Artificial intelligence (AI) applications in medical imaging have the potential to revolutionize healthcare by improving diagnostic accuracy, reducing errors, and increasing efficiency. Deep neural networks (DNNs) have achieved a great success in diagnosing various diseases, often reaching or even surpassing the performance levels of human experts. However, DNNs remain vulnerable to adversarial attacks — carefully crafted inputs that can manipulate the neural networks into incorrect predictions. Current adversarial defense techniques, such as adversarial training, are limited in their effectiveness against specific attacks and necessitate model re-training, thereby hindering their widespread application. Adversarial purification is an emerging defense method that leverages generative models to refine training data and eliminate adversarial examples before classification. This paper explores various generative models and proposes AdvPurify-GAN, a novel Generative Adversarial Network (GAN) for purifying medical images against adversarial attacks. Our work stands out as the first to utilize the adversarial purification techniques for enhancing robustness in medical image analysis. We evaluate AdvPurify-GAN on multiple datasets and show favorable performance under standard attacks compared with existing adversarial training and purification baselines.

Background:Conventional surgical skill assessment, conducted by expert surgeons in operating rooms, is often time-consuming and subjective. An automated and objective system capable of differentiating surgeon expertise levels (novice, intermediate, and expert) and providing feedback would significantly enhance surgical training. However, most existing automated frameworks prioritize deep learning performance over interpretability. Materials and methods:We propose I3D-SAP, an interpretable video-based surgical skill assessment framework of basic procedural elements using a 3D convolutional neural network. The model processes surgical videos and classifies surgeons into three expertise levels. Experiments are conducted on the public JIGSAWS dataset, evaluating performance across suturing, needle passing, and knot tying tasks. Results:Our method has achieved high accuracy: 100% (suturing), 96.3% (needle passing), and 97.2% (knot tying), comparable to the state-of-the-art models. Our framework's interpretability is demonstrated through Class Activation Map (CAM) visualization and snippet-wise analysis, offering dual perspectives to explain classification decisions. However, evaluation under a more rigorous Leave-One-User-Out (LOUO) protocol revealed a significant performance drop, and testing on an external dataset (ROSMA) resulted in near-chance-level accuracy, indicating that the current model's ability to generalize to entirely unseen surgeons is limited. Conclusion:I3D-SAP provides both high accuracy and interpretability, addressing a critical gap in automated surgical assessment of basic procedural elements. By visualizing the model decisions, it enhances transparency and trust, paving the way for practical adoption in surgical training. Moreover, the model's current generalizability challenges should be addressed through future work involving larger, more diverse datasets and techniques for surgeon-invariant feature learning before practical clinical deployment can be considered.

BACKGROUND:Surgical skill assessment relies on in-person observation by expert surgeons, which is time-consuming and subjective. Therefore, an automated objective surgical skill assessment system that could discriminate between surgeons with different levels of expertise would be helpful for surgeon training. MATERIALS AND METHODS:In this paper, we propose a multimodal surgical skill classification framework, which contains a combination of video and kinematic classifiers. We present a 3D Convolutional Neural Network (CNN)-based video classifier and a 1D CNN-based kinematic classifier. Furthermore, we apply Class Activation Map (CAM) to provide an explainability of the model in decision-making process. We conduct the experiments on the public datasets JIGSAWS and ROSMA. JIGSAWS contains 108 trials, and it is collected from 8 surgeons with different surgical skill level. ROSMA contains 206 trials, and it is recorded from 12 subjects. RESULTS:The experimental results have shown that our framework has achieved the comparable performance on the two public datasets. The proposed multimodal learning framework outperforms single-modality in most surgical tasks. CONCLUSION:This study contributes a multimodal learning approach for automatic robotic surgical skill evaluation with kinematic data and video data. It also explains the model in decision-making process using CAM.

