Abstract:The presence of metal artifacts significantly compromises the clarity and clinical diagnostic value of computed tomography(CT)images,primarily manifesting as streaks and shadows that obscure key anatomical structures and impede the accurate assessment of pathological conditions by physicians.To address this issue,we proposes a novel CT image metal artifact reduction algorithm based on tissue processing and non-linear diffusion filtering.The algorithm initially employs multi-scale iterative threshold segmentation to precisely extract metal regions and generate metal projection trajectories,and subsequent wavelet denoising and interpolation of metal components in the projection domain contribute to producing preliminary corrected images.A tissue processing technique is then utilized to restore bone structure deficiencies caused by linear interpolation,whereas non-linear diffusion filtering is applied to suppress residual artifacts and effectively preserve edge details.Experimental results revealed that compared with conventional methods,the application of this algorithm facilitates a significant reduction in metal artifact intensity,prevents the generation of secondary artifacts,and maintains the integrity of bone and soft tissue structures surrounding metal implants,thereby enhancing CT image quality and clinical diagnostic utility.

Abstract:Objective:This study proposes an unsupervised learning model,ViTD-CycleGAN,based on an improved CycleGAN to synthesize computed tomography(CT)images from cone-beam computed tomography(CBCT)images.Our aim is to enhance the quality and realism of synthetic CT(sCT)images.Methods:ViTD-CycleGAN incorporates a U-Net framework based on a vision Transformer(ViT)and depth-wise convolution(DW)into its generator,where the self-attention mechanism of the Transformer is leveraged to extract and preserve crucial features and detailed information.Additionally,a gradient penalty and pixel-wise loss function are introduced to enhance the stability of the model training and image consistency.Results:Quantitative evaluation metrics(MAE,PSNR,and SSIM)for head and neck as well as chest datasets indicate the superior performance of the proposed model compared with existing unsupervised learning methods.Ablation experiments show that the DW significantly improved the model performance.Visual-display analysis confirms that the sCT images generated using the ViTD-CycleGAN exhibit higher image quality and realism.Conclusion:Compared with other unsupervised learning methods,the proposed method can improve the quality of CBCT-synthesized CT images and thus offer potential clinical application value.

Abstract:Objective:To explore the impact of Mono technology and resolution parameter in Mono+technology on image quality with the goal of providing reference for the appropriate selection of these post-processing parameters in virtual monoenergetic images(VMI).Methods:The data included a computed tomography(CT)quality control phantom scanned in energy mode and portal venous phase data of contrast-enhanced upper abdomen energy CT scan of a sample patient.Mono and Mono+technologies with different resolution parameters(1–10)and a default setting of 2 were used to reconstruct 70 keV VMIs.The noise power spectrum(NPS)was employed to analyze the image noise characteristics of both phantom and patient.The CT value,standard deviation(SD),signal-to-noise ratio(SNR),and contrast-to-noise ratio(CNR)of the liver parenchyma,and edge rise slope(ERS)between the liver parenchyma and portal vein were measured and calculated in the sample patient.Results:The NPS curves of the phantom showed that image noise was most pronounced with Mono technology and gradually increased with higher resolution parameters in Mono+technology.The NPS curves for the patient exhibited a trend similar to that of the phantom.The average and peak spatial frequency values of NPS curves did not differ significantly and CT values of the liver parenchyma remained stable across different reconstruction techniques.However,the SD values of the liver parenchyma were lower at lower resolution parameters in Mono+technology,corresponding to relatively higher SNR and CNR values.The ERS value between the liver parenchyma and portal vein was the lowest when the resolution parameter in Mono+technology was 2.The ERS gradually increased as resolution increased.Conclusion:Mono+technology can reduce image noise compared to Mono technology without significantly altering image texture.The increased resolution in Mono+technology can improve image sharpness at the expense of higher image noise.Optimized SNR and CNR values can be obtained using the default resolution parameter settings of Mono+technology,but the corresponding ERS may be low.

