Micrometer-Resolution X-ray Imaging Enabled by a Flexible Perovskite Screen.

Spatial resolution improvement has been keenly sought recently in the perovskite-based scintillation community. Here, micrometer resolution (∼2.0 µm) was achieved by using an X-ray imaging screen of self-assembled perovskite nanosheets. The assembly behavior of nanosheets was applicable to many substrates, including glass, metal, and polymer surfaces. The use of a polymer substrate not only eliminated the parasite absorption of X-ray but also enabled a flexible screen with robust bending stability. The assembly behavior, on the other hand, provided vicinity for an efficient energy transfer between nanosheets of varied thicknesses, as evidenced by both transient absorption and photoluminescence lifetime measurements. Importantly, the ensuing large Stokes shift (∼316 meV) significantly mitigated the reabsorption issue, leading to a comparable light yield to LYSO/Ce crystals. With the aid of the synchrotron-based collimated X-ray beam, the fine structure of two-dimensional objects, such as microchips, was clearly visualized with the flexible scintillation screen. Furthermore, those challenging biological samples were also scanned by phase-contrast imaging, whereby a three-dimensional reconstruction was obtained successfully. Despite the labile nature of the perovskite screen, this work represents the state-of-the-art spatial resolution for perovskite scintillation.

Machine learning based on structural and FTIR spectroscopic datasets for seed autoclassification.

A single feature set is often unable to effectively classify complex biological samples due to their similar morphology and sizes. This paper proposes a protocol for the fast identification of seed medicinal materials based on micro-structural and infrared spectroscopic characteristics. Three different feature datasets, namely micro-CT, FTIR, and mixed datasets, were established via principal component analysis (PCA) and competitive adaptive reweighted sampling (CARS) and then used to train a back-propagation neural network. The mixed dataset consists of 34-dimensional micro-CT eigenvalues and 13-dimensional FTIR eigenvalues, optimized by PCA and CARS processing and then used to train a BP neural network. The results showed that the classification accuracy reached 89.5% for the micro-CT dataset and 93.3% for the FTIR dataset, and the classification accuracy of the mixed dataset achieved 99.2%, much higher than those of the traditional single feature datasets. This study provides a new protocol for multi-dimensional characteristic architecture with excellent performance for the classification and identification of Chinese medicinal materials.