Introduction
X-rays are electromagnetic radiation characterized by high energy and penetration. When X-rays penetrate an object, different materials absorb them to varying degrees. Materials with high density (such as bones) exhibit stronger X-ray absorption than those with low density (such as soft tissue) [1]. X-ray detectors receive the X-rays that penetrate an object and convert them into electrical signals. These collected electrical signals are processed and reconstructed to generate visual X-ray images. The two main types of X-ray detectors are indirect and direct conversion detectors. The former typically uses scintillator materials to convert X-ray signals into photon signals [2, 3]. These photons are then converted into charge signals by a photodetector (such as a photomultiplier tube or photodiode). The latter may experience photon diffusion during the transmission process, resulting in a relatively low spatial resolution and potential challenges in capturing fine structures and details [4]. Direct conversion detectors typically use materials with high atomic numbers, such as silicon or selenium. When X-rays penetrate such a detector, their energy is absorbed, creating electron–hole pairs, which are collected by an electric field and converted into charge signals. Direct conversion detectors offer a higher spatial resolution than indirect conversion detectors [5].
To achieve high detection efficiency and spatial resolution, many semiconductor materials such as amorphous silicon/selenium, cadmium zinc telluride, and perovskites have been used in direct conversion X-ray detectors. In 1993, the first amorphous silicon flat panel detector with a PIN structure was proposed [6]. Subsequently, amorphous selenium X-ray detectors using inorganic materials have demonstrated high X-ray sensitivity and spatial resolution, leading to widespread commercial applications [7-9]. However, amorphous selenium detectors are sensitive to temperature changes, and rapid fluctuations (≥ 5 °C h) can result in crystallization, leading to degraded detector performance or even detachment of the selenium layer or film from the detector surface. Furthermore, a high electric field bias can cause irreparable damage to the detectors [10, 11]. Cadmium telluride zinc (CZT) crystals, which have a large atomic number, high resistivity, and a wide bandgap, are among the most promising room-temperature semiconductor detector materials. Compared with traditional material detectors, they offer advantages such as high energy resolution, small size, and portability. However, CZT X-ray detectors face challenges in crystal growth and the fabrication of large-area detectors. Furthermore, nonuniform crystal growth results in a nonuniform detector response, directly affecting imaging quality [12]. In recent years, perovskite materials have gained interest as a promising choice for the next generation of direct conversion semiconductor materials in X-ray detectors owing to their high X-ray attenuation coefficient [13-16], long carrier diffusion distance, and radiation stability. However, issues such as dark current and noise caused by ion migration in perovskite materials severely limit the detection limits and operational stability of the devices [17, 18]. Moreover, perovskite materials are susceptible to high-dose radiation damage, leading to decreased sensitivity and performance of the detectors [19, 20]. The fabrication process for perovskite detectors is still under development, and the achievement of large-area high-resolution applications is challenging. In addition, the relatively high cost of perovskite materials may limit their widespread use in certain fields. The complex preparation process of perovskite materials can result in significant performance variations between different detectors, posing challenges for calibration and interchangeability among multiple devices. Regardless of the semiconductor material used in X-ray detectors, none of them are perfect. However, as long as the corresponding application scenarios are identified, further research is worthwhile.
In this study, FC40 electronic fluorinated liquid was employed as a new alternative to address the limitations of traditional semiconductor materials in direct conversion X-ray detectors. These limitations include susceptibility to temperature effects, delamination, nonuniform lattice growth, and vulnerability to radiation damage. FC40 electronic fluorinated liquid is a colorless and transparent perfluorinated chemical with excellent chemical inertness, electrical insulation properties, thermal conductivity, and an extremely low dark current. It is commonly used as a thermal conductive medium in semiconductor manufacturing facilities. Owing to its liquid form, the use of the FC40 fluorinated liquid in the fabrication of X-ray detectors addresses issues such as delamination and radiation damage. Furthermore, the X-ray absorption thickness can be customized to suit specific requirements [21-23]. This novel approach provides possibilities for improving the performance and reliability of direct conversion X-ray detectors by addressing the limitations of traditional semiconductor materials.