Abstract:Objectives:We explored the value of combining coronary computed tomography angiography(CCTA)-derived features with clinical factors to predict all-cause mortality after(transcatheter aortic valve replacement,TAVR).Methods:A total of 380 candidates for CCTA examined between September 2017 and September 2023 were enrolled in this multicenter retrospective cohort study with clinical end point being all-cause mortality.Univariate and multivariate Cox regression analyses were performed to identify independent clinical and CCTA-derived predictors of post-TAVR mortality to construct clinical or imaging factor only models and a combined factor model.Model performance was evaluated using time-dependent area under the receiver operating characteristic curve(time-AUC)and Harrell's concordance index(C-index).To compare nested models,χ2 likelihood ratio analyses were performed.Results:Median follow-up duration was 706 days.All-cause mortality occurred in 13%(48/380)of the patients.The combined model(age+diabetes+coronary artery disease reporting and data system,(CAD-RADS)+(segment stenosis score,SSS)+(left anterior descending artery-fat attenuation index,LAD-FAI))demonstrated superior predictive performance for all-cause mortality with higher C-index(0.690,95%confidence interval[CI]:0.590-0.772)compared with the clinical model(age+diabetes)(C-index,0.630,95%CI:0.538-0.718).Likelihood ratio tests confirmed significantly better fit of the combined model than the clinical model.SHAP analysis indicated that LAD-FAI and age were the most influential factors contributing to model prediction accuracy.Conclusion:A combined model incorporating CCTA features and clinical risk factors can improve all-cause mortality prediction efficacy in patients after TAVR.

Abstract:Objective:To explore the application value of the combined use of different keV virtual monoenergetic images(VMI)and orthopedic metal artifact reduction(O-MAR)technologies to optimize the image quality of oropharyngeal computed tomography(CT)in patients with oral metal implants.Methods:Phantom study based on the clinical oropharyngeal scanning protocol for spectral CT;the CTP528 module of the Catphan600 phantom was scanned.Images at 66 keV and 80-160 keV(20 keV intervals)were reconstructed with and without the O-MAR,and the modulation transfer function(MTF)was measured.Clinical study:A retrospective analysis was conducted on 55 oropharyngeal CT examinations performed using Philips IQon Spectral CT,including 25 cases with high-density artifacts and 30 cases with low-density artifacts.Conventional images(CI),66 keV,and 80-160 keV(20 keV intervals)VMI were reconstructed with and without the O-MAR.The layer with the most severe artifacts was selected as the artifact layer,and the adjacent layer without artifact interference was selected as the reference layer.The artifact index(AI)of the region of interest(ROI)in the artifact layer,CT value deviation of the ROI between the artifact and reference layers,absolute contrast(AC),and contrast-to-noise ratio(CNR)of the ROI in the reference layer were measured.The images were subjectively evaluated in three dimensions using a three-point Likert scale:artifact removal ability,new artifact formation,and tissue credibility in the artifact removal area.Results:Phantom study:Enabling the O-MAR had no effect on MTF values.The MTF values of the VMI images were approximately 3%lower than those of the CI images.When the O-MAR and VMI(VMIMAR)were used together,the artifact coefficient of high/low-density artifacts and the CT value deviation of high-density artifacts decreased with increasing energy levels.The average values decreased from 125.89/192.47 and 166.04 HU in CI images to 27.61/18.30 and 10.36 HU at 160 keV.The CT value deviation of low-density artifacts was the smallest at 100 keV,with an average of-4.35 HU.When VMI was used alone,the average artifact coefficient of high/low-density artifacts was the lowest at 120/160 keV(65.16/125.31),and the average CT value deviation was the smallest at 160 keV 12.87/-46.11 HU.The AC/CNR ratio of the images decreased with increasing energy levels.When the O-MAR and VMI were used together,the subjective score reached 3 points in all three dimensions when the energy level reached 120 keV.When VMI was used alone,the highest subjective score was 2 points.Conclusion:The combined use of VMI and O-MAR can effectively reduce the impact of oral metal implant artifacts on the observation of surrounding tissues.Selecting VMIMAR images with a virtual energy level around 120 keV balances the artifact removal effect and image contrast.

