (16) allows outputting a photon detection image, and may comprise a full electronic photo-detector plane or may only comprise a front-end of such electronics. Typically this readout electronics comprises a number of pixel detectors, preferably a large number of pixel detectors, but other structures may be possible. Also, the pixels may be square or not at all, namely staggered, hexagonal or even comprising subpixels. Alternatively, the detector may dispense with discrete pixels and even be continuous.

The implementation of this apparatus in a clinical RT environment consists in a single, or multiple, detector head positioned according to dispositions, in particular, in order to form an angle between the collimating direction, usually defined by the holes of the collimator, and the direction of the beam central axis preferably within the range 45° to 135°, preferably within the range 60° to 120°, more preferably in the range 85° to 95°. Each detector head allows for photons that are scattered or emitted by the target (e.g. the patient) to be collected if they escape preferably approximately perpendicular to the direction of the beam central axis. This function is accomplished by means of a collimator positioned in front of the photon detector. The collimator may either be prolonged into the detector, or it may finish before the detector.

Preferably, best results are obtained by substantially perpendicular angles between the collimating direction, usually defined by the holes of the collimator, and the direction of the beam central axis, and thus orthogonal operation, but other angles are possible, in particular for accommodating specific LINAC configurations and physical building constraints, or even specific physiological or patient requirements.

Preferably, the detector is a square plane but other shapes are perfectly possible, like a rectangle plane or a curvilinear cylindrical surface preferably with a long axis parallel to the beam central direction (2), or like a spherical or elliptical shape, being obviously the construction of the collimator necessarily adapted to these shapes.

In addition to the detector details mentioned above, the readout electronics of the system may be utilized to assist in the imaging strategy of this RT- monitoring modality. It is the task of the data acquisition system (DAQ) to provide a time correlation between each detected event (r _y ) and the beam incoming position (rb) either in a step-and-shot approach or in a continuous beam delivery radiotherapeutic system. This correlation, hereafter referred to as slow time-correlation, allows for the 2D information given by each detected event (r _Y ) to be spatially correlated with the image-space of the target (e.g. patient) in 1 D, 2D, 3D or 4D (Figure 8A - detector 2D image and Figure 8B - 1 D in the image-space of the target). This 2D image information may further be used to allow for a real-time image to be constructed as the beam is being delivered at each position.

This is valid both for small irradiation fields, as well as for larger fields. The latter will profit from multiple detector heads positioned in particular according to the above referred dispositions in order to improve the spatial resolution at the image-space. Alternatively, the detector head may be movable between different positions in order to achieve the intended spatial resolution.

In addition to the referred slow time-correlation, a hereafter referred to as fast time-correlation with the beam delivery may be preferably utilized in order to provide the system with background suppression capability. The background noise here refers mainly to events generated in the LINAC head, which may comprise passive and active beam shaping materials (absorbers) that are patient-specific or any multi-leaf collimator. When background events are detected, they generate a signal that is fully correlated with the LINAC time structure. Consequently, timing techniques may be used in order to discard these LINAC-associated background events from dose-correlated events being generated in the target (patient).

In Figure 6A the comparison of events emitted by the target (patient 4) and arriving to the detector at the instant (to+t-i), in respect to events that come from the LINAC head are depicted. The latter may arrive directly at the detector at an earlier time t ₂ . The moment t ₂ can be therefore directly correlated with the timed radiofrequency used by the LINAC to accelerate the electron beam. Each dose- correlated event arriving at (to+ti) is also modulated by that time structure, but suffers a delay At = (to+ti) - (to ² +ti ² ) ^½ in respect to t ₂ . Considering the distances corresponding to to and ti as typically 1 m and 30 cm, respectively, the time delay At becomes measurable and again typically of the order of 1.5 ns. Discrimination can therefore be accomplished by means of one or a combination of the following: (1 ) signal shaping with appropriate fast time constants at the front-end electronics, (2) delayed gated-integration applied to events seen by the detector and in correlation with the dose signal, or (3) through digital pulse processing applied to each signal delivered by each pixel of the detector head.

The Figure 6B has typical details about this discrimination signal processing and depicts the typical signal evolution, S, resulting from background events, B, and dose-correlated events, D, and a pulse repetition period of 10 a.u. (arbitrary unit) is considered, together with a delay between B and D of 1.5 a.u. (signal timings not to scale and are for illustrative purposes). The origin and timings of B and D are shown in Figure 6A and as also previously explained. The Figure 6B shows directly that digital pulse processing and/or differentiation of S allows for background events B to be separated from dose-correlated events D. Another possibility is the utilization of gated integration during the rise and fall of pulse D only (signal forced to ground outside the gate).