Materials and methods
Design of the detector
Figure 1(a) shows the prototype of a direct conversion X-ray detector using FC40 electronic fluorinated liquid, comprising the detector and readout system. Figure 1(b) shows the schematic of the direct conversion X-ray detector using FC40 electronic fluorinated liquid, primarily comprising the Topmetal II- chip [24, 25], an FC40 electronic fluorinated liquid device, and a beryllium layer. The top layer of the Topmetal II- chip consists of exposed electrodes, directly in contact with the FC40 electronic fluorinated liquid. The FC40 electronic fluorinated liquid device is fixed to the top metal layer of Topmetal II- using an epoxy resin, creating a cavity of 2 mm (the height can be customized). The top of the cavity is sealed with a 200 μM thick beryllium layer using epoxy resin, enabling a convenient bias connection. The chip bonding board supplies power, drives the chip, and provides interfaces for connection with the readout board. When X-rays penetrate an object and reach the FC40 electronic fluorinated liquid through the beryllium layer, a reaction occurs. Under the influence of a negative bias voltage, an electric field is generated, causing negative ions to be collected by the Topmetal II pixel array within the electric field. The collected charges are then output through the readout circuit through a polling process.
Topmetal chip
The Topmetal II- chip is a pixel array chip, which collects spatial charges directly using its top metal layer. Each pixel has an exposed metal electrode at the top, enabling the collection of spatial charges generated by the ionization of charged particles. Under the influence of an electric field, these charges drift to the surface of the chip and are collected by the metal electrodes on each pixel, forming a two-dimensional distribution of spatial charges. As shown in Fig. 2, the Topmetal II- chip consists of a 72× 72 pixel array, with a total area of 8 mm × 9 mm. The area of the charge-sensitive region is 6 mm × 6 mm. Each pixel has dimensions of 83 μm × 83 μm, and the size of the top metal electrode exposed to air owing to the opening is 25 μm × 25 μm. The design noise target for the Topmetal II- chip is 30 e-. Figure 3(a) shows the overall structure of a single pixel; the Topmetal II- chip offers two signal readout modes: analog and digital readout modes. Because these readout modes are independent, the chip can operate in either an analog or a digital readout mode or have all pixel analog readouts connected to a single analog output buffer. In an analog read-out mode, scanning the entire pixel array enables the voltage signals from the charge-sensitive amplifiers of each pixel to be read out by the analog output buffer. Alternatively, the signal from any single pixel can be selectively output. The principle of chip simulation readout is shown in Fig. 3(b). The digital readout is stored at the bottom of each column and then output through column polling.
ScanModule controls the selection of rows and columns and outputs only the signal from one pixel to the output buffer during a specific time. Each pixel has Ms1, Ms2, and row select switches (ROW). Each column shares a column select switch (COL). The entire pixel array shares one output buffer, Mr1, and Mr2. The current of the first stage of the source follower, Ms1 and Ms2, is provided to all pixels. However, the second stage of the source follower is used for pixel selection in scanning mode. Because of the limited driving force of charge-sensitive amplifiers within pixels, they can be easily influenced by the row and column switches. The first-stage source follower is necessary to isolate the row/column selection switches from the charge-sensitive amplifiers [26].
FC40 electronic fluoride liquid
In this study, we attempted to explore the use of electronic fluorinated liquid FC40 as an alternative material for direct conversion X-ray detectors to address the limitations of traditional semiconductor materials. FC40 electronic fluorinated liquid is a colorless and transparent perfluorinated chemical with several advantageous properties including excellent chemical inertness, electrical insulation performance, thermal conductivity, and extremely low dark current. It is commonly used as a thermal conductive medium in semiconductor manufacturing facilities. The temperature range of this liquid extends from -57 ℃ to 155 ℃, and it has wide applications in etching equipment, ion implantation devices, testing equipment, and other devices. The electrical resistivity of FC40 electronic fluorinated liquid has been determined [27], revealing the IV curve characteristics of the material. A polyethylene container with inner dimensions of 1 cm × 1 cm × 1 cm was used. Copper plates were placed at both ends of the container to serve as electrodes. The container, along with the electronically fluorinated liquid FC40, was placed inside a lead box and exposed to X-ray radiation. The X-ray focus was positioned at a distance of 42.0 mm from the container with X-ray tube settings of 50kVp voltage and 100 μA tube current. The objective was to measure the resistivity of FC40 electronic fluorinated liquid under X-ray irradiation.
