Objective:To investigate the denoising performance of different deep learning (DL) strategies on low-dose brain 18F-FDG PET images. Methods:This retrospective methodological study was conducted on brain PET/CT images of 50 patients (35 males, 15 females, age 20-87 years) who received 3.7MBq/kg 18F-FDG at the Fifth Affiliated Hospital of Sun Yat-sen University between May 2023 and January 2024. Full-dose PET data were acquired with 2min scan. CT scans were acquired before PET scanning. Low-dose PET sinograms were generated by down-sampling the full-dose list mode data to 1/2, 1/4, and 1/20 of full-dose count level. Both full-dose and low-dose sinograms were reconstructed with random, CT-based attenuation and scatter corrections using the three-dimensional (3D) ordered-subsets expectation maximization (OSEM) algorithm (2 iterations, 20 subsets). A total of 4 DL denoising methods were established: (1) 3D conditional generative adversarial networks (GAN) using only low-dose PET as input (GAN-1); (2) 3D attention-based GAN (AttGAN) with low-dose PET input (AttGAN-1); (3) 3D AttGAN with low-dose PET and CT inputs (AttGAN-2); (4) 3D AttGAN with frequency-separation using low-dose PET and CT inputs (AttGAN-FS-2). For AttGAN-FS-2, during the frequency division process, high- and low-frequency components were extracted from the PET reconstructed images via Fourier transform, then inversed Fourier transform, denoised separately, and finally combined to produce the final denoised images. The dataset was separated into training (70%), validation (10%) and testing (20%) sets using simple random sampling without replacement with a fixed random seed. A 5-fold cross-validation scheme was then applied to test all 50 patients. Performance was evaluated against full-dose PET using normalized mean square error (NMSE), structural similarity (SSIM), peak signal-to-noise ratio (PSNR), contrast-to-noise ratio (CNR), SUV mean and SUV max bias of selected brain ROIs. Wilcoxon signed rank test was used to analyze the differences between the denoising methods. Results:AttGAN-FS-2 showed the best performance among all dose levels, with statistical difference as compared by low-dose PET and GAN-1 denoised images for NMSE, SSIM, PSNR, and CNR ( Z values: 2.92-6.15, all P<0.005). NMSE, SSIM quantitative evaluation results (median) of each model at 1/20 dose were: GAN-1: 0.08, 0.87, AttGAN-1: 0.08, 0.88, AttGAN-2: 0.07, 0.89, AttGAN-FS-2: 0.06, 0.91, respectively ( Z values: 3.24-5.77, all P<0.005). Conclusion:The DL-based method combined with multiple strategies AttGAN-FS-2 shows improved denoising performance for low-dose brain PET images.

Objective:To investigate the denoising performance of different deep learning (DL) strategies on low-dose brain 18F-FDG PET images. Methods:This retrospective methodological study was conducted on brain PET/CT images of 50 patients (35 males, 15 females, age 20-87 years) who received 3.7MBq/kg 18F-FDG at the Fifth Affiliated Hospital of Sun Yat-sen University between May 2023 and January 2024. Full-dose PET data were acquired with 2min scan. CT scans were acquired before PET scanning. Low-dose PET sinograms were generated by down-sampling the full-dose list mode data to 1/2, 1/4, and 1/20 of full-dose count level. Both full-dose and low-dose sinograms were reconstructed with random, CT-based attenuation and scatter corrections using the three-dimensional (3D) ordered-subsets expectation maximization (OSEM) algorithm (2 iterations, 20 subsets). A total of 4 DL denoising methods were established: (1) 3D conditional generative adversarial networks (GAN) using only low-dose PET as input (GAN-1); (2) 3D attention-based GAN (AttGAN) with low-dose PET input (AttGAN-1); (3) 3D AttGAN with low-dose PET and CT inputs (AttGAN-2); (4) 3D AttGAN with frequency-separation using low-dose PET and CT inputs (AttGAN-FS-2). For AttGAN-FS-2, during the frequency division process, high- and low-frequency components were extracted from the PET reconstructed images via Fourier transform, then inversed Fourier transform, denoised separately, and finally combined to produce the final denoised images. The dataset was separated into training (70%), validation (10%) and testing (20%) sets using simple random sampling without replacement with a fixed random seed. A 5-fold cross-validation scheme was then applied to test all 50 patients. Performance was evaluated against full-dose PET using normalized mean square error (NMSE), structural similarity (SSIM), peak signal-to-noise ratio (PSNR), contrast-to-noise ratio (CNR), SUV mean and SUV max bias of selected brain ROIs. Wilcoxon signed rank test was used to analyze the differences between the denoising methods. Results:AttGAN-FS-2 showed the best performance among all dose levels, with statistical difference as compared by low-dose PET and GAN-1 denoised images for NMSE, SSIM, PSNR, and CNR ( Z values: 2.92-6.15, all P<0.005). NMSE, SSIM quantitative evaluation results (median) of each model at 1/20 dose were: GAN-1: 0.08, 0.87, AttGAN-1: 0.08, 0.88, AttGAN-2: 0.07, 0.89, AttGAN-FS-2: 0.06, 0.91, respectively ( Z values: 3.24-5.77, all P<0.005). Conclusion:The DL-based method combined with multiple strategies AttGAN-FS-2 shows improved denoising performance for low-dose brain PET images.

