Purpose: To develop a deep learning-based Bayesian inference for MRI reconstruction.Methods: We modeled the MRI reconstruction problem with Bayes's theorem, following the recently proposed PixelCNN++ method. The image reconstruction from incomplete k-space measurement was obtained by maximizing the posterior possibility. A generative network was utilized as the image prior, which was computationally tractable, and the k-space data fidelity was enforced by using an equality constraint. The stochastic backpropagation was utilized to calculate the descent gradient in the process of maximum a posterior, and a projected subgradient method was used to impose the equality constraint. In contrast to the other deep learning reconstruction methods, the proposed one used the likelihood of prior as the training loss and the objective function in reconstruction to improve the image quality.Results: The proposed method showed an improved performance in preserving image details and reducing aliasing artifacts, compared with GRAPPA, 1 -ESPRiT, and MODL, a state-of-the-art deep learning reconstruction method. The proposed method generally achieved more than 5 dB peak signal-to-noise ratio improvement for compressed sensing and parallel imaging reconstructions compared with the other methods. Conclusion:The Bayesian inference significantly improved the reconstruction performance, compared with the conventional 1 -sparsity prior in compressed sensing reconstruction tasks. More importantly, the proposed reconstruction framework can be generalized for most MRI reconstruction scenarios.

ObjectiveThere is recent evidence that demonstrates worse oncologic outcomes associated with minimally invasive radical hysterectomy when compared with open radical hysterectomy, particularly in patients with tumors >2 cm. The aim of our study was to retrospectively evaluate the oncological outcomes between laparoscopic and open radical hysterectomy in International Federation of Gynecology and Obstetrics(FIGO) 2009 stage IB1 (FIGO 2009) cervical cancer patients with tumor size ≤2 cm.MethodsA retrospective review of medical records was performed to identify patients who underwent either laparoscopic or open radical hysterectomy during January 2010 and December 2018. Inclusion criteria were: (1) histologically confirmed cervical cancer including all histological types; (2) FIGO 2009 stage IB1; (3) tumor size ≤2 cm (determined by pelvic examination, magnetic resonance imaging or transvaginal ultrasound); (4) had undergone radical hysterectomy (type II or III) with pelvic and/or para-aortic lymphadenectomy as primary surgical treatment; (5) had follow-up information. Patients with FIGO 2009 stage IA1 or IA2, tumor size >2 cm, or who received neo-adjuvant chemotherapy before surgery, those with cervical cancer incidentally found after simple hysterectomy, or with insufficient data were excluded. Concurrent comparison between the laparoscopic and open cohorts was made for disease-free survival and overall survival.ResultsA total of 325 cervical cancer patients were included; of these, 129 patients underwent laparoscopic surgery and 196 patients had open surgery. The median follow-up times were 51.8 months (range 2–115) for laparoscopic surgery and 49.5 months (range 3–108) for open surgery. Patients in the laparoscopic group had significantly worse 5 year disease-free survival than those in the open group (90.4% vs 97.7%; p=0.02). There was no significant difference in 5 year overall survival between groups (96.9% vs 99.4%, p=0.33). The Cox proportional hazards regression analysis indicated that laparoscopic surgery was associated with lower disease-free survival compared with open surgery (adjusted hazard ratio 4.64, 95% CI 1.26 to 17.06; p=0.02). In patients with non-squamous cell carcinoma or with grade II–III, laparoscopic surgery had a significantly worse 5 year disease-free survival compared with the open surgery group (74% vs 100%, p=0.01, and 88.8% vs 98.0%, p=0.02, respectively).ConclusionLaparoscopic radical hysterectomy was associated with worse disease-free survival for stage IB1 (FIGO 2009) cervical cancer patients with tumor size ≤2 cm compared with open radical hysterectomy. Further studies may shed additional light on the impact of minimally invasive surgery in this low-risk patient population.

The improvement of social support promotes the mental health and improves the health status. The study aimed to examine the influence of the social support on symptoms of anxiety and depression among patients with silicosis and provide the scientific basis to further alleviate anxiety and depression and to monitor their whole quality of life. We investigated 324 inpatients with silicosis between April 2011 and September 2011. The HADS (the Hospital Anxiety-Depression Scale) was the major methodology used to evaluate anxiety and depression, and the MSPSS (the Multidimensional Scale of Perceived Social Support) to evaluate the social support level. Among patients with silicosis, 99.1% had anxiety symptoms, and 86.1% had depression symptoms. Meanwhile, the social support significantly influenced symptoms of anxiety and depression. The study suggested that patients with silicosis presented more anxiety and depression symptoms, while the social support levels of the patients were relatively low. The influence of social support on symptoms of anxiety and depression among patients with silicosis implied that improving the level of social support and the effective symptomatic treatment might alleviate anxiety and depression symptoms and improve physical and mental status.

The SWB and QOL of patients with silicosis are still relatively low and their health status needs improvement. At the same time, longer distances walked by patients suggest more positive influences on their SWB and QOL. It indicates that when the 6MWT cooperates with SWB and QOL, it may be able to get more accurate evaluation results of patients' survival status.

Speckle variance in ultrasound images limits the detection of low-contrast targets. In conventional compounding, multiple correlated sub-images are generated and then averaged to reduce the speckles at the cost of resolution loss. In this paper, a decorrelation procedure was applied to the correlated sub-images to further reduce speckle variance. Lesion signal-to-noise-ratio (lSNR), which combines the effect of speckle reduction and resolution loss, was used as an indicator of the detectability of lesions. The lSNR of the hyperechoic lesion in the simulated and experimental images using decorrelated compounding was increased by 122% and 89%, respectively, compared to the delay-and-sum method.