Preamble

An approach to overcome the resolution-efficiency trade-off in X-ray scintillator imaging Hong-Quan Zhou Yan-Qing Wu 1, 3, Lu Wang, Hao Shi, Cheng-Qiang Zhao, You He, Jia-Li Long, Yong Wang, Zhi Guo, and Ren-Zhong Tai 1, 3, 1 Shanghai Institute of Applied Physics, Chinese Academy of Sciences, Shanghai 201800, China University of Chinese Academy of Sciences, Beijing 100049, China Shanghai Synchrotron Radiation Facility, Shanghai Advanced Research Institute, Chinese Academy of Science, Shanghai, 201204, China Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences Shanghai 201800, China To address the well-known "resolution–photon efficiency" trade-off, we developed a lens-coupled X-ray tube- based indirect imaging system that incorporated a thick scintillator plate and lens with a large numerical aper- ture. This configuration provided sufficient photon flux while maintaining a theoretically high spatial resolu- tion, thereby reducing the core challenge of resolving the defocusing issue induced by a thick scintillator and approaching the theoretical resolution limit. Two key techniques were developed: (1) generalized point spread function (PSF) restoration, which extended the single PSF recovery method to geometrically magnified X-ray imaging systems, demonstrating its suitability for large-NA configurations; and (2) truncated PSF correction, which eliminated the imaging artifacts caused by severe fabrication defects in ultrathin scintillators through PSF truncation, followed by resolution restoration using experimentally measured PSFs. The experimental re- sults showed that in the high-frequency range, the power spectral density was improved by up to 8.45 times for an image on a thick scintillator. High image quality and photon efficiency were achieved simultaneously, demonstrating the feasibility of this integrated strategy. These results provide a critical pathway for overcom- ing the long-standing resolution efficiency dilemma in indirect X-ray imaging using an X-ray tube source or synchrotron radiation facility.

Keywords

Scintillator, X-ray imaging, indirect imaging

INTRODUCTION

Both direct and indirect imaging techniques are used for high-resolution X-ray detection [ Direct imaging uses semiconductor materials such as CdZnTe to generate electri- cal signals via ionization effects, offering high sensitivity [ However, the high cost of CdZnTe crystal growth and man- ufacturing [ ] limits its large-scale applications. Using cur- rent micro-nano fabrication processes, the pixel size of direct imaging can be reduced to , which is still larger than pixel size of the standard sCMOS cameras used in high-resolution indirect imaging systems. Despite chal- lenges such as light scattering [ ], secondary electron scatter- ing, and fluorescence field depths exceeding the depth of field (DOF) of the optical system, the resolution of indirect X-ray imaging remains higher than that of direct imaging systems.

Therefore, high-resolution indirect imaging systems are still widely used in synchrotron radiation imaging stations and in industrial and biomedical fields because of their high stability and low cost [ In the last decade, micrometer/submicrometer lens- coupled X-ray indirect imaging technology has advanced sig- nificantly. (1) Detector performance has been enhanced by

equally contributed This work was supported by the Shanghai Municipal Science and Technol- ogy Program Project (24JD1402900) and the National Key R&D Program of China (Grant No. 2022YFB3503904, 2021YFA1601003)

optimizing the materials and thickness of scintillator screens, with new scintillators such as LuAG:Ce and GAGG:Ce boosting the light output and resolution [ ZnO:In nanorod arrays suppress optical crosstalk via a pixelated design, combining a sub-nanosecond decay time and high light yield (> photons ), and simultaneously enhancing the temporal and spatial resolutions [ ]. Per- ovskites are excellent indirect imaging detectors because of their high carrier mobility and flexible processability.

The CsPbI Br/CsPbIBr film developed at Zhejiang Uni- versity achieved a dark current of < and sen- sitivity of The flexi- ble Cs detector of NUAA demonstrated a sensitivity with 60-day stability, surpass- ing rigid detector limits [ ].The structured CsI scintillation screen was further optimized, and its comprehensive perfor- mance was significantly improved [ ]. (2) Image recorders have evolved from CCD to CMOS and sCMOS types, which has greatly improved the sensitivity, dynamic range, and imaging speed. (3) Optical lens systems have also been de- veloped. The ZEISS Xradia series employs a two-stage mag- nification architecture, achieving submicrometer resolution (RaaD technology), while maintaining a large working dis- tance (> 10 mm ). Optical lens structures have also been de- veloped with designs such as large-numerical-aperture (NA) and long-working-distance lenses [ ], increasing the light-coupling efficiency and resolution.

This technology has broad applications in X-ray micro-computed tomogra- phy (CT) imaging, materials science, and medical imaging.

When combined with synchrotron radiation sources, 3D mi- cro–nanostructural reconstruction and dynamic observations

can be performed. The Shanghai Synchrotron Radiation Facility (SSRF) was instrumental in this technological progress.

