The MPMIB project (Multispectral photon-counting for medical imaging and beam characterization), funded via the Academy of Finland RADDESS programme, focuses on the development of a next generation radiation detection system operating in a photon-counting (PC) mode . The extraction of spectrum per pixel data will lead to higher efficiency and image quality, as well as the possibility to identify different materials and tissue types. Our approach is to construct direct-conversion semiconductor detectors hybridized with read-out chips (ROC) capable of operating in PC mode. Currently, we focus on Cadmium Telluride (CdTe) as a detector material candidate, which is a high-Z material with excellent photon radiation absorption properties.
However, CdTe is a fragile material that can include large concentrations of extended crystallographic defects, such as grain boundaries and tellurium (Te) inclusions, necessitating material assessment prior to the complex procedure of detector processing . As CdTe is nearly transparent in the near-infrared we employ infrared microscopy (IRM) to make Te inclusions inside the crystal visible. Employing a neural network , we identify and classify the defects in the obtained IRM images and visualise the defect distribution in 3D-maps (Fig.1). We are currently comparing the defect distributions to measurement results with transient current technique (TCT) to study the relation between areas of higher defect densities and the charge collection effiency of the detector.
We will give an update on the MPMIB project, discuss IRM results showing defect distributions of CdTe crystals and the possible implications on the performance of the processed detector, as well as present first experimental data obtained with one of our prototype PC detectors in a small tomographic setup. Further analysis and advanced image reconstruction will give us a clearer picture of how our detectors are performing in respect to multispectral imaging.
 E. Brücken et al., Journal of Instrumentation 15 (2020), doi:10.1088/1748-0221/15/02/C02024
 A. Gäddä et al., Journal of Instrumentation 12 (2017), doi:10.1088/1748-0221/12/12/C12031
 Ohtu neural networks. https://github.com/Ohtu-project/Ohtu-neural-networks