Abstract:As fine-scale exploration of unconventional oil and gas reservoirs advances,the need for high-resolution imaging of complex structures,such as fractures and cavities near boreholes has become increasingly critical.To address the resolution limitation of conventional acoustic logging tools,this paper presents a collaborative imaging method based on shear wave transformation and Kirchhoff integral migration.The approach is tailored for complex reservoirs and leverages a newly developed eight-directional acoustic logging instrument.The proposed method features multi-scale sparse representation capabilities,enabling effective extraction of weak reflected wave signals within the borehole environment and accurate imaging and localization of reflected wave energy.Its ability to identify karst cave systems and double-inclined interfaces has been validated through both theoretical simulations and real well data analysis.Numerical results demonstrate that the method enhances the accuracy of reflected wave identification and improves the resolution of structural imaging.These findings offer technical support for the application of eight-directional logging instruments in geologically complex reservoir settings.

Abstract:Objective:Exploring the potential of optimizing Mono+technology resolution parameters for the reduction of image noise and radiation dose in virtual monoenergetic imaging(VMI).Methods A computerized tomography(CT)quality control phantom was scanned in energy CT mode at different radiation doses(0.8 to 15.0 mGy).Images were reconstructed from VMI acquisitions at 70 keV using Mono technology and Mono+technology with different resolution parameters(2,6,10).A conventional CT scan was performed at 15.0 mGy,and corresponding images were reconstructed from poly-energetic imaging(PEI).The noise power spectrum(NPS)was used to analyze the image noise characteristics.The relationship between image noise and radiation dose was fitted to estimate the potential of VMI reconstructions using different technologies to reduce scan radiation dose.Results As the radiation dose decreased,the image noise in VMI increased.The images reconstructed from VMI using Mono technology had the highest image noise,whereas those reconstructed using Mono+technology with a resolution parameter of 2 had the lowest image noise.The average and peak spatial frequency values of NPS curves of VMI were close to those of PEI.The images reconstructed from VMI using Mono technology and Mono+technology with resolution parameters of 2,6,and 10 achieved the noise level of PEI with radiation dose reductions of 12.0%,37.3%,23.6%,and 19.8%,respectively.Conclusion Optimizing the resolution settings in Mono+technology has the potential to further reduce radiation dose,with an expected reduction of more than one-third compared to that of conventional CT scans.

Abstract:Objective:To visually analyze the literature on computed tomography virtual monoenergetic imaging(VMI)clinical research and discuss the hotspots and trends in this field.Methods:We searched VMI clinical research-related literature that published in China National Knowledge Infrastructure(CNKI)between 1 January 2020 and 28 February 2025,then visually analyzed the number of publications,authors,institutions,journals,and the keywords using CNKI as well as CiteSpace and VOSviewer software tools.Results:We included a total of 441 articles in this study.The number of publications was stable in recent five years at a level of more than 80 articles per year.The authors and institutions of the articles were mainly concentrated in affiliated medical college and university hospitals.The cooperation between authors and institutions remained limited.After search word exclusion,the co-occurrence maps of keywords revealed that the three most frequent keywords were"image quality"(52 articles),"radiation dosage"(17 articles),and"metal artifact"(14 articles),while the three most central keywords were"image quality"(0.23),"lung adenocarcinoma"(0.08),and"radiomics"(0.08).Keyword clustering and burst word mapping suggested that recent research in this field has gradually shifted from technical validation,(e.g.,VMI-mediated image quality improvement)to clinical applications(e.g.,specific disease diagnosis).Conclusion:In recent five years,VMI clinical research in China displays stable outputs and has been published in high-quality journals.However,collaboration among authors and institutions remains weak and needs to be strengthened.The research hotspots are gradually shifting from technical validation to clinical VMI applications.

Abstract:Objective:To explore the independent risk factors and predictive efficacy of acute necrotic collection(ANC)conversion to encapsulated necrosis walled-off necrosis(WON).Methods:A Retrospective analysis of CT/MRI features in 53 cases of ANC,divided into the WON and absorption groups,based on the outcome after 4 weeks of ANC formation.The chi square test or T-test were used to compare the statistical significance of differences in etiology and laboratory tests between the two groups.Regression analysis was used to identify independent risk factors affecting the outcome of ANC.Receiver operating characteristic(ROC)curves were used to obtain the area under the curve(AUC)and evaluate the predictive efficacy of each risk factor for the outcome of ANC to WON.Results:Logistic regression analysis showed that the P-values for necrotic volume≥30%and MCTSI score>6 were both<0.05,with OR values of 9.21 and 16.04,respectively.ROC curve analysis showed that the P-values for necrotic volume≥30%and MCTSI score>6 were both<0.05,with AUC values of 0.86 and 0.88,respectively.Conclusion:Necrosis volume≥30%and MCTSI score>6 points are independent risk factors for the progression of ANC into WON,and the predictive performance of both is significant.