Its high- performance beamlines provide high-brightness, highly col- limated X-rays for various X-ray imaging methods [ ] and their applications [ ], particularly indirect micrometer/submicrometer-resolution X-ray imaging. Over the past decade, the hard X-ray imaging beamline at the SSRF has achieved breakthroughs in indirect imaging technolo- gies, with micro-CT serving as the core technique (supporting >70% of the experiments). By integrating absorption, phase contrast, and fluorescence imaging with rapid algorithms, high-precision 3D reconstructions can be realized [ Furthermore, multimodal approaches, such as fluorescence/- dynamic/diffraction CT, have been implemented, with dy- namic CT capturing real-time images of microscopic evolu- tion in living insects [ Deep learning is widely used in optical imaging. In re- cent years, it has also been introduced in direct-coupled indirect X-ray imaging fields to eliminate the influence of optical blurring under various conditions [ lens-coupled systems, we developed deep neural network- assisted high-spatial-frequency enhancement and reconstruc- tion (DA-HSFER), which is an information optics-based in- novation that addresses high-frequency information loss in X- ray microscopic imaging [ ]. This technique combines an optical encoder on the scintillator surface with a deep- learning decoding module that converts X-ray-induced high- frequency fluorescence signals into low-frequency signals transmitted through the scintillator-air interface. The deep- learning model reconstructs high-frequency details from the low-frequency data. Through encoder-decoder synergy, DA- HSFER recovers lost information, overcomes traditional physical limitations, and significantly enhances the imaging performance.

Specifically, the optical encoder employs specialized mi- cro/nanostructural patterning on the scintillator surface to modulate the visible light generated by X-ray excitation, en- coding high-frequency details that are typically lost during conventional imaging processes. Subsequently, deep-learning algorithms decode and reconstruct the encoded image data, leveraging large-scale training datasets and optimized neural network architectures (e.g., hybrid convolutional neural net- work (CNN)-transformer models) to accurately restore high- frequency information, which significantly enhances the im- age detail fidelity. This technology overcomes the limitations of traditional indirect X-ray imaging, where the high refrac- tive index of the scintillator restricts the bandwidth of the op- tical system and degrades the image resolution.

However, in practical applications and further research and development, we must prioritize resolving the fundamental optical bottleneck arising from the contradiction between the imaging depth and DOF of an optical system, which causes defocus blurring and surpasses the impact of high-order aber- rations and high-frequency information loss induced by in- ternal reflections in scintillators. The imaging depth refers to the longitudinal distribution of fluorescence patterns gen- erated by X-ray penetration within a scintillator, which is typically on the scale of hundreds of micrometers or more.

This often leads to blurred features in sample regions distant from the focal plane in high-NA optical systems designed for high resolution, thereby compromising the overall imaging quality. This challenge remains a critical unresolved issue in many scenarios, particularly in biomedical applications such as imaging thick tissue sections or 3D cell cultures, as well as in industrial inspections of multilayered electronic com- ponents, where defocus blurring significantly degrades the imaging accuracy and detection precision.

Experimental data show that as the CsI:Tl thickness in- creases from , the absorption rate of 200 keV X-rays increases from 38% to 92% [ ], while the modulation transfer function value at half maximum (MTF50) decreases from ]. This contrasting thickness-performance relationship stems from two factors. First, the light yield (LY) increases with the X- ray excitation depth in the scintillator. Second, the longitudi- nal distribution of fluorescence points caused by X-ray pen- etration broadens the full width at half maximum (FWHM) of the point spread function (PSF). However, although tradi- tional thinning methods (e.g., laser cutting to ) improve the spatial resolution, they also introduce high costs, instabil- ity, machining defects, and a low photon yield in thin scin- tillators, creating an "efficiency-quality-reliability" trilemma.

The present study overcame this optical bottleneck through the synergistic integration of optimized optical structural de- signs and novel image restoration methods, thereby advanc- ing the development and application of micron/submicron- scale indirect X-ray imaging technologies.

Wavefront coding has been employed for general optical imaging DOF issues [ ], where phase masks modulate the wavefronts to maintain consistent PSFs across the entire DOF, followed by PSF deconvolution for image recovery. However, the phase plate in this technology is wavelength-sensitive and introduces design, manufacturing, or computational complex- ities when applied to broad-spectrum light sources [ ]. Stud- ies on synchrotron-based indirect X-ray imaging [ ] have indicated that a pure-PSF method can be used to restore im- ages under parallel light conditions because the fluorescence point distributions at different depths in the scintillator are sufficiently uniform. Ref. [ ] employed a generalized PSF function from the simulation, which was overly idealized.