Early brain injury (EBI) is the vital factor in determining the outcome of subarachnoid hemorrhage (SAH). Schizandrin A (Sch A), the bioactive ingredient extracted from Schisandra chinensis, has been proved to exert beneficial effects in multiple human diseases. However, the effect of Sch A on SAH remains unknown. The current study was designed to explored role and mechanism of Sch A in the pathophysiological process of EBI following SAH. A total of 74 male C57BL/6 J mice were subjected to endovascular perforation to establish the SAH model. Different dosages of Sch A were administrated post-modeling. The post-modeling assessments included neurological test, brain water content, RT-PCR, immunofluorescence, Nissl staining. Oxygenated hemoglobin was introduced into microglia to establish a SAH model in vitro. Sch A significantly alleviated SAH-induced brain edema and neurological impairment. Moreover, application of Sch A remarkably inhibited SAH-induced neuroinflammation, evidenced by the decreased microglial activation and downregulated TNF-α, IL-1β and IL-6 and expression. Additionally, Sch A, both in vivo and in vitro, protected neurons against SAH-induced inflammatory injury. Mechanismly, administration of Sch A inhibited miR-155/NF-κB axis and attenuated neuroinflammation, as well as alleviating neuronal injury. Our data suggested that Sch A could attenuated EBI following SAH via modulating neuroinflammation. The anti-inflammatory effect was exerted, at least partly through the miR-155/NF-κB axis, which may shed light on a possible therapeutic target for SAH.

Surgical skill assessment currently hinges on the manual observations of senior surgeons, and the assessment process is inherently time-consuming and subjective. Hence, there is a need to develop machine learning-based automated robotic surgical skill assessment. However, the existing machine learning-based works are only built in either the time domain or frequency domain but have never considered the investigation on the time-frequency domain. To fill the research gap, we explore the representation of the surgery motion data in the time-frequency domain. In this study, we propose a novel automated robotic surgical skill assessment framework called Continuous Wavelet Transform-Vision Transformer (CWT-ViT). We apply continuous wavelet transform, i.e., a time-frequency representation method, to convert robotic surgery kinematic data to synthesis images. Furthermore, by taking advantage of the prior knowledge of the da Vinci surgical system, we design a four branches-based architecture, each branch representing a robotic manipulator. We have conducted extensive experiments and achieved comparable results on the benchmark robotic surgical skill dataset JHU-ISI Gesture and Skill Assessment Working Set (JIGSAWS). Our proposed CWT-ViT framework has demonstrated the feasibility of applying time-frequency representation for automated robotic surgical skill assessment using kinematic data. The code is available at https://github.com/yiming95/CWT-ViT-Surgery.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

Background:Inflammatory bowel disease (IBD) is affected by interactions between intestinal microbial factors, abnormal inflammation, and an impaired intestinal mucosal barrier. Neutrophils (NE) are key players in IBD. Fusobacterium nucleatum (F. nucleatum) is reported to contribute to IBD progression. However, the relationship between F. nucleatum, abnormal inflammation, and intestinal barrier impairment should be interpreted to understand the role of F. nucleatum in IBD. Methods:Dextran sulfate sodium (DSS)-induced colitis model was established and mice were orally administered with F. nucleatum. F. nucleatum colonization was confirmed by fluorescence in situ hybridization (FISH) and PCR. Intestinal barrier impairment was investigated by tight junction protein expression. Immuno-histochemistry (IHC) for Ly6G and flow cytometry detection to measure NE chemotaxis in mouse colon tissues. Caco-2 monolayers were used to evaluate epithelial integrity and permeability in vitro. A transwell model involving caco-2 cells and NE co-culture was used to assess NE chemotaxis. NE chemokines were measured by ELISA. A mouse model of NE exhaustion using an anti-Ly6G antibody was used to identify the role of NEs in F. nucleatum-induced colitis. Transcriptome sequencing and bioinformatics analysis were applied to screen cytokines and signaling pathways. Results:Administration of F. nucleatum aggravated colitis in the DSS model. F. nucleatum infection downregulates ZO-1 and Occludin expression and increases intestinal permeability. Additionally, F. nucleatum-induced NE chemotaxis decreases the integrity and permeability of the caco-2 monolayer. F. nucleatum-induced NE chemotaxis is dependent on IEC-derived interleukin 8 (IL-8) secretion, mediated by the TLR2/ERK signaling pathway. In addition, NE exhaustion in mice inhibited F. nucleatum-induced intestinal barrier impairment and colitis. Conclusion:F. nucleatum improves NE chemotaxis by infecting intestinal epithelial cells (IECs) to secrete IL-8 and aggravate intestinal barrier impairment, contributing to the progression of intestinal inflammation. Examining and eliminating F. nucleatum could be a valuable microbiome-based method for IBD surveillance and prevention.