Abstract:Objective:To explore the feasibility of using a low-dose contrast agent volume and a low flow rate in dual-layer detector spectral computed tomography(CT)for CT pulmonary angiography(CTPA).Methods:In total,80 patients clinically diagnosed with pulmonary embolisms were collected for CTPA examination and randomly divided into two groups,comprising 40 cases each.The low-dose group received a contrast agent dose of 15 mL at a flow rate of 2 mL/s,and the images were reconstructed using a 40 keV monochromatic spectrum.The conventional group underwent a small-dose test with a contrast agent dose of 50 mL at a flow rate of 5.0 mL/s.The volume CT dose index(CTDIvol),dose-length product(DLP),effective dose(ED),CT values,signal-to-noise ratio(SNR),and contrast-to-noise ratio(CNR)were recorded and calculated for both groups,and the image quality was subjectively scored.Results:(1)The differences in these indicators between the two groups were statistically significant.The pulmonary artery CT values for the conventional and the low-dose groups were 368.450 HU and 569.050 HU,respectively,with a statistically significant difference between the two groups.There were statistically significant differences in SNR and CNR between the two groups.(2)There was no statistically significant difference in the subjective scoring of image quality between the two groups.(3)The DLP values for the conventional and low-dose groups were 376.500 mGy/cm and 294.500 mGy/cm,respectively;the CTDIvol values were 11.150 mGy and 8.350 mGy,respectively;and the ED values were 5.271 mSv and 4.123 mSv,respectively,with statistically significant differences between the groups.Conclusion:The scanning protocol with the"low"contrast agent dose and flow rate using dual-layer detector spectral CT for CTPA examinations met the diagnostic requirements while reducing risks such as contrast extravasation and contrast-induced nephropathy.

Abstract:Objective:To optimize the virtual monoenergetic imaging(VMI)energy level for pulmonary computed tomography(CT)angiography(PCTA)based on image quality and thrombus visualization and evaluate its impact on the confidence of radiological technologists in detecting pulmonary embolism(PE).Methods:Patients who underwent PCTA because of clinical suspicion of PE at our hospital between February 2025 and March 2025 were prospectively screened.All examinations were performed using dual-source dual-energy CT,and the images were reconstructed into 120 kVp equivalent polyenergetic imaging(PEI)and VM images at energy levels of 70,60,50,and 40 keV.One radiological technologist conducted objective measurements to assess the ascending aorta,main pulmonary artery,left pulmonary artery branch,right pulmonary artery branch,right lower lobe pulmonary artery branch,left lower lobe pulmonary artery branch,liver parenchyma,and erector spinae muscle to obtain CT values(Hounsfield units[HU])and their standard deviations(SD)to calculate the signal-to-noise ratio(SNR).The contrast-to-noise ratio(CNR)was calculated using the SD value of the erector spinae muscle as the background noise.Two radiological technologists evaluated the images of 23 patients with thrombi using a 5-point scale to assess their confidence in alerting them to PE.A score of 5 denoted a high certainty of thrombus presence,necessitating immediate notification to the diagnostic physician for review and issuance of a critical value,whereas a score of 1 denoted no thrombus detection,requiring no prior review by the diagnostic physician.Differences in objective and subjective assessments between the different image groups were compared.Results:The study included 98 patients,with a mean age of(72.06±13.41)years,comprising 47 male and 51 female patients;23 had thrombi and 75 did not.Objective measurements revealed that in the 40,50,and 60 keV VMI images,as the VMI energy level decreased,the corresponding CT values increased,providing improved tissue contrast.However,the SD values also increased,indicating higher noise levels.In the 40 and 50 keV VMI images,the CNR values for the vessels and thrombi were higher than those in the PEI images.Subjective assessment showed that the two technologists detected almost all thrombi(225/230,97.8%),but had greater confidence in alerting PE when observing 50 keV VMI images than when observing PEI images((4.41±0.88)vs.(4.72±0.58)).Conclusion:For the clinical task of detecting PE using PCTA,50 keV VMI images are preferred to enhance the confidence of radiological technologists in alerting patients with PE,facilitating timely physician review and issuance of critical values,with the potential to improve clinical diagnoses and the treatment process for patients.