Our study theoretically extended this method to the non- parallel geometric magnification configurations commonly used in systems based on X-ray tubes and often employed in imaging stations at synchrotron radiation facilities, espe- cially those with large NAs. Crucially, prior simulations [ ignored some non-ideal factors, such as crystal lattice de- formation, which may lead to deviations in images. To ad- dress this problem, we developed a PSF-based image restora- tion method. Taking a Ce -doped yttrium aluminum gar- net (YAG) crystal as an example, we used a -thin YAG crystal film, with a depth close to the DOF of the optical sys- tem, as the control group. By deconvolving the -thick YAG imaging results with the control group, we extracted the actual PSF and restored the high-resolution image. In indi- rect X-ray imaging, the direct measurement of the PSF for

image restoration has long faced significant challenges be- cause of limitations in experimental setups and material prop- erties [ 36 – 38 ]. In this study, the challenges included severe defects that severely degraded the reliability of the measured PSF, along with low photon efficiency in the thin YAG, which required a long exposure time to induce sample drift. We uti- lized these defects as alignment markers and exploited the differences in defect correlations between thick and thin scin- tillators to eliminate the effects of the defects entirely. By combining this method with large-NA optics, a microfocus X-ray imaging system was developed, in which both high- photon-flux and high-definition imaging were achieved.

IMAGING PRINCIPLE An extended PSF-based restoration method for addressing depth-related challenges in fluorescent pattern imaging All of the studies were conducted using a microscope- coupled high-resolution indirect X-ray imaging system. A tungsten target X-ray source was employed and operated at an accelerating voltage of 50 kV , producing X-ray photons with energies predominantly in the range of 10 keV 30 keV shown in Fig.

The penetration depth of these photons in a yttrium alu- minum garnet (YAG) crystal varied with photon energy, rang- ing from . To ensure the maximum absorp- tion of X-ray photons, a -thick YAG:Ce scintillator was utilized for indirect X-ray imaging.

The results of an error analysis of the optical imaging sys- tem are shown in Fig. , where optical magnification numerical aperture = sin , absorption length ∆=100 10 cm 10 cm , and Then, we obtained . Therefore, image point shift owing to the X-ray geometric magnification, and the other shift because of the thickness of the scintil- lator. The edge parts of the image shifted by , or by approximately one pixel ( ), for a field of view (Fig. a). This error can be ignored in the present stage of research. In Fig. = sin , where is the imaging point size of the fluorescent object on the front surface of the scintillator. Here, represents the size of the defocused im- age spot of the fluorescent object at the absorption depth ( of the scintillator, = 2∆tan . When ∆=100 . Assuming is the size of one pixel, , then 13 . This means that the PSF function for the fluorescent points varies greatly with depth z in the scin- tillator. This variation is much greater than the variation in the overall light intensity distribution on the imaging surface (Fig. a). The image distribution could be viewed as having different depths in the scintillator in this study. (b) the point spread of the image was caused by the depth of the fluorescent object scene, which was at most 13 times.

Under the above estimation, fluorescent patterns at differ- ent depths along the optical axis ( z direction) exhibited ge- ometrically similar spatial distributions, and the finite thick- ness ( ∆ z ) of the scintillator led to multifocal superposition (defocusing) and considerable X-scattering from the z -planes downstream.

The imaging process was then modeled as shown in Eq. ( 1 ):

I i = � ∆ z

x, y, z where denotes the convolution, is the distance from the anterior surface of the scintillator to the X-ray source, is the depth-dependent intensity-weighting function, is the ideal image under parallel light, and is the geomet- ric magnification. The combines internal photon diffu- sion ( internal ) and optical system effects ( optical internal is caused by X-ray scattering, while optical is the result of the pattern at not being in the center of the When the field coverage angle for the X-ray source is much smaller than the angular acceptance of the optical system downstream, which is usually satisfied by a high-NA system, varies much less than PSF in the integral above. In our sys- tem, the two angles were approximately . Thus, we obtain Eq. (

I i ( x, y ) = I g ( αz 0 · x, αz 0 · y ) ∗ � ∆ z

x, y, z

= I g ( αz 0 · x, αz 0 · y ) ∗ PSF all ( x, y ) (2)

Therefore, a single PSF-based convolution is sufficient, even under geometric magnification in our system.

Imaging by a 20 µ m -depth YAG crystal was chosen as the standard for the experimental PSF measurement. By analyz- ing the modulation transfer function (MTF) for 20 µ m and 200 µ m scintillators, we derived Eq. ( 3 ).

F{ PSF 20 } = F{ PSF 200 } · F{ PSF 20 } F{ PSF 200 } (3)

Based on this analysis, we further defined the deconvolu- tion kernel, , using Eq. (

Applying PSF 20 − 200 to 200 µ m -thick YAG imaging re- stored the high-frequency details while retaining the high photon efficiency. As shown below, the defects in the thin and thick YAG layers were uncorrelated and could be suppressed via PSF truncation.