A time-correlation complimentary to those presented before consists in an hereafter referred to as ultra-fast time-correlation that preferably may be used to further aid resolving the spatial location of each detected photon that scattered in the target (patient 4). The principle of operation is depicted in Figure 7A, which depicts an implementation of an ultra-fast time-correlation between the moment of detection of each photon that escapes from the patient (represented by t-i , t ₂ or t.3, with t-i < t ₂ < t.3 in respect to to) and an instant to proportional to the moment of beam input into the target region. The current time structures of linear accelerators used for photon radiotherapy do not allow resolving to with a precision inferior to typically 3 μβ. If this time resolution decreases to 100 ps full- width half maximum (FWHM) or less, e.g. LASER-driven accelerators (Leemans et al. 2006) (Figure 7B) but not only, the principle of ultra-fast time correlation then allows for a localization of each photon along the line-of-sight of the collimator that the photon traversed.

Brief Description of Drawings

The following figures provide preferred embodiments for illustrating the description and should not be seen as limiting the scope of invention.

Figure 1 depicts the principle of operation of the RT monitoring system proposed in the present invention.

Figure 2A shows a sagittal view of an alternative embodiment of an RT monitoring system according to the present invention, corresponding to its positioning in the front part of the LINAC.

Figure 2B shows an anterior transverse view of the embodiment depicted in Figure 2A.

Figure 2C shows a posterior transverse view of the embodiment depicted in Figure 2A. Figure 3A shows a sagittal view of an embodiment of an RT monitoring system according to the present invention, where its supporting arm is attached to the LINAC support structure (1 1 ) with telescopic positioning capability (12).

Figure 3B shows an anterior transverse view of the embodiment depicted in Figure 3A.

Figure 3C shows a top view of the embodiment depicted in Figure 3A.

Figure 4A and figure 4B show sagittal views of other alternative embodiments of an RT monitoring system according to the present invention, where its positioning is independent from both the LINAC and the patient couch.

Figure 5A depicts a possible configuration of an RT monitoring system according to the present invention, where the collimator (7) septa are prolonged into the detector (6).

Figure 5B depicts an alternative configuration of an RT monitoring system according to the present invention, where the collimator (7) septa are prolonged into the detector material (6), but are thinner inside the detector (6) than outside it.

Figure 5C depicts another alternative configuration of an RT monitoring system according to the present invention, where the collimator (7) septa finish before the detector material (6). Figure 6A depicts a schematic representation of the fast time-correlation principle.

Figure 6B shows an example of signal analysis using fast time-correlation principle.

Figure 7A depicts a schematic representation of the ultra-fast time-correlation principle.

Figure 7B shows a comparison between a conventional linear accelerator and a LASER-driven accelerator in terms of accelerator timing macrostructure.

Figure 8A depicts counts in a detector placed behind a collimator.

Figure 8B shows the integration in the X-axis of the graph in Figure 8A between -15 mm < X < 15 mm and the comparison with the dose profile curve obtained with the same phantom as the used for the Figure 8A and Figure 8B counts curve.

Figure 9 depicts four graphs where several morphological changes were tested of device performance in detecting such morphological changes for 6-MV and 18-MV LINACs.

Figure 10 shows depth-dose profiles versus expected photon yield and measured relative pulse height in arbitrary units. The measurements were performed under a clinical radiotherapeutical environment. The experimental agreement with the expected results and with the depth-dose profile is remarkable and reveals the technological advantages of this radiotherapy monitoring technique.

Modes for Carrying Out the Invention

With reference to the attached drawings, a device 1 is provided consisting of a detector 6 plus a collimator 7 and readout electronics 16, and a data acquisition system (DAQ) 5. The readout electronics (16) may comprise a full electronic photo-detector plane or may only comprise a front-end of such electronics.

Figure 1 shows the general scheme of implementation of the device 1 into a clinical RT environment. The radiotherapeutic photon beam 2 impinges the patient 4 from the left. Photons escaping the patient 4 will hit a pixel of the device 1 only if they leave the patient 4 with an angle close to 90°, in reality as much as determined by the collimator 7, in respect to the photon beam 2 central axis direction. The spatial correlation of the 2D position of each photon hitting the device 1 (r _Y ) and the 2D position of the photon beam 2 (rb) allows an image to be constructed in the target-space in real-time. The 2D position of the photon beam 2 (rb) that must be delivered to the DAQ 5 must preferably contain the position of the LINAC support structure 1 1 , the expected exit position of the photon beam 2 (including each position of the multi-leaf collimator for IMRT), and the beam current being delivered.

Although the above referred scheme shows one single-head device 1 only, additional devices 1 positioned according to the previously referred dispositions in respect to the photon beam 2 central axis direction can be used. In addition, each single-head may or may not be made of multi-pixel detectors coupled to each collimator 7 hole. Therapeutic monitoring without information from the beam delivery is also possible with the present proposed apparatus (Figure 8A and Figure 8B), but in this case dispensing with the previously referred preferable slow time-correlation aspects.