Abstract:Objective:To construct an improved deep learning computer-aided diagnosis model based on convolutional neural network(CNN)and attention mechanism proposed as inception spatial and channel attention network(ISANET)and evaluate the model's performance,comparing it with the traditional CNN model.Methods:A total of 619 lung CT images of 60 patients with lung squamous cell(LUSC)carcinoma or lung adenocarcinoma(LUAD)confirmed by surgical pathology and 30 patients with normal lungs were collected retrospectively to form Dataset A;a total of 737 public dataset images were collected to form Dataset B.The two datasets were randomly divided into training and test sets at a 6︰4 ratio.Construct the ISANET model and conduct training and verification,then record the precision ratio and recall ratio to calculate the F1 score and evaluate the performance of the ISANAT model.Finally,the ISANET model was compared with the traditional CNN models such as AlexNet,VGG16,InceptionV3,MobilenetV2,and ResNet18 by drawing the Precision-Recall(P-R)curve and calculating the area under the P-R curve to evaluate the classification performance of different models for LUSC and LUAD.Results:Compared with the traditional CNN model,the accuracy of the ISANET model for non-small cell lung cancer classification improved significantly,reaching 99.6%and 95.2%in Dataset A and Dataset B,respectively.Conclusions:The ISANET model provides better non-invasive prediction of LUSC and LUAD,improves the accuracy of CT imaging identification of lung squamous cell carcinoma and lung adenocarcinoma,and can help diagnosticians quickly and accurately classify non-small cell lung cancer.

Abstract:Objective:To compare the image quality and radiation dose of conventional images obtained from single-source computed tomography(CT)scans with those of polyenergetic imaging(PEI)and virtual monoenergetic imaging(VMI)obtained from the dual-source upper abdominal contrast-enhanced CT scans of the same patients.Methods:Patients who underwent upper abdominal contrast-enhanced CT examinations using both single-source(120 kVp)and dual-source(70/Sn 150 kVp)protocols were retrospectively screened considering the change in the scanning protocol between November 2024 and March 2025.During this interval,no surgical procedures were performed,and there was no significant change in the patient's condition.Associated with the portal venous phase were conventional 120 kVp PEI images(CON)from a single-source CT scan as well as mixed 120 kVp equivalent PEI images(MIX)and 70keV VMI images from a dual-source CT scan.To calculate the signal-to-noise ratio(SNR),CT values in Hounsfield units(HU)and their standard deviations(SD)were measured at six anatomical sites,including the liver,spleen,abdominal aorta,portal vein,psoas major muscle,and anterior abdominal subcutaneous fat.The contrast-to-noise ratio(CNR)was calculated using the CT value and SD of the anterior abdominal subcutaneous fat as background noise.Subjective image quality was evaluated using a 5-point Likert scale to assess image noise,clarity,sharpness,and contrast.The scan radiation dose was recorded and the size-specific dose estimate(SSDE)was calculated.The differences in image quality and radiation dose among the different groups were compared.Results:The study included 34 patients with mean age of(68.1±10.8)years,of which 28 were male(82.4%).The average interval between the two examinations was(75.6±28.0)days.Objective measurements revealed significant differences in the SNR values of the abdominal aorta and portal vein among the different imaging groups,with higher SNR values in the MIX and VMI groups than in the CON group.No significant differences in the CNR values were observed for the liver,spleen,abdominal aorta,portal vein,or psoas major muscle among the different imaging groups.Subjective evaluations showed that all the images received scores above the requirements for clinical diagnostic use.No significant differences in image clarity were observed between the different imaging groups.Image noise was lower in the CON group than in the MIX and VMI groups.Image sharpness was lower in the VMI group than in the CON and MIX groups,but image contrast was higher in the VMI group than in the CON and MIX groups.The SSDE of dual-source CT scans was significantly lower than that of the single-source CT scans.Conclusion:The comparison study on the same individuals eliminated the potential influence of individual differences and further confirmed that mixed 120 kVp equivalent PEI images and 70 keV VMI can reduce the radiation dose with image quality comparable to those of 120 kVp conventional PEI images.This study underscores the necessity of dual-source scanning and VMI reconstruction for contrast-enhanced CT of the upper abdomen.