Readout system of detector
The back-end signal readout circuit comprises a complementary metal-oxide-semiconductor (CMOS) chip bonding board, mixed-signal conversion board, and control board using a field-programmable gate array (FPGA) [28]. The bonding board houses the bare die of the CMOS chip. Filter capacitors are placed on the bonding board to suppress noise. The bonding board is connected to a mixed-signal conversion board using pin headers. The single-ended analog signals output from the CMOS chip are input to the mixed-signal conversion board and then converted into differential signals to match the input of the analog-to-digital converter (ADS5282, TI, USA). ADS5282 features a 12-bit digital resolution, sampling rate of 65 Msps, and serialized low-voltage differential signaling output interface. A 16-bit digital-to-analog converter (DAC8568, TI, USA) on a mixed-signal conversion board is used to configure the bias voltage of the CMOS chip. The mixed-signal conversion board is connected to the FPGA-based control board using an FPGA mezzanine card connector. The control board, which uses a Xilinx Kintex-7 FPGA, serves as a scalable and minimal hardware core logic controller, including Ethernet transceiver modules, two double-data rates, three synchronous dynamic random-access memory modules (DDR3 SDRAM), and flash configuration modules. The FPGA-based control board receives serialized data output from the ADS5282 and transmits the data to a computer via Ethernet [29-31].
Digital radiography (DR) imaging platform
The self-developed X-ray DR imaging platform is shown in Fig. 4. It incorporates two types of X-ray sources: a medical-grade X-ray source (Varex G-242, Varex Imaging Corporation, UT, USA) and a microfocus X-ray source (L9181-02, HAMAMATSU, Japan). Moreover, it includes a horizontal linear stage (model: LS12-X200, Hanjiang, China) with a travel of 200.0 mm and a vertical linear stage (model: STS06-X20, Hanjiang, China) with a travel of 40.0 mm. The medical-grade X-ray tube G-242 has a maximum tube voltage of 125kVp and is equipped with a beam collimator to modulate the beam shape using a fixed-beam filter of 1.5 mm aluminum [32]. Spatial resolution testing, modulation transfer function (MTF) evaluations, and PCB DR images were obtained using the medical-grade G-242 X-ray source, and noise power spectrum (NPS) and cockroach DR images were obtained using the micro-focus X-ray source. The detector enabled continuous scanning along the vertical and horizontal directions on the linear stages, enabling rapid imaging of samples larger than 5 cm × 10 cm. Specific experimental parameter settings for each test group are presented in the Results section. Multiple X-ray projection images obtained from the CMOS were post-processed and stitched together to form complete DR projections of objects. In particular, signals corresponding to the same spatial positions were combined, whereas those corresponding to different spatial positions were systematically rearranged [33, 34]. For comparison, MTF and NPS measurements were conducted using a commercial scintillator (CsI:TI)-based indirect conversion CMOS flat-panel detector (model: Dexela 2329 NDT, PerkinElmer, USA). Notably, the experiment did not adhere to IEC standards because of the prototype nature of the CMOS direct conversion X-ray detector coupled with a fluorinated liquid, which had a small size (72 × 72 pixels), preventing compliance with IEC measurement protocols and imposing restrictions on distance measurements. The experiments were conducted under identical conditions to ensure comparable results.
Experimental results
Electronic properties of FC40 electronic fluorinated liquid
To understand the electronic characteristics of FC40 electronic fluorinated liquid, the IV (current-voltage) curve was measured using the test setup shown in Fig. 5(a). The FC40 electronic fluorinated liquid exhibited an ideal linear current response under X-ray radiation. With the X-ray tube current maintained at 100 μA, the tube voltage was scanned at bias voltages of 100 V, 150 V, and 200 V. As shown in Fig. 5(b), the current response of the FC40 electronic fluorinated liquid was susceptible to high tube voltages.
Spatial resolution
The spatial resolution was obtained by acquiring standard striped pattern DR images of the line-pair card (model: Type 18-d, QUART GmbH, Germany) through continuous exposure to a medical-grade X-ray tube (G-242, Varex Imaging Corporation, USA). The X-ray tube was operated at 50kVp with a tube current of 12.5 mA. The distance from the X-ray focal spot to the center of the line pair card was 474.0 mm and that from the X-ray focal spot to the photosensitive layer of the CMOS detector was 476.0 mm. While keeping the line-pair card stationary, the detector moved at a speed of 3.0 mm/s in the vertical direction. The image of the line-pair card obtained by stitching is shown in Fig. 6(a), demonstrating the capability of the fluorinated liquid detector to resolve structures with spatial frequencies greater than 4.0 lp/mm.