Optical System Design for High-Resolution Imaging with Thick Scintillator and High-NA Lens In the microscope there were two lenses with optical mag- nifications of 5 and 10 and NAs of 0.4 and 0.9, respec- tively. The camera pixel size was . As previously mentioned , we employed a thick scintillator to convert the X- ray patterns into visible light. A high-NA optical microscope lens was utilized to capture more photon flux and achieve high diffraction-limited resolution. The depth-of-field of the optical system was not particularly restricted, allowing for the reduction of various aberrations under the conditions of a high NA, thick scintillator, and suitable working distance.

To resolve depth-dependent fluorescence field distortions in the thick scintillator, a PSF-based restoration method was adopted using thin scintillator-derived high-resolution images as reference standards. Therefore, images with both a high photon efficiency and high definition were obtained.

X-rays can degrade optical lenses over time, thereby re- ducing their optical efficiency. A conventional solution in- volves installing a lead-glass filter at the front of the lens to protect subsequent optics. However, conventional high-NA microscope lenses, particularly those with , have short working distances (typically, approximately Although Nikon offers an lens with a working distance of up to , installing a lead glass filter directly would significantly affect axial aberrations, such as spheri- cal and chromatic aberrations, because of not considering the thickness of the filter during commercial lens design, thereby reducing the imaging resolution. Hence, there is a need to develop dedicated, long working distance, and high-NA mi- croscope lenses that can accommodate lead glass.

To meet these requirements, we designed and developed two long working distance microscope lenses with 5 magnifications, which had a conjugate distance of 500 mm and NA values of 0.4 and 0.9, respectively. Both lenses exceeded the 40% MTF values at 384 lp 769 lp , corresponding to a camera pixel size of A replaceable -thick ZF7 lead-glass filter with a lead equivalence of 66 mmpb 33 mmpb per millimeter of ZF7 glass) was installed between the lens and scintillator.

In addition, the effects of different scintillator thicknesses on high-NA lens aberrations cannot be overlooked. There- fore, during the lens design, aberration optimization was per- formed for a commonly used -thick scintillator. Simu- lations showed that with an MTF decrease to 30% as the crite- rion, the 5 lens could accommodate scintillator thicknesses ranging from 0 to without adjusting the back-focal distance, whereas the 10 lens could only accommodate 45 . With an adjustable back-focal distance, the range for the 5 lens was extended to 0 to , and that for the lens was extended to 0 to . Figures a and show the optical layouts of the two lenses and their corre- sponding MTF curves.

To further protect the CCD/sCOMS camera, a mirror was added between the lens and camera to redirect the overall op- tical path by 90°. These designs ensured a high photon collec- tion efficiency while effectively blocking X-rays from damag- ing the optical components and CCD camera, thereby extend- ing the lifespan of the equipment. The high NA enabled the system to theoretically achieve pixel-level resolution, but at the cost of a shorter DOF.

As shown in Fig. e, for a 5 lens at a spatial fre- quency of 384 lp (corresponding to two imaging pixels, ), the contrast decreased to zero when the defocus distance was (Fig. e), which provided a DOF of approximately . Similarly, the DOF for the 10 tem was approximately for a spatial frequency of 384 lp (Fig.

The thinnest scintillator foil obtained had a thickness of . The fluorescence images within the top had an actual resolution exceeding the pixel resolution of the camera 384 lp , 2-pixel criterion), whereas in the deeper region, it fell below this threshold. Images on the thick YAG foils were employed as standard images for the PSF-based recovery method. Using the high-NA lenses, a new recov- ery method was investigated in this study that overcame the DOF and scintillator thickness contradiction to achieve high- resolution and high-definition imaging.

The optical DOF, X-rays, and electron scattering may all degrade the imaging resolution. In a YAG crystal, electron scattering has a mean free path of 100 nm for secondary electrons, which is too small to induce significant image blur.

Therefore, the degradation of the image resolution is mainly caused by optical defocusing and X-ray scattering. Under the X-ray photon energy range and DOF of the optical system

The optical layouts of the 5 and 10 lenses, respectively; (c,d) MTFs for 5 and 10 lenses, respectively. MTF simulation re- sults for lenses with defocus: (e) 5 lens with defocus and (f) 10 lens with defocus. The colored lines indicate the contrast of line pairs (lp) at different positions on the image plane, while the black line represents the value corresponding to the optical diffraction limit. Here, “T” and “S” represent tangential and sagit- tal, respectively; and 0000 mm 6500 mm , and 4000 mm the distances between the test positions on the lens and its center. employed in this study, X-ray scattering was not the primary factor affecting the resolution (as discussed later).

EXPERIMENTS, RESULTS, AND ANALYSIS Experiments X-ray indirect imaging system.

The experimental setup is shown in Fig. 4 [FIGURE:4] . It consisted of an X-ray-tube-based source, a scintillator conversion layer, and a high-NA visible-light microscope.

