Description
Fine-pitch semiconductor detectors, including CdTe and CZT arrays, are establishing themselves as the leading technology for high-resolution medical imaging. However, reducing pixel pitches below 200 µm worsens charge-sharing effects, where incoming photon energy distributes across multiple adjacent pixels. This spatial dispersion, more pronounced in sub-200um pixels and gamma-ray energy range, causes signals in many peripheral pixels to fall below the electronic detection threshold, inducing significant spectral degradation and systematic distortion.
To overcome this issue, this study introduces a correction methodology that evaluates the spatial symmetry of pixel patterns using a novel circularity metric. Tested using a 2-mm-thick CdTe crystal coupled to a Timepix3 readout chip, individual photon interactions from Am-241 and Co-57 radioisotopes were classified based on both cluster size (multiplicity) and shape symmetry/roundness (circularity). Morphological assessment revealed that spatial asymmetry directly reflects sub-threshold charge losses.
A two-stage correction algorithm was then implemented to restore spectral performance. First, we categorized the detected events by clustering them according to their multiplicity and circularity, to obtain distinct per-pixel energy spectra for each combination of these parameters. Next, in the intra-multiplicity step, we aligned the spectra with lower circularity values to the reference spectrum with perfect symmetry, for each multiplicity level. Finally, an inter-multiplicity step unifies the spectral response across different cluster sizes. As results, the average per-pixel FWHM improved from 16.70% to 10.05% at 59.5 keV (Am-241) and from 9.64% to 6.73% at 122 keV (Co-57). Given its closed-form geometric formulation, this method offers a practical path toward real-time event-by-event spectral correction.