Efficacy and safety of robotic versus laparoscopic liver resection for hepatocellular carcinoma: a propensity score-matched retrospective cohort study.

BACKGROUND: Evidence concerning long-term outcome of robotic liver resection (RLR) and laparoscopic liver resection (LLR) for hepatocellular carcinoma (HCC) patients is scarce. METHODS: This study enrolled all patients who underwent RLR and LLR for resectable HCC between July 2016 and July 2021. Propensity score matching (PSM) was employed to create a 1:3 match between the RLR and LLR groups. A comprehensive collection and analysis of patient data regarding efficacy and safety have been conducted, along with the evaluation of the learning curve for RLR. RESULTS: Following PSM, a total of 341 patients were included, with 97 in the RLR group and 244 in the LLR group. RLR group demonstrated a significantly longer operative time (median [IQR], 210 [152.0-298.0] min vs. 183.5 [132.3-263.5] min; p = 0.04), with no significant differences in other perioperative and short-term postoperative outcomes. Overall survival (OS) was similar between the two groups (p = 0.43), but RLR group exhibited improved recurrence-free survival (RFS) (median of 65 months vs. 56 months, p = 0.006). The estimated 5-year OS for RLR and LLR were 74.8% (95% CI: 65.4-85.6%) and 80.7% (95% CI: 74.0-88.1%), respectively. The estimated 5-year RFS for RLR and LLR were 58.6% (95% CI: 48.6-70.6%) and 38.3% (95% CI: 26.4-55.9%), respectively. In the multivariate Cox regression analysis, RLR (HR: 0.586, 95% CI (0.393-0.874), p = 0.008) emerged as an independent predictor of reducing recurrence rates and enhanced RFS. The operative learning curve indicates that approximately after the 11th case, the learning curve of RLR stabilized and entered a proficient phase. CONCLUSIONS: OS was comparable between RLR and LLR, and while RFS was improved in the RLR group. RLR demonstrates oncological effectiveness and safety for resectable HCC.

Enhancing adversarial attacks with resize-invariant and logical ensemble.

In black-box scenarios, most transfer-based attacks usually improve the transferability of adversarial examples by optimizing the gradient calculation of the input image. Unfortunately, since the gradient information is only calculated and optimized for each pixel point in the image individually, the generated adversarial examples tend to overfit the local model and have poor transferability to the target model. To tackle the issue, we propose a resize-invariant method (RIM) and a logical ensemble transformation method (LETM) to enhance the transferability of adversarial examples. Specifically, RIM is inspired by the resize-invariant property of Deep Neural Networks (DNNs). The range of resizable pixel is first divided into multiple intervals, and then the input image is randomly resized and padded within each interval. Finally, LETM performs logical ensemble of multiple images after RIM transformation to calculate the final gradient update direction. The proposed method adequately considers the information of each pixel in the image and the surrounding pixels. The probability of duplication of image transformations is minimized and the overfitting effect of adversarial examples is effectively mitigated. Numerous experiments on the ImageNet dataset show that our approach outperforms other advanced methods and is capable of generating more transferable adversarial examples.

Clinical characteristics and corticosteroids application of different clinical types in patients with corona virus disease 2019.

To describe the epidemiological and clinical characteristics of patients with Corona Virus Disease 2019 (COVID-19) in Beijing. To analyze the application of corticosteroids in patients with severe pneumonia. We collected information on demographic characteristics, exposure history, clinical characteristics, corticosteroids use, and outcomes of the 65 confirmed cases of COVID-19 at Fifth Medical Center of PLA General Hospital from Jan 20 to Feb 23, 2020. The final follow-up date observed was April 15th, 2020. The number of patients with mild, general, severe, and critical type were 10 (15.38%), 32 (49.23%), 8 (12.31%), and 15 (23.08%), respectively. The median incubation period was 6 days. Notable outliers were 1 patient at 16 days and 1 patient at 21 days. In lymphocyte subgroup analysis, decreases in total, T, CD4, and CD8 lymphocytes were more common as the disease worsened (All P < 0.05). Methylprednisolone (mPSL) was applied to 31 (47.69%) patients with pneumonia, including 10 (31.25%) general, 8 (100%) severe, and 13 (86.67%) critical patients, respectively. Corticosteroids inhibited Interleukin-6(IL-6) production (P = 0.0215) but did not affect T lymphocyte (P = 0.0796). There was no significant difference between patients using lower dose (≤ 2 mg/kg day) and higher dose (> 2 mg/kg day) mPSL in inhibiting IL-6 production (P = 0.5856). Thirty of 31 patients (96.77%) had stopped mPSL due to improvement of pneumonia. Virus RNA clearance time lengthened with disease progression (P = 0.0001). In general type, there was no significant difference in virus clearance time between patients with (15, 12-19 days) and without (14.5, 11-18 days) (P = 0.7372) mPSL use. Lymphocyte, especially T lymphocyte, in severe and critical patients showed a dramatic decrease. Application of lower dose corticosteroids (≤ 2 mg/kg day) could inhibit IL-6 production (a representative of cytokines) as effectively as a higher dose. Proper use corticosteroids in general type patients did not delay virus clearance.