Single-layer Compton camera based on a 1 mm thick pixelated CDTE detectors

Compton camera is a novel γ-camera paradigm that relies upon the kinematics of Compton scattering for image reconstruction. A Conventional Compton camera uses two layers of sensors - the Absorber and the Scatterer. This work reports the proof of concept of a single-layer Compton camera (SLCC) where simultaneous Compton pairs events are registered in a single High-Z semiconductor sensor. The Hybrid pixel detector (HPD) is 1 mm thick, 256 x 256 square pixels with 55 μm pixel pitch CdTe sensor bonded to a Timepix3 readout ASIC. The superior spectroscopic imaging and fast timing capabilities of the Timepix3 readout ASIC coupled with a microscopic and highly pixelated CdTe enable simultaneous event detection of multi-energies occurring at multiple positions. The concept was exemplified by measuring the 122 KeV γ-ray emitted from a 57Co radioisotope source at two positions. Data were captured in the Timepix3 data-driven mode with the KATHERINE readout system via Gigabyte Ethernet data transfer. A bespoke Compton kinematics criterion algorithm implemented in Python 3 IDE was used for data analysis and Compton’s image reconstruction. Numerous events (5.2 million) were captured for 30-minute acquisitions. However, due to the thin nature of CdTe, fewer events (≈ 0.01 %) met the Compton event selection criteria. Nevertheless, the algorithm accurately pinpointed the radioisotope’s location, demonstrating proof of concept of the SLCC system.


Introduction
Compton camera is a novel gamma camera concept that relies on Compton scattering kinematics for image reconstruction instead of physical collimation as in a conventional gamma camera system [1,2,3].This feature gives the Compton cameras greater detection efficiency and a wider field of view.A conventional Compton camera system uses two detectors; the first detector acts as Scatterer, while the second is the Absorber.During the Compton scattering process, the recoil electrons deposit their energies in the scatterer detector, and the scattered photons that escaped the first detector are absorbed and deposit their energies through the photoelectric effect in the absorber detector.The single-layer Compton camera (SLCC) is a Compton camera variant where both the recoil electron and scattered photon were detected in a single detector system [4].A hybrid pixel detector with a high-Z CdTe sensor bonded to a Timepix3 readout ASIC makes the SLCC possible.This is because the CdTe is sufficient to stop the Compton event in a single volume, and the Timepix3 permits simultaneous detection of energy and time occurrence of the event with high precision (Time of Arrival (ToA) = 1.562 ns).In an SLCC system (figure 1), Compton pairs deposit their energies in two 3D coordinates, i.e., scattering (x1, y1, z1) and absorption (x2, y2, z2) coordinates inside the detector volume.The imaging principle of an SLCC (see figure 1) relies upon the Compton scattering angle � = arccos �1 − ���, where    2 = 511 keV is the electron rest mass energy whilst  1 ,  2 are the deposition energies of scattering and absorption events, respectively.Solving the scalar product of vectors connecting the absorption to the scattering coordinates for each Compton pair generates a Compton cone on the projection plane.Iterating the process for the identified valid Compton pair events produced many Compton cones, and the intersections of these cones represent the origin of the incident photon.Since the summed energy of scattering and absorption events is equal to the incident photon, thus, the radioisotope that produces the incident photons is also known.This paper presents the demonstration of an SLCC with a 1 mm thick CdTe detector coupled to a Timepix3 readout ASIC.The data acquisition system, the experiment setup, and the data processing method are explained, and the results will be discussed.1 shows the data acquisition system and photon source used in this paper.The hybrid pixel detector is a 1 mm thick, 256 × 256 pixels with 55 μm-pitch CdTe bonded to a Timepix3 readout ASIC.The hybrid pixel detector is calibrated per pixel for energy and corrected for time anomalies (time-walk in the Time-of-Arrival (ToA) and column delay).Data comprising photon energy (ToT: Time-over-Threshold) and time occurrence of the event (ToA: Time-of-Arrival) were acquired in Timepix3's data-driven mode with the Katherine readout and Burdaman software [5].Two experiments were conducted; EXP 1, the 57 Co was placed 50 mm in the x-axis away from the detector at a vertical distance, ZL = 64 mm (figure 2) and EXP 2, where the 57 Co is placed 50mm below the detector with the same vertical distance.The acquisition time for the EXP 1 and EXP 2 are 3600 s and 1800 s, respectively.