目的 探讨神经内镜经鼻筛蝶窦入路视柱磨除视神经管扩大减压术治疗外伤性视神经病变(TON)的可行性和治疗效果.方法 回顾性分析2018年3月至2022年10月宁波大学附属李惠利医院神经外科连续收治并行手术治疗的43例TON患者的临床资料.采用神经内镜经鼻筛蝶窦入路进行常规视神经管减压手术24例(对照组),应用神经内镜经鼻筛蝶窦入路视柱磨除视神经管扩大减压术式19例(试验组).比较两种不同术式的手术时长、术中出血量及住院时长等,术后观察患者的视力变化以评估手术的有效性,安全性评估指标为手术相关并发症的发生情况.结果 对照组的手术时长为 56.3～103.5 min[(76.6±23.4)min],试验组为 75.1～121.6 min[(93.2± 17.9)min],差异有统计学意义(t=2.55,P=0.009).两组间术中出血量及住院时长的差异均无统计学意义(均P＞0.05).对照组的术前视力:无光感6例、光感10例、眼前手动6例、眼前指数2例;术后3个月,无光感5例、光感7例、眼前手动7例、眼前指数3例、LogMAR视力表≥0.02 2例,手术有效比例为9/24.试验组术前视力:无光感4例、光感7例、眼前手动6例、眼前指数2例;术后3个月,无光感3例、光感4例、眼前手动3例、眼前指数6例、LogMAR视力表≥0.02 3例,手术有效比例为13/19.两组间手术有效比例的差异有统计学意义(x2=4.06,P=0.036).两组无一例发生脑脊液鼻漏、颈内动脉损伤及嗅觉丧失等手术相关并发症.结论 神经内镜经鼻筛蝶窦入路视柱磨除视神经管扩大减压术治疗TON在临床实践中安全、可行,并能提高对TON的临床疗效.

Artificial intelligence (AI) has been steadily developing in the medical field in the past few years, and AI-based applications have advanced cancer diagnosis. Breast cancer has a massive amount of data in oncology. There has been a high level of research enthusiasm to apply AI techniques to assist in breast cancer diagnosis and improve doctors' efficiency. However, the wise utilization of tedious breast cancer-related medical care is still challenging. Over the past few years, AI-based NLP applications have been increasingly proposed in breast cancer. In this systematic review, we conduct the review using preferred reporting items for systematic reviews and meta-analyses (PRISMA) and investigate the recent five years of literature in natural language processing (NLP)-based AI applications. This systematic review aims to uncover the recent trends in this area, close the research gap, and help doctors better understand the NLP application pipeline. We first conduct an initial literature search of 202 publications from Scopus, Web of Science, PubMed, Google Scholar, and the Association for Computational Linguistics (ACL) Anthology. Then, we screen the literature based on inclusion and exclusion criteria. Next, we categorize and analyze the advantages and disadvantages of the different machine learning models. We also discuss the current challenges, such as the lack of a public dataset. Furthermore, we suggest some promising future directions, including semi-supervised learning, active learning, and transfer learning.