MTF
The MTF measurement was performed using a medical-grade G-242 X-ray source operating at a tube voltage of 50kVp and tube current of 12.5 mA. A 1.0 mm tungsten plate was tilted at 3 ° and secured onto a linear horizontal stage. The distance from the X-ray focal point to the tungsten plate was 638 mm and that from the tungsten plate to the surface of the CMOS detector was 2.0 mm. The tungsten plate remained stationary while the detector moved horizontally and vertically at a speed of 1.0 mm/s, capturing 5000 frames per scan and conducting a total of 10 scans. The experiment employed the tilted edge method to measure the edge spread function (ESF) for fine sampling. Subsequently, the ESF was differentiated to obtain the line spread function (LSF). The obtained LSF underwent a fast Fourier transform, and its value at zero spatial frequency was normalized to compute the MTF [35, 36]. Under the same experimental conditions, compared with the commercial indirect conversion Dexela 2329 NDT detector, the FC40 electronic fluorinated liquid detector exhibited superior MTF at low frequencies, as shown in Fig. 6(b). This demonstrates an advantage of the FC40 electronic fluorinated liquid material in direct conversion applications. The parameters of the FC40 electronic fluorinated liquid detector and the Dexela 2329 detector are presented in Table 1.
| Design feature | FC40 detector | Dexela 2923N detector |
|---|---|---|
| Pixel size (μM) | 83 | 74.8 |
| Array size | 72 × 72 | 3888 × 3072 |
| Sensitive area (mm2) | 6 × 6 | 290.8 × 229.8 |
| Material thickness (μM) | 2000 | 500 |
| Detector type | Direct conversion | Indirect conversion |
| Material | FC40 | CsI |
NPS
NPS measurements were conducted using a microfocus X-ray source (L9181-02, HANAMATSU, Japan) with a tube voltage of 50kVp and tube current of 0.1 mA. A PCB was fixed on a horizontal linear stage, with the distance from the X-ray focal point to the surface of the CMOS detector at 159 mm and that from the PCB to the CMOS detector at 82 mm. While maintaining the PCB stationary, the detector moved at a speed of 2.0 mm/s in both the vertical and horizontal directions, collecting 8000 frames per scan. The resulting NPS curve was not ideal, as shown in Fig. 6(c), indicating a certain level of correlation within the CMOS detector. Under identical experimental conditions, it might exhibit slightly weaker characteristics than the current Dexela commercial detector, particularly displaying a reduced correlation at high frequencies. This could potentially be attributed to the direct conversion process, where, because of the thickness of the material, ion drifting might cause some crosstalk among ions arriving at different pixels on the detection surface [37, 38].
Signal to Noise Ratio (SNR)
To measure the SNR, a 2.0 mm aluminum plate and 4.5 mm PMMA plate were stacked non-overlapped and fixed on a horizontal platform. The distance from the X-ray focal point to the surface of the CMOS detector was 476.0 mm and that from the PMMA plate to the CMOS detector was 76 mm. The PMMA plate remained stationary while the detector moved at a speed of 3.0 mm/s in both the vertical and horizontal directions, acquiring 5000 frames per movement. The CMOS detector was biased at -3 kV, and the chip Vrest was set at 700 mV. Six tube currents and their corresponding dose rates were scanned six times using a medical-grade G-242 X-ray source with a tube voltage of 50 kV. The detector moved at a speed of 3 mm/s on a vertical linear stage for each scan, collecting 5000 frames per scan. Because of excessive experimental interference, the obtained experimental results were unsatisfactory, and the SNR results were not computed [39].
DR imaging results
The tube voltage of the microfocus X-ray source was set to 20kVp, with a tube current of 200 μA. The cockroach was placed inside a transparent plastic bag and suspended on a platform. The distance from the X-ray focal point to the CMOS detector was 159.0 mm and that from the cockroach to the CMOS detector was 82.0 mm. Keeping the cockroach stationary, the detector moved at a speed of 3.0 mm/s in both the vertical and horizontal directions, collecting 10,000 frames per movement. The stitched DR image of the cockroach obtained through this process is shown in Fig. 7(a), displaying a clear resolution of the spines on the legs of the cockroach.