For high-definition images, a microfocus X-ray source

(HamamatsuL10101, tungsten target) was employed. It had an adjustable accelerating voltage range of 40 kV 100 kV with a maximum output power of

20 W

. At an output power

4 W

, the source spot size was . The X-ray source was operated at an accelerating voltage of 50 kV to achieve a small source spot size. In future studies, higher accelerating voltages will be utilized for imaging with thick scintillators to achieve a higher photon flux. Using a 5 lens and an sC- MOS camera with a pixel size of , a pixel resolution was obtained. YAG:Ce crystal films were used to convert X-rays into visible light with high luminous effi- ciency in the 600 nm wavelength range.

A fixed visible-light imaging system was used in the exper- iment, and the scintillator plate position was finely adjusted for focusing. To accurately determine the optimal focus posi- tions for the two YAG scintillators with different thicknesses, we used a displacement stage with a step size of to scan the samples axially. This process required precise mechani- cal adjustments and repeated verification of the focal position to ensure the optimal imaging quality. After locating the best focal plane, each sample and its corresponding background were imaged ten times to enhance the data reliability and sta- bility. LabVIEW software controlled the CCD for synchro- nized data acquisition and storage. The raw data were then imported into MATLAB for averaging, to reduce the noise and highlight the true features of the samples. Background subtraction was also performed to eliminate interference from YAG surface defects and system noise, thus clarifying the details of the samples.

These rigorous experimental steps and data processing methods ensured the acquisition of high- quality image data for subsequent analyses.

The image on the -thick YAG scintillator film had a high resolution but weak signal, requiring an exposure time , whereas was needed for the -thick YAG scintillator film. Accurate alignment of the sample and back- ground images was crucial to avoid information bias and en- sure a reliable analysis.

As shown in Figs. a and b, the scale-invariant feature transform (SIFT) algorithm was used, which is known for its noise resistance in image feature matching. SIFT is widely used for image registration because of its stability and relia- bility. It identifies feature points by recognizing distinct im- age features. In this study, the unique contours, positions, and higher intensities of the defect features rendered them suit- able reference points for registration. Weaker image infor- mation was treated as “noise” and ignored. Image alignment was performed using SIFT feature point matching, and some obviously incorrect matches were manually excluded, which eventually eliminated the influence of image drift.

Background subtraction, which involves subtracting a background image from a sample image, is feasible because of the rich frequency components of the resolution target im- age. These components help achieve high contrast and sharp line edges, allowing the background-subtracted image to re- tain more detail. As shown in Figs. c and d, when compar- ing the images of the -thick YAG film with those of the -thick YAG sample, the former still show significant defect impacts. This indicates that while background subtrac-

tion can effectively enhance the image quality, the inherent defects in thinner YAG scintillator films may still interfere with the imaging results. These defects likely stem from the material properties or unavoidable factors in the fabrication process. -depth YAG and (b) its background images, and resolution target images (c) on -depth (d) -depth YAG samples.

Data Analysis As shown in Figs. a and was obtained by deconvolving the resolution target image (Fig. d) on -thick YAG with that (Fig. c) on -thick YAG.

The Lucy-Richardson algorithm was employed, and the num- ber of iterations was limited to eight. The entire structure, Gaussian-like peaks, and some adjacent structures can be found in the central region. PSF truncation was employed to remove the influence of defects on a -thick YAG plate used as a standard sample (Fig. c), but these were not com- pletely removed during background removal. The imaging process in the frequency domain is described in Eqs. ( ) and

F 20 = F · M 20 + G 20 (5)

F 200 = F · M 200 + G 200 (6)

Here, , and represent ideal imaging, imaging on -thick scintilla- tors, and the corresponding modulation transfer functions and point-like defects on -thick scintillators, respectively. is much smaller than and thus can be ignored. The measured MTF was obtained using Eq. (

M = F 200 F 20 ≈ M 200 M 20 · (1 + G 20 F · M 20 ) ≈ M 200 M 20 · (1 − G 20 F · M 20 )

In the spatial domain, the second item, , corresponds , where is the pseudo-impulse response con- tributed by the point-like defects attached to the real-image PSF. This additional has a small total power, wide dis- tribution, and small amplitude in the central region, which is relatively large in the outer range. Therefore, in princi- ple, PSF truncation can remove the influence of defects in a standard image (on a -thick YAG plate). Further- more, the blurred image spot on the -thick YAG was approximately 13 pixels; thus, the truncation boundary had to be significantly larger than 13 pixels. However, to reduce the frequency-domain pollution caused by PSF truncation, the truncation boundary needed to be far from the center. There- fore, a central region (60 pixels 60 pixels) containing almost all these characteristics was finally selected. . (a) The en- tire area, (b) central area, (c) image obtained by deconvolving the resolution target image on -thick YAG (Fig. d) with the truncated PSF, and (d) image from Fig.