Abstract:Objective:In this study,energy spectrum CT optimal virtual single energy image(VMI)with virtual flat scan(VUE)phase subtraction was compared with the matched mask bone elimination(MMBE)method in the subtraction of head and neck CT angiography(CTA).Methods:This retrospective study comprised 80 patients who underwent head and neck CTA between January 2024 and February 2025 at our institution.Reconstruction of image data was performed at 10 keV intervals,with eleven groups of VMI data from 40～140 keV ultimately reconstructed.The CT values and standard deviations of the common carotid artery(CCA),internal carotid artery(ICA),middle cerebral artery(MCA),and anterior cerebral artery(ACA)were measured,and the signal-to-noise ratio(SNR)and the contrast-to-noise ratio(CNR)were calculated.Then,subjective scoring of the 11 groups of VMI images was performed by two senior attending physicians.The outcomes of the objective evaluation indexes and subjective scoring were statistically analyzed to identify the best VMI images.The subtraction images obtained from the optimal VMI subtraction VUE were compared with the MMBE subtraction images,including arterial and venous system vascularization,debulking,overall image quality grade,and effective radiation dose.Results:The differences in the SNR and CNR of CCA,ICA,MCA,and ACA between the VMIs of each energy level were statistically significant,with the SNR and CNR of 60 keV images found to be higher than those of other VMI images.The radiologists showed good agreement on their subjective evaluations of VMI image quality at each energy level(Kappa>0.61).The differences in subjective scores between the VMI at each energy level were statistically significant,with the 60 keV images showing the highest subjective scores.The subtracted images of ICA intracranial segment,vertebral artery(VA)intracranial segment,and intracranial artery after subtraction of the 60 keV images and VNC were significantly better than those of the MMBE method.The debulking effect of its subtraction was also significantly better than that of the MMBE method.The percentage of the overall image quality of Grade I was 88.75%(71/80),significantly higher than that of the MMBE method(77.50%,62/80).Its average effective radiation dose was(0.96±0.02)mSv,approximately 44.24%lower than that of the MMBE method-a difference that was statistically significant.Conclusion:We identified 60 keV as the best virtual single energy for displaying head and neck CTA.Its subtracted image was better in terms of head and neck arterial vascular display,debridement effect,and overall image quality grade.This highlights its potential as an alternative to the traditional matched-mask debridement method.By omitting the true flat scan,the radiation dose can be significantly reduced,providing a novel mechanism for the optimization of head and neck CTA.

Abstract:Marchenko imaging is a method that utilizes full scattering information.Compared with conventional migration methods,it can reduce artefacts caused by internal multiples.This paper first introduces the retrieval of the Marchenko Green's function in detail.Subsequently,based on an analysis of imaging conditions applicable to the Marchenko Green's function,it proposes an improved multidimensional deconvolution imaging condition to enhance migration quality.To address the challenges of large data volumes and low computational efficiency in 3D Marchenko imaging,the paper proposes master-slave process group parallel strategies,as well as aperture and frequency band selection strategies.Numerical experiments verify the effectiveness of the proposed method.