The tube voltage for the medical-grade G-242 X-ray source was set to 50kVp, with a tube current of 12.5 mA. A PCB was fixed onto a horizontal linear stage, with the distance from the X-ray focal point to the CMOS detector set to 204.0 mm and that from the PCB to the CMOS detector set to 43.0 mm. While maintaining the PCB stationary, the detector moved at a speed of 3.0 mm/s in both the vertical and horizontal directions, collecting 6000 frames per movement. The stitched DR image of the PCB obtained after this process is shown in Fig. 7(b), displaying a clear resolution of the pin information of the chip on the PCB.
conclusion
In this study, a CMOS direct conversion X-ray detector coupled with FC40 electronic fluorinated liquid was developed. The setup employed a specialized high-resolution and high-readout speed CMOS pixel array, the Topmetal II- chip, coupled with FC40 fluorinated liquid. FC40, an electronic fluorinated liquid, circumvents performance issues, such as delamination and uneven crystal growth, observed in traditional semiconductor materials, offering the advantage of customizable X-ray absorption thickness. To validate the feasibility of coupling CMOS chips with this liquid arrangement and offer a new direction for researching novel liquid direct conversion X-ray detectors, preliminary experiments demonstrated imaging performances consistent with expectations. The spatial resolution reached 4.0 lp/mm, showing an advantage in the low-frequency domain for MTF compared with commercial Dexela scintillator indirect detectors under the same experimental conditions. However, because of the design-related thickness of the fluorinated liquid layer, the NPS did not meet expectations. The experiments also produced relatively visible DR images of the cockroach and PCB. These preliminary findings demonstrate the viability of using FC40 electronic fluorinated liquid coupled with CMOS chips to fabricate X-ray detectors. This alternative approach to traditional semiconductor materials has distinct advantages when coupled with CMOS chips, presenting an extremely low dark current and negligible radiation damage. These characteristics suggest promising potential applications in low-energy and high-radiation environments.
However, some limitations are also associated with CMOS fluorinated liquid X-ray detectors. The slow charge migration resulting from the use of FC40 electronic fluorinated liquid, which is inert, may limit the response speed and sensitivity of the detector. Additionally, because of the relatively low atomic number of FC40 electronic fluorinated liquid, its ability to block X-rays is comparatively weak. Consequently, the performance of CMOS fluorinated liquid detectors may be constrained in high-energy X-ray detection.
Future studies will investigate the electrical characteristics of the FC40 fluorinated liquid and explore methods to test the charge-transfer efficiency. Moreover, to leverage its advantages when coupled with CMOS chips, further research can focus on combining FC40 with smaller pixel Topmetal chips to create high-precision X-ray detectors that are suitable for low-energy applications. This will assist in expanding the application prospects of CMOS fluorinated liquid detectors in areas such as medical imaging, materials analysis, and other low-energy and high-radiation detection fields. The investigation of the electrical properties of FC40 and the development of improved detector designs will contribute to enhancing the performance and capabilities of CMOS fluorinated liquid detectors. Further studies and development will enable the exploration of new applications and advancements in the field of X-ray detection.
Halide perovskites for high-performance X-ray detector
. Mater. Toda. 48, 155–175 (2021). https://doi.org/10.1016/j.mattod.2021.01.028X-ray detectors based on Ga2O3 microwires
. Materials 16, 4742 (2023). https://doi.org/10.3390/ma16134742Fast-response X-ray detector based on nanocrystalline Ga2O3 thin film prepared at room temperature
. Appl. Surf. Sci. 554,Halide perovskite, a potential scintillator for X-ray detection
. Small Methods 4,X-ray detectors in medical imaging
. Nucl. Instrum. Meth. A 731, 57–63 (2013). https://doi.org/10.1016/j.nima.2013.05.174Semiconductor of distinction
. Physics World 6, 54 (1993). https://doi.org/10.1088/2058-7058/6/4/27/metaEarly crystallization of amorphous selenium under high pressure studied by synchrotron XRD method