Also in Figure 1 , a photon absorber 17 and a shielding material 18 are depicted. The former is to absorb photons that come from the patient with a certain amount of energy, and the latter serves to block the passage of photons from the LINAC head 9. The latter is usually used in this technical field to improve signal and reduce noise.

Figures 2A, 2B, and 2C depict three views of a possible implementation of the device 1 into a clinical RT environment. This also comprises generally applicable preferable aspects of the invention. The setup corresponds to the positioning of the device 1 onto the LINAC support structure1 1 and respecting the patient 4 spacing. Figure 2A depicts the sagittal view, where the LINAC support structure 1 1 , the patient 4 position, and the device 1 configuration are shown. The LINAC support structure 1 1 is vertically displayed and contains the LINAC head 9 where the electron beam is accelerated and the radiotherapeutic photon beam 2 is generated. Vertically opposite to it lies the counterbalance 10, potentially containing the electronic portable imaging device (EPID) (not shown). The photon beam 2 comes from the top and interacts with the treatment area in patient 4, resulting in photons that hit the device 1.

In Figure 2A, a schematic front view of the device 1 is shown, represented by a multi-hole collimator 7 of variable geometries (e.g. hexagonal, squared, circular, triangular) filled in the back by a detector material 6. The device 1 is substantially aligned with the LINAC support structure 1 1 , forming with it an angle according to dispositions previously referred, regardless the rotation angle of the LINAC support structure 1 1 . The patient couch 3 and the device 1 can be adjusted in height in respect to each other.

Figure 2B depicts the transverse view. The side view of the device 1 , containing the layer of the detector material 6 divided by the collimator 7, is shown. As observed in all the views, the device 1 is connected to the LINAC support structure 1 1 by an arm 8 that is attached or encrusted into it, potentially telescopic. Figure 2C shows the arm 8 in form of an encrusted claw that embraces the LINAC support structure 1 1 and holds the device 1 . An alternative hypothesis is to attach the support of the device 1 to one side only of the LINAC support structure 1 1 (not shown). Still an alternative configuration is to implement the device 1 supporting arm 8 encrusted into the front of the LINAC support structure 1 1 , instead of in its back (also not shown).

Figures 3A, 3B, and 3C show a possible implementation of the device 1 into a clinical RT environment, where it is located at the front of the LINAC support structure 1 1 . This also comprises generally applicable preferable aspects of the invention. The radiotherapeutic photon beam 2 impinges the patient 4 from the top. Photons escaping the target (patient 4) will hit the device 1. In Figure 3A the device 1 is attached to the LINAC support structure 1 1 , and it has a telescope 12 that takes the device 1 close to the patient 4. The device 1 park position may be encrusted into the LINAC support structure 1 1 , as partly represented in Figure 3A. Figure 3C corresponds to the BEV (BEV is an acronym for beam's eye view) and represents the possibility of the device 1 to rotate in particular up to ±90° through an extensible runway 13. The patient couch 3 can also be moved in order to optimize therapy. Figure 3C shows the longitudinal view of the setup, where the multi-hole collimator 7 and detector 6 composing the device 1 can be seen.

Figure 4A and Figure 4B show still another implementation of the device 1 into a clinical RT environment. Again, this also comprises generally applicable preferable aspects of the invention. This configuration, by means of a self- supporting structure 14, grants the device 1 independence from the LINAC support structure 1 1 , from the LINAC head 9, and from the patient couch 3 without further changes neither to the LINAC support structure 1 1 , to the LINAC head 9 nor to the patient couch 3. This will facilitate its installation at therapeutic sites unable to perform modifications to the LINAC support structure 1 1 and/or to the LINAC head 9. The small wheels 15 at the bottom of the structure 14 for device 1 support are only representative and they may or may not constitute a final solution.

Figures 5A, 5B, and 5C show three basic collimator 7 and associated detector 6 and readout electronics 16 configurations, each representing different compromises between detection efficiency and image resolution. Figure 5A shows the collimator 7 - detector 6 and readout electronics 16 combination that provides the best image resolution. It only accepts substantially orthogonal photons, meaning that its efficiency of photon detection may be worse than other configurations. In Figure 5B the detector 6 volume is slightly increased at the expense of a reduction of the thickness of the collimator 7 walls (only in the detector-filled region). Finally, in Figure 5C a continuous detector 6 allows for all photons crossing the collimator 7 hole to have a similar detection probability (increased detection efficiency). This is achieved at the cost of inferior image resolution due to 1 ) a larger acceptance for oblique photons, and 2) a larger acceptance of photons that Compton-scatter in the detector 6.

The following claims set out particular embodiments of the invention.