Compared to the image (Fig. d) on the -thick YAG film, the image (Fig. c) deconvolved from the resolution target image on the -thick YAG using the above- mentioned truncated PSF was much better. The details were clearer and sharper; the edges were more distinct; the overall image quality was significantly improved; and the influence of defects was fully eliminated. This showed that the trun- cated PSF method used in this study is effective in removing the influence of these sparsely distributed sharp defects, while keeping the image details intact.

Figs. a and b present the deconvolution results from ze- brafish images captured using a YAG indirect imag- ing system utilizing the extracted central PSF region. Fig. shows the original zebrafish image after the background sub- traction. The basic outline and shape of the zebrafish are dis- cernible, but the details are obscured by background noise,

resulting in a blurry image. Fig. b) presents the deconvolved image. Visually, the processed images showed remarkable improvements, with enhanced clarity and sharper details for the scales and fins of the zebrafish. This indicated that decon- volution effectively reduced blurring and boosted the image contrast and resolution. (Coloronline) Zebrafish imaging: three 65 mm 65 mm areas showing details from the original image on -thick YAG, (b) the corresponding deconvolved results, (c) 2D PSD of the raw image, (d) 2D PSD of the decon- volved image, and (e) 1D PSD comparison of raw and deconvolved results.

It is important to note that while the YAG substrate has inherent defects, such as surface impurities, processing flaws, and minor crystal structure inconsistencies, these de- fects are less impactful. The greater imaging depth and pho- ton count associated with thicker scintillators indicate that these defects do not significantly interfere with the key imag- ing features. By contrast, defects such as surface scratches and internal bubbles in the YAG control group are theoretically more detrimental. However, after PSF function truncation, their impact was effectively controlled and high- frequency information enhancement remained largely unaf- fected.

For specimens with complex details, low intrinsic contrast, and weak visible light signals (e.g., zebrafish specimens), division-based background subtraction was used. Unlike the traditional subtraction methods, this approach removes the background by dividing the image pixel values by the esti- mated background signal values. In the zebrafish images, this method preserved crucial details such as fine surface textures and weak internal fluorescence signals, which might have been lost with traditional subtraction. This prevented issues such as excessive contrast and detail loss during background correction, provided richer and more accurate visual informa- tion for subsequent image analysis, and facilitated a deeper understanding of the specimen structures and properties.

PSD analysis (Figs. c and d) clearly showed that decon- volution significantly boosted high-frequency detail recovery.

The 1DPSD (Fig. e) indicated a significant SNR improve- ment in the high-frequency range (up to 8.45 times), with no low-frequency information loss. This confirmed the effective- ness of the deconvolution method based on a generalized PSF function for improving image quality. However, the SNR in the high-frequency region did not improve as expected. This was likely because the -thick ultrathin scintillator used exceeded the imaging lens DOF ( ). Consequently, thin- ner scintillators will be used in future experiments to enhance the SNR in the highest frequency band and further optimize the image quality.

In comparison, when imaging zebrafish samples using a -thick YAG crystal, the high X-ray absorption rate of the samples combined with the low transmitted beam inten- sity from the 4W X-ray source resulted in severely degraded image contrast. This made it impossible to determine the op- timal focal position by adjusting the focus settings based on the observed image quality.

To verify the fidelity advantage of this method, based on the real PSF measured, the perception based image quality evaluator (PIQE) method was adopted to determine the image fidelity. A lower score for the PIQE method implied better fidelity. The results were as follows.

For the image of the pair of resolution targets on the -thick YAG, the PIQE value was 63.2. Thus, the PIQE value of 48.9 for the image after deconvolution showed an ob- vious improvement. The value for the -thick YAG plate was 75.4 because of serious defects. When imaging the fish samples with the YAG, the PIQE value was 14.0, and the PIQE value after deconvolution was 9.13, which proved that the image fidelity had been significantly improved.

In conclusion, this approach achieved a high resolution and high SNR with -thick YAG. It is worth noting that this method effectively circumvents the conventional "efficiency- quality" trade-off. In other words, an improvement in resolu- tion does not result in a significant decrease in detection effi- ciency, which is a common challenge in the field. The com- putational overhead of our deconvolution algorithm is signif- icantly lower than that of the hardware upgrade. Thus, it is a cost-effective solution for maximizing the performance of ex- isting detectors. By using computational methods to enhance imaging quality, we can save substantial resources that might have been spent on upgrading hardware. In future work, we will use thinner scintillator films, such as GGG:Tb films with a depth of ], as control samples to further explore the performance of this method in 10 magnification sys- tems.

This will help us to better understand the potential and limitations of our approach in different experimental and imaging scenarios.