. J. Phys. Condens. Matter. 35,Recent development in X-ray imaging technology: Future and challenges
. Research 2021,Detecting ionizing radiation using halide perovskite semiconductors processed through solution and alternative methods
. Nat. Photon. 16, 14–26 (2022). https://doi.org/10.1038/s41566-021-00909-5Applications of halide perovskites in X-ray detection and imaging
. CrystEngComm 24, 2201–2212 (2022). https://doi.org/10.1039/D1CE01575CDesign, Fabrication and characterization of a unipolar charge sensing amorphous selenium X-ray detector
. Dissertation,Fabricating flexible wafer-size inorganic semiconductor devices
. J. Mater. Chem. C 8, 1915–1922 (2020). https://doi.org/10.1039/C9TC05781AHighly efficient light management for perovskite solar cells
. Scientific Reports 6,Supple formamidinium-based low-dimension perovskite derivative for sensitive and ultrastable X-ray detection
. ACS Nano. 17, 13638–13647 (2023). https://doi.org/10.1021/acsnano.3c02476X-ray detectors based on halide perovskite materials
. Coatings 13, 211 (2023). https://doi.org/10.3390/coatings13010211Zero-dimensional lead-free FA3Bi2I9 single crystals for high-performance X-ray detection
. J. Phys. Chem. Lett. 12, 1778–1785 (2021). https://doi.org/10.1021/acs.jpclett.1c00090Dark current modeling of thick perovskite X-ray detectors
. Front Optoelectron 15, 43 (2022). https://doi.org/10.1007/s12200-022-00044-1Thick-junction perovskite X-ray detectors: processing and optoelectronic considerations
. Nanoscale 14, 9636-9647 (2022). https://doi.org/10.1039/D2NR01643EHighly stable organic antimony halide crystals for X-ray scintillation
. ACS. Mater. Lett. 2, 633–638 (2020). https://doi.org/10.1021/acsmaterialslett.0c00133Lead-free halide Rb2CuBr3 as sensitive X-ray scintillator
. Adv. Mater. 31,Adhesion and friction properties of fluoropolymer brushes: on the tribological inertness of fluorine
. Langmuir 30, 12532–-12540 (2014).https://doi.org/10.1021/la501802bUsing volatile liquid FC-40 as pressure transmitting medium for pressure gauge calibrations
. J. Phys: Conf. Ser. 1065,Performance analysis of single-phase immersion cooling system of data center using FC-40 dielectric fluid
. Int. Commun. Heat Mass Transf. 145,Topmetal-II–: a direct charge sensor for high energy physics and imaging applications
. JINST. 11,A low-noise CMOS pixel direct charge sensor, Topmetal II-
. Nucl. Instrum. Methods Phys. Res. A. 810, 144–150 (2016). https://doi.org/10.1016/j.nima.2015.11.153Topmetal-M: A novel pixel sensor for compact tracking applications
. Nucl. Instrum. Meth. A 981,Measurement of electrical resistivity of hydrofluorocarbons – A review
. J. Electrost. 109,Electronics system for the cosmic X-ray polarization detector
. Nucl. Sci. Tech. 34, 64 (2023). https://doi.org/10.57760/sciencedb.j00186.00072Preliminary test of topmetal-II sensor for X-ray polarization measurements
. Nucl. Instrum. Meth. A 1008,A highly pixelated CdZnTe detector based on Topmetal II–sensor
. Chin. Phys. C 41,Advances in nuclear detection and readout techniques
. Nucl. Sci. Tech. 34, 205 (2023). https://doi.org/10.1007/s41365-023-01359-0CMOS based high-resolution dynamic X-ray imaging with inorganic perovskite
. physics.app-ph. 2210,The realisation of fast X-ray computed tomography using a limited number of projection images for dimensional metrology, NDT
. E International 137,ADAPTIVE-NET: deep computed tomography reconstruction network with analytical domain transformation knowledge
. Quant. Imaging Med. Surg. 10, 415–417 (2020). https://doi.org/10.21037/qims.2019.12.12MTF measurement by slanted-edge method based on improved zernike moments
. Sensors 23, 509 (2023). https://doi.org/10.3390/s23010509Modulation transfer function calculation and image restoration based on slanted-edge method
. Acta. Optica. Sinica. 30, 2891 (2010). https://doi.org/10.3788/aos20103010.2891An unsupervised dual contrastive learning framework for scatter correction in cone-beam CT image
. Comput. Biol. Med. 165,Deep learning image reconstruction algorithms in low-dose radiation abdominal computed tomography: assessment of image quality and lesion diagnostic confidence
. Quant. Imaging Med. Surg. 13, 3161–3173 (2023). https://doi.org/10.21037/qims-22-1227A comparison of two methods for measuring the signal to noise ratio on MR images
. Phys. Med. Biol. 44,The authors declare that they have no competing interests.