Compton event selection criteria
Data were processed with an algorithm written in Python 3 IDE utilising Numpy, Pandas, Scipy and Matplotlib packages for data handling, processing, and visualisation.Only events comprising two interactions that coincide within the clustering time were selected as potential candidates for the valid Compton events.The clustering time is the travel time of the electron-hole pair drifting across the detector thickness.For this case, the clustering time is assumed at around 32 ns, calculated from the electron mobility   , bias voltage   and detector thickness,  relationship �clustering time =  2     � �.From these coincidence groups, energy cuts were implemented to select events with summed energy equal to incident photon energy: and exclude events belonging to the CdTe X-rays fluorescence lines: The Compton edge: was used to identify events belonging to the scattering ( 1 ) or absorption ( 2 ).The hybrid pixel detector provides information on the energy, time (ToA) and 2D pixel coordinate (x, y) of the Compton pair events.The scattering event was assumed to occur at the surface (z1 = 0), and the z2 is calculated from the depth difference between the two events (z2 = −1 * depth difference).The depth difference is given by  �  1 −  2   �, where  1 ,  2 are the arrival time of  1 and  2 .The   is the time taken by the charge cloud to drift across the detector thickness, which is equal to the clustering time of 32 ns.Using the Compton scattering equation, the scattering angle θ was computed from the 3D coordinate of the Compton events.In addition to the energy cuts and XRF removal described earlier, two crucial criteria were implemented to obtain accurate image projection: i.
A sizeable distance between the Compton event: 2D pixel distance ≥ 15 pixels.
Criteria (i) assumed the scattering event occurred at the top following the convention used in the traditional Compton camera system and (ii) to filter the XRF events.

Compton cone projection
In figure 1, vectors: connect the scattering and absorption events to the Compton cone.The 3D coordinate (x, y, z = ZL) is the variable coordinate on the Compton cone surface, and the Compton cone is the scalar product of  �⃗ and  ⃗ given as: Squaring Equation ( 6) yields: Equation ( 7) can be expressed into a second-order quadratic equation as: where A, B, C, D, E and F are quadratic coefficients computed from 3D coordinates and scattering angle  for each identified valid Compton events.The solution of equation ( 8) is a conic section representing the intersection between the Compton cone and the projection plane.Iterating this process for the identified Compton event produced many conic sections, and the intersection of these conic sections produced an image revealing the origin of the incident photon.Figure 3 shows the energy spectra of 57 Co taken with the 1 mm CdTe bonded to Timepix3 readout ASIC.The photopeak of the 57 Co γ-rays is at 123.97 ± 8.46 keV, whereas the XRF peak is at 21.86 ± 6.00 keV.The energy resolution (FWHM) at respective peaks was used as the range of the energy and XRF cuts to select the valid Compton event.Table 2 shows the analysis of the number of events detected in both experiments.The 1 mm thick CdTe recorded many events, but mostly from the photoelectric interactions since this process is the dominant photon interaction at 122 keV in a Z = 48 CdTe sensor.Less than 0.2% (EXP 1 = 8283 & EXP 2 = 7823) were considered as candidates for valid Compton events as these events are two events that coincide within the 32 ns clustering time.Only ≈ 11% of the possible candidates pass the energy and XRF cuts, and ≈ 600 events passed the θ and 2D pixel distance criteria.

Result and discussion
Figure 4 shows the reconstructed Compton images from the two experiments.Despite limited events, the source origin was sufficiently pinpointed in reconstructed images in both experiments.The reconstructed image's lack of sharpness was due to fewer data and the uncertainty in the  measurement.Longer measurement time and (or) thicker sensors would improve the data statistics.The detector's resolution (spatial and energy) and the Doppler broadening affect the  measurement.The spatial resolution generates divergence between the measured  and actual , which can be higher than 6 ° for a 55 μm-pitch and at 50 mm source to detector distance [6].The energy resolution and Doppler broadening affect the accuracy of the energy measurement.In this paper, the detector was biased directly from the Katherine readout, which is maximum at −300 V.The detector is specified for −500 V; therefore, the obtained resolution is not optimal.The Doppler broadening occurs in the energy spectra of the scattering and absorption events.This effect is due to the incident photon interacting with moving electrons that are bound to atoms [7].

Conclusion
A demonstration of a single-layer Compton camera system with a 1mm thick CdTe bonded to a Timepix3 readout ASIC irradiated with 122 keV γ-rays from 57Co has been presented.Due to the thinness of the CdTe sensor used, only a limited number of valid Compton events were obtained as the photoelectric absorption is the dominant interaction at 122 keV.A careful selection of the scattering angle and a large separation between the two coincidence events were used besides the energy and XRF cuts to improve the accuracy of the image projection.

Figure 2 .
Figure 2. The experiment setup for the SLCC demonstration.

Figure 3 .
Figure 3. Energy spectra of 57 Co showing the main photopeak at 123.97 keV and the CdTe X-ray fluorescence around 21.86 keV.

Figure 4 .
Figure 4.The reconstructed Compton image from the two experiments.

Table 1 .
Data acquisition system and photon source used in the SLCC demonstration.

Table 2 .
Analysis of number of events obtained from the two experiments.