DISCUSSION

Effect of X-ray scattering on imaging X-ray and electron scattering, and the DOF of optical am- plification systems, may degrade the imaging resolution. X- ray scattering includes Rayleigh and Compton scattering. In YAG X-rays, Rayleigh scattering occurs within a small an- gle range, which reduces the imaging resolution. Compton scattering scatters X-ray photons with a large angular distri- bution and forms a background, which does not directly affect the image resolution. In the range of 10 keV 50 keV , the

Rayleigh scattering probability is very small, and Compton scattering accounts for a considerable proportion. The spec- trum of the X-ray source used in the experiment contained a set of characteristic lines near 10 keV , as well as a broad con- tinuum of background radiation that extended mainly from 10 keV 30 keV . Thus, X-ray scattering could be ignored.

As previously mentioned, electron scattering induces an ex- tremely small image blur with a size of 100 nm . Therefore, for a large-NA optical lens, the degradation of the fluorescent image resolution is mainly caused by optical defocusing, and the scattering mentioned above can be ignored to a certain degree.

Material Adaptability and Scintillator Selection Based on these research findings, relaxing the defect tol- erance requirements facilitates the adoption of ultrathin scin- tillators (e.g., -thick GGG:Tb in Ref. [ ]), although the material diversity remains constrained. To address this, we propose a hybrid strategy: standard imaging may em- ploy process-compatible ultrathin conventional scintillators, whereas practical imaging can utilize scintillators with a high photon efficiency. The material differences lie only in the X-ray absorption rate, luminescence efficiency, and refrac- tive index. The influence of the X-ray absorption rate and luminescence efficiency of different materials is described by W.B. Ma, C.F. Kuang, X. Liu et al., Research progress of X- ray detection and imaging based on emerging metal halide semiconductors and scintillators. Acta Opt. Sin. , 1704002 (2022).

Y.R. Li, G.Q. Zha, D.K. Wei et al., Effect of deep-level de- fects on the performance of CdZnTe photon counting detectors.

Sens. , 2032 (2020). G.Q. Zha, T. Wang, Y.D. Xu et al., The development of CZT semiconductor X-ray and g-ray detectors. Phys. , 862–869 (2013).

K. Tabata, R. Ohtake, T. Aoki, High-spatial-resolution X-ray imaging by scintillator in silicon collimator. Sens. Mater. 4037–4045 (2020).

A. Maier, S. Steidl, V. Christlein et al. (eds.), Medical Imaging Systems:

An Introductory Guide (SpringerCham, 2018). Nikl, Yoshikawa, Kamada Devel- opment LuAG-based scintillator crystals view. Prog. Cryst. Growth Charact. 47–72 (2013).

Y. Zorenko, V. Gorbenko, T. Voznyak et al., LuAG:Pr, LuAG:La, and LuAP:Ce thin film scintillators for visualisation of X-ray images. Proc. SPIEInt. Soc. Opt. Eng. , 731007 (2009).

M. Kobayashi, J. Komori, K. Shimidzu et al., Development of vertically aligned ZnO-nanowiresscintillators for high spa- tial resolution X-ray imaging. Appl. Phys. Lett. , 081909 (2015). ically grown ZnO vertical nanowirescintillator with light- , and the entire process can be described by a generalized PSF. The refractive index must be considered for optical path changes such that the spherical aberration remains within a controllable range. The entire process can then be performed directly, regardless of material differences.

CONCLUSION

This study proved theoretically and experimentally that an image-recovery method using a single PSF can be applied to a high-resolution X-ray scintillation imaging system with a high-NA lens. By integrating a measured-PSF-based decon- volution method with a high-NA optics design, we developed a high-resolution, high-SNR, indirect X-ray imager using an X-ray-tube-based system. The image details suppressed by the thick scintillator could be restored using this method.

Combining the above two new technologies and an imaging system with a large NA lens and thick scintillator plate, both high resolution and a high SNR were simultaneously achieved in the X-ray tube-based system. This study showed that our method does not require an ultrathin scintillator with a per- fect surface for standard imaging, which greatly reduces the difficulty of the acquisition process. These results provide a critical pathway for overcoming the long-standing resolution- efficiency dilemma in indirect X-ray imaging based on X-ray tube sources or synchrotron radiation facilities.

guiding effect. Phys. Status Solidi A , 1700285 (2017).

W. Li, D.Y. Xin, S.J. Tie et al., Zero-dimensional lead- free FA single crystals for high-performance X-ray detection. J. Phys. Chem. Lett. 1778–1785 (2021).

C. Li, S. Zhou, J. Nie et al., Durable flexible polymer- encapsulated sensitivity X-ray detection.

Lett. (2021). Y. Teng, M. Gu, Z.X. Sun et al., Optimization of SiO reflective layer thickness for improving the performance of structured CsI scintillation screen based on oxidized Si micropore ar- ray template in X-ray imaging. Opt. Express , 24097–24109 (2023).

C.S. Choi, K.D. Song, S.S. Kang et al., Optical imaging featur- ing both long working distance and high spatial resolution by correcting the aberration of a large aperture lens. Sci. Rep. 9165 (2018).