In the era of big data, text-based medical data, such as electronic health records (EHR) and electronic medical records (EMR), are growing rapidly. EHR and EMR are collected from patients to record their basic information, lab tests, vital signs, clinical notes, and reports. EHR and EMR contain the helpful information to assist oncologists in computer-aided diagnosis and decision making. However, it is time consuming for doctors to extract the valuable information they need and analyze the information from the EHR and EMR data. Recently, more and more research works have applied natural language processing (NLP) techniques, i.e., rule-based, machine learning-based, and deep learning-based techniques, on the EHR and EMR data for computer-aided diagnosis in oncology. The objective of this review is to narratively review the recent progress in the area of NLP applications for computer-aided diagnosis in oncology. Moreover, we intend to reduce the research gap between artificial intelligence (AI) experts and clinical specialists to design better NLP applications. We originally identified 295 articles from the three electronic databases: PubMed, Google Scholar, and ACL Anthology; then, we removed the duplicated papers and manually screened the irrelevant papers based on the content of the abstract; finally, we included a total of 23 articles after the screening process of the literature review. Furthermore, we provided an in-depth analysis and categorized these studies into seven cancer types: breast cancer, lung cancer, liver cancer, prostate cancer, pancreatic cancer, colorectal cancer, and brain tumors. Additionally, we identified the current limitations of NLP applications on supporting the clinical practices and we suggest some promising future research directions in this paper.

Objective:To investigate the clinical characteristics and risk factors of delayed cerebral hyperperfusion syndrome (dCHS) after decompressive hemicraniectomy (DHC) for massive cerebral infarction (MCI).Methods:A retrospective analysis of 44 patients with MCI who underwent DHC at Ningbo Medical Center Lihuili Hospital from June 2016 to September 2020 was performed. The patients were divided into three groups: dCHS group (n = 10), acute cerebral hyperperfusion syndrome (aCHS) group (n = 14) and non-CHS group (n = 20), based on whether CHS occurred and when it occurred. The general data of the three groups were compared, and the risk factors related to dCHS were analyzed by multivariate Logistic regression.Results:There were significant differences in middle cerebral artery (MCA) occlusion, hemorrhagic transformation, and reoperation after hemorrhagic transformation among these three groups (χ2 = 17.720, 24.144, 18.324, all P < 0.05). The results of multivariate Logistic regression analysis showed that delayed recanalization after MCA occlusion [odds ratio (OR) = 16.750, 95% confidence interval (CI) (1.111, 252.468), P = 0.042] and hemorrhagic transformation [OR = 10.206, 95%CI (1.095, 95.123), P = 0.041] were independent risk factors for the occurrence of dCHS.Conclusion:Delayed recanalization after MCA occlusion and hemorrhagic transformation are independent risk factors for dCHS in patients with MCI undergoing DHC.

Background Both stent retriever (SR) and contact aspiration (CA) are widely used as first-line strategies for acute posterior circulation strokes (PCS). However, it is still unclear how CA and SR compare as the first-line treatment of acute PCS. Several new studies have been published recently, so we aimed to perform an updated meta-analysis. Methods The meta-analysis was conducted according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) statement. Random-effects models were performed to pool the outcomes and the value of I-2 was calculated to assess the heterogeneity. Results Ten observational studies with 1189 patients were included, among whom 492 received first-line CA and 697 received first-line SR. The pooled results revealed that first-line CA could achieve a significantly higher proportion of modified Thrombolysis In Cerebral Infarction (mTICI) 2b/3 (OR 1.90, 95% CI 1.33 to 2.71, I-2=0%), mTICI 3 (OR 1.95, 95% CI 1.15 to 3.31, I-2=59.6%), first-pass effect (OR 2.91, 95% CI 1.51 to 5.58, I-2=0%), lower incidence of new-territory embolic events (OR 0.20, 95% CI 0.05 to 0.83, I-2=0%), and shorter procedure