Objective To explore the effect of deep learning image reconstruction(DLIR)algorithm combined with"double low"[low voltage(kV)and low contrast agent dosage]technique on the image quality of coronary computed tomography angiography(CCTA)in overweight patients compared with adaptive statistical iterative reconstruction(ASIR-V)and filtered back projection(FBP).Methods Fifty-two patients with body mass index(BMI)between 25.1 kg/m2and 28 kg/m2 who underwent CCTA scanning were prospectively selected,all of whom scanned on a GE Revolution APEX-CT with a tube voltage of 80 kV,a smart mA(500-1 300 mA),a noise index of 30,and a contrast dosage of 0.5 mL/kg;four groups of images were reconstructed for each patient,FBP,50％ASIR-V,DLIR-M,and DLIR-H.The CT and SD values of the aorta(AO)root,the proximal segment of the right coronary artery(RCA),the left circumflex(LCX),the left anterior descending branch(LAD)and the pericardial fat were measured,and the signal-to-noise ratio(SNR)and contrast-to-noise ratio(CNR)were calculated.Two doctors experienced in the diagnosis of cardiovascular disease were selected to subjectively score the reconstructed images using a double-blind method.The SD value,SNR value,CNR value and subjective scores of images in the four groups were compared.Results In the objective image quality evaluation,there were statistically significant differences in SD value,SNR value and CNR value of reconstructed images in the four groups(P＜0.05).As the four groups of reconstruction algorithms FBP,50％ASIR-V,DLIR-M,and DLIR-H were changed sequentially,the image SD value gradually decreased,the SNR value and CNR value gradually increased,and the DLIR-H group had the lowest SD value and the highest SNR and CNR values.In the subjective image quality evaluation,the subjective scores of the two doctors had good consistency(Kappa value=0.900),and the difference between them was statistically significant(P＜0.001).The subjective scores of DLIR-M and DLIR-H groups were higher.Conclusion DLIR algorithm combined with"double-low"technique can significantly improve the CCTA image quality of overweight patients,which is better than 50％ASIR-V and FBP.

As a safe,cheap and effective diabetes drug,metformin has been used for many years.Diabetes increases the risk of liver cancer and affects its prognosis.In recent years,it is found that metformin reduces the pancreatic cancer risk in the treatment of diabetic patients,a large of experiments also prove that it has anti-cancer and synergistic anticancer effect.This paper focused on the effects of metformin on treatment of Ⅱ type diabetes,discussed the curative effect on liver cancer,suggested the molecular biology mechanism of inhibiting tumor,listed the latest experiment researches,analyzed the existed clinical data,proposed the further study of anticancer mechanism and clinical treatment.Metformin for a future role in prevention of hepatocellular carcinoma in patients with type Ⅱ diabetes are briefly summarized and future prospects,which in type Ⅱ diabetic patients with liver cancer in a prospective study of the effect of treatment.Mefformin for application in other cancer prevention also raises possibilities.

Objective To characterize the natural history of psoriasis vulgaris.Methods A retrospective study was carried out.Totally,245 patients admitted to hospitals within three months after the first episode of psoriasis vulgaris were selected from 1136 patients with psoriasis vulgaris who had been followed up for more than 20 years.Changes in disease severity during the long-term follow-up were traced,and information on the shape and distribution of skin lesions,family history,use of anticancer drugs,vitamins and traditional Chinese medicines was collected and analyzed.SPSS13.0 software package was utilized to assess factors associated with the evolution of psoriasis vulgaris.Results The natural course of psoriasis vulgaris could be classified into six types:immediate healing,slow healing,intermittent relapse,frequent mild relapse,frequent moderate relapse,and frequent severe relapse.The immediate healing type and slow healing type amounted to 30％ of these patients,and the frequent severe relapse type to less than 10％.Statistical analysis revealed that the clinical severity of psoriasis was associated with the age of onset and family history,and was negatively correlated with the use of anticancer drugs.Conclusions The long-term follow-up study reveals the natural course of psoriasis vulgaris,which may be helpful in guiding the prediction of prognosis,prevention of recurrence and selection of treatment.

Objective In order to explore the relationship between the annual incidence rate of psoriasis and meteorological factors. Methods An investigation was carried out using single factor correlation analysis, multiple factor regression analysis and correlation analysis among meteorological factors. Results Single factor analysis revealed that the annual incidence of psoriasis showed a significant negative correlation with the mean annual air temperature, atmospheric pressure, relative humidity and whole year rainfall, but a positive correlation with the whole year sunshine time(P

Objective To verify whether there is any difference between early-onset and late onset subtypes of psoriasis and the rationality of this classification. Methods A total of 1 632 patients with psoriasis vulgaris were included in this study. The distribution of age at onset was calculated. The disease severity at first visit and follow up, and the family history were assessed according to different ages at onset. The results were evaluated in the light of the data from a national psoriasis survey in 1984. Results There was only one peak regarding to the age at onset in psoriasis vulgaris, rather than two peaks according to our study, it was consistent with the results of the national large scale survey reported in 1984. It was found that the earlier the age at onset, the more frequently the patient had positive family history. There was some relationship between the disease severity and the age at onset, however, a clear cut age at onset by which the disease serverity could be determined was not identified in this study. Conclusion It is suggested that the early age at onset be related only to the increased possibility of family history. Its value in clinical management is not significant.