Z.J. Qiu, K. Li, H.L. Xie et al., Study of 20 Hz high spatial- temporal resolution monochromatic X-ray dynamic micro-CT.

Nuclear Techniques , 070101 (2023). C.P. Wang, X.Y. Li, R.Z. Wan et al., Accelerating imag- ing research at large-scale scientific facilities through scien- tific computing. J. Synchrotron Radiat. , 1317–1326 (2024).

Y.D. Wang, Y.Q. Ren, G.Z. Zhou et al., Equally sloped tomog- raphy based X-ray full-field nano-CT at Shanghai Synchrotron Radiation Facility. Nucl. Instrum. Methods Phys. Res. A

108–112 (2018). D.P. Yin, Y.B. Liu, Y. Zhang et al., Synchrotron-based X-ray molecular probes for imaging in intelligent biomedicine. Small

Methods

, 2401890 (2025). L. Zhang, F. Tao, J. Wang et al., The 3Dnanoimagingbeamline at SSRF. Nucl. Sci. Tech. , 201 (2023).

R.Y. Gao, L. Zhang, F. Tao et al., Transmission X-ray microscopy-based three-dimensional XANES imaging. Ana- lyst. , 4506–4513 (2024).

C.H. Fan, Z.T. Zhao (eds.), Synchrotron Radiation in Materials Science: Light Sources, Techniques, and Applications (Wiley- VCH, Weinheim, 2018).

Y. Zhu, J.C. Zhang, A.G. Li et al., Synchrotron-based X-ray microscopy sub-100 resolution imaging. Curr. Opin. Chem. Biol. 11–16 (2017).

H.L. Xie, B. Deng, G.H. Du et al., Methodology development and application of X-ray imaging beamline at SSRF. Nucl. Sci.

Tech. , 102 (2020). R.C. Chen, H.L. Xie, B. Deng, G.H. Du et al., X-ray micro- tomography at Shanghai Synchrotron Radiation Facility. Proc. , 99671B (2016).

H.L. Xie, B. Deng, G.H. Du et al., Development of X-ray imaging methodology and its applications on material science at Shanghai Synchrotron Radiation Facility. Failure Analy- sis and Prevention , 46–59 (2021).

K.A. Mohan, R.M. Panas, J.A. Cuadra, SABER: a systems approach to blur estimation and reduction in X-ray imag- ing. IEEE Trans. Image Process. , 7751–7764 (2020).

M.C. Chiu, C.J. Wei, Integrating DeblurGAN and CNN to improve the accuracy of motion blur X-Ray image classification. J. Nucl. Sci. Technol. , 403–416 (2024).

H.J. Xia, Y.Q. Wu, L. Zhang et al., Great enhancement of image details with high fidelity in a scintillatorimager using an optical coding method. Photon. Res. , 1079–1085 (2020).

H. Shi, Y.Q. Wu, L. Wang et al., DA-HSFER: empower- ing high-performance incoherent X-ray scintillation encoded imaging with deep neural networks. ACSPhotonics. , 3652– 3661 (2024).

Z.J. Li, H. Shi, B.N. Li et al., Fabrication of large-area photonic crystal-modified X-ray scintillatorimager for opti- cal coding imaging. Opt. Express. , 8877–8886 (2024).

H. Shi, Y.H. Sun, Z.F. Liang et al., Feature-enhanced X- ray imaging using fused neural network strategy with des- ignable metasurface. Nanophotonics , 3793–3805 (2023).

B. Wei, M. Zhou, P. Feng et al., Study on CsI(Tl) scin- tillating crystal for X-ray high-resolution detection with montecarlo method. Acta Opt. Sin. , 1429–1434 (2006).

V.V. Nagarkar, T.K. Gupta, S. Miller et al., Structured CsI(Tl) scintillators for X-ray imaging applications. IEEE Trans. Nucl. , 492–496 (1998).

E.R. Dowski, W.T. Cathey, Extended depth of field through wave-front coding.

Appl. (1995). Z.L. Cao, C.J. Zhai, J.H. Li et al., Combination of color coding and wavefront coding for extended depth of field. Opt. Com- , 252–257 (2017).

Y.P. Wang, G. Li, J. Zhang et al., Improving the detection ef- ficiency and modulation transfer function of lens-coupled indi- rect X-ray imaging detectors. J. Synchrotron Radiat. , 1093– 1105 (2018).

P. Janout, P. Páta, P. Skala et al., PSF estimation of space- variant ultra-wide field of view imaging systems. Appl. Sci. , 151 (2017).

J. Yang, Z.J. Zhang, Q.M. Cheng. Resolution enhancement in micro-XRF using image restoration techniques. J. Anal. At.

Spectrom. , 750–758 (2022). J.M. Shao, H.N. Lu, H. C, A point spread function model for X-Ray imaging. Acta Opt. Sin. , 1148 (2005).