Abstract:The Xujiahe Formation fractured oil and gas reservoirs in the Sichuan Basin have broad exploration prospects.The existing seismic data mainly faces challenges of low signal-to-noise ratio and chaotic reflections.Traditional offset vector tile(OVT)domain imaging techniques are affected by acquisition factors,resulting in uneven coverage in various directions,which is not conducive to observing fractures from various directions.To address these issues,this study systematically optimizes pre-stack noise suppression,pre-stack depth domain velocity modeling,and imaging technology to significantly improve the imaging quality of the Xujiahe Formation reservoir and provide high-precision basic data for fracture prediction.The main innovative mainly include three components:(1)Optimizing the traditional domain based pre-stack noise suppression method,based on the rearrangement of shot domain data,reducing data redundancy and time of domain based sorting noise,and significantly improving the efficiency and effectiveness of pre-stack noise suppression;(2)On the basis of grid tomography modeling of common offset gathers,combined with omnidirectional reflection angle domain grid tomography modeling,reasonably applying multiple offset gathers improves modeling accuracy and efficiency;and(3)The use of omnidirectional angle domain imaging technology compensates for shortcomings of traditional OVT domain imaging,such as uneven coverage in various directions and low imaging quality in far offsets,and improves the imaging quality of data.Empirical research based on the JY block in the Sichuan Basin shows that our method significantly improves imaging quality compared to traditional OVT domain results.The Xujiahe Formation imaging quality is significantly improved,the correlation between well seismic activity is enhanced,the ability to identify small fractures is enhanced,and the characteristics of fracture development are clearer.This research provides an efficient data processing technology system and practical framework for the exploring similar fractured oil and gas reservoirs.

Abstract:Objective:This study investigated the impact of deep learning image reconstruction(DLIR)combined with gemstone spectral imaging(GSI)CT scanning and metal artifact reduction(MAR)on metal artifacts from lumbar spine implants.Methods:A retrospective analysis was conducted on 40 patients with lumbar metal implants who underwent abdominal spectral CT scans at our hospital.After spectral abdominal scanning,images were reconstructed using Adaptive Statistical Iterative Reconstruction-V(ASiR-V)at 50%(AR50 group)and DLIR at a high level(DH group).MAR was applied to both groups to obtain reconstructions(AR50-MAR group and DH-MAR group).Regions of interest(ROI)were delineated and measured for metal artifacts and surrounding tissues on the four sets of images.CT values and standard deviations(SD)were recorded,and the signal-to-noise ratio(SNR),contrast-to-noise ratio(CNR),and artifact index(AI)of the images were calculated and compared.Two radiologists subjectively evaluated the quality of the images,the severity of metal artifacts,and the area of the artifacts on a 4-point scale.Results:Under the same energy level of 68 keV,statistically significant differences were observed in SD,CNR,and AI values among the four groups.The SD values in the MAR groups were lower than those in the non-MAR groups.The images reconstructed with DH-MAR had the highest CNR and SNR and the lowest AI values.Conclusion:For patients with lumbar metal implants,the combination of DLIR and MAR in GSI CT can significantly reduce metal artifacts while greatly improving the signal-to-noise ratio of images of the tissue,which is more conducive to clinical diagnosis.

Abstract:Objective:Using hepatic magnetic resonance imaging proton density fat fraction(MRI-PDFF)as the reference standard,we constructed a 3D-ResNet model based on dual-echo Dixon fat-only images to explore the value of deep learning in the grading of fatty liver disease.Materials and Methods:Overall,132 participants were prospectively enrolled to undergo both dual-echo and multi-echo Dixon liver MRI.Among them,94 participants were diagnosed with fatty liver disease by MRI-PDFF,and 38 participants were without fatty liver disease.The inter-sequence agreement of fat-only images between the two Dixon protocols was evaluated across hepatic lobes to determine the optimal lobe for region of interest(ROI)placement.Circular ROIs were drawn on dual-echo Dixon fat-only images,and the corresponding PDFF values from multi-echo Dixon served as the grading reference.In total,3,104 ROIs were included:Grade 0(normal),1,026;Grade 1(mild),1,042;and Grade 2/3(moderate-to-severe),1,036.A 3D-ResNet model was developed using dual-echo Dixon fat-only images with hyperparameter optimization and overfitting prevention.The model performance was assessed on the test set via receiver operating characteristic(ROC)and precision-recall(P-R)curves.The values of area under the curve(AUC),average precision(AP),accuracy,precision,recall,specificity,and F1-score were calculated.Results:The following were achieved by the 3D-ResNet model based on dual-echo Dixon fat-only images:a macro-average AUC of 0.971,mean AP of 0.943,accuracy of 88.60%,and weighted F1-score of 88.59%for distinguishing normal,mild,and moderate-to-severe fatty liver.Conclusion:The 3D-ResNet model built on dual-echo Dixon fat-only images demonstrates high accuracy and robustness for grading fatty liver disease and can serve as an effective screening tool for assessing hepatic steatosis.

Abstract:Objective:Organ dose modulation(ODM)was used in male pelvic electronic computed tomography(CT)scans.To avoid affecting clinical diagnosis,the effective radiation dose of pelvic tissues and organs in men should be reduced to provide a theoretical basis for the future application of this technology in the epidemiological large-scale CT screening of male pelvic diseases and reduce radiation damage in the population.Methods:150 adult male subjects with planned pelvic CT scans were selected in a randomized controlled clinical trial.In this study,the subjects were divided into AODM,BATCM,and CLOW groups according to the randomization table method.In the ODM group,the ODM utilized automatic tube current modulation(ATCM);only ATCM and the other scanning conditions were consistent;in the CLOW group,the ATCM was set to fixed tube current,and the other conditions were consistent.The peak human square tubular current was measured and recorded by placing the DLVA-ACT voltammetric characteristic tester in the anterior,left,posterior,and right sides.The effective dose(ED)was calculated from the radiation dose table data automatically generated after the CT scan.The effective radiation dose of the prostate was estimated by calculating SSDEprostate from the AAPM220 report.At the central level of the pelvic prostate,the region of interest(ROI)was selected in the prostate and gluteus muscle regions to compare standard deviation(SD)and signal-to-noise ratio(SNR).The reconstructed transaxial and coronal images were scored under the pelvic soft tissue window.Results:The objective data showed that the anterior tube current was significantly lower than that in the BATCM group;no statistical difference was observed in the other directions.Pairwise comparisons between the SD and SNR levels in the three prostate regions were statistically significant;the SD and SNR levels in the gluteus maximus region were not significant between the AODM and BATCM groups.The difference in the signal organ EDOS and overall ED between the volume CT dose index(CTDIvol).Compared with the BATCM group,the tube current and ED in the AODM group decreased.Conclusion:Compared with traditional CT scanning,ODM is used in male pelvic electronic CT,which can reduce the radiation intensity of radiation-sensitive organs in the pelvic cavity and ED of single organs to obtain medical images that satisfy clinical standards.

Abstract:Objective:To investigate the imaging characteristics and differences among pancreatic serous cystic neoplasm(SCN)subtypes and their pathological basis,aiming to improve preoperative diagnostic accuracy.Methods:Imaging and pathological data of 65 patients with pathologically confirmed pancreatic SCN were analyzed retrospectively.The lesions were categorized into four subtypes based on morphological features:macrocystic,microcystic,mixed,and solid.Computed tomography/magnetic resonance imaging(CT/MRI)imaging features of each subtype were compared and correlated with histopathological findings.Results:No statistically significant differences were observed among the age,lesion location,clinical manifestations,or comorbidities of subtypes.Significant differences were observed in lesion density and enhancement patterns.All SCN exhibited hyperintensity on T2-weighted imaging,absence of communication with the pancreatic duct,and no metastatic signs.In addition,other shared imaging features included lobulated contours and multicystic structures,with internal calcifications(on CT)being a specific but less frequent finding(28.3%).Macrocystic SCN lacked central scars,while microcystic and mixed subtypes displayed higher MRI detection rates of central fibrous scars(63.2%).Solid-type SCN were smaller in size,demonstrating marked homogeneous enhancement.Pathological analysis revealed:macrocystic SCN had thin stroma and sparse vasculature between cysts;microcystic SCN were rich in fibrous scar tissue;mixed-type combined features of both;solid-type SCN were characterized by high vascular density and dense fibrous stroma,with microcystic structures undetectable by conventional imaging.Conclusion:Each pancreatic SCN subtype exhibited characteristic imaging features,and these imaging findings correlate well with pathological characteristics.A thorough understanding of their pathological basis,combined with multimodal imaging techniques,can significantly enhance the accuracy of imaging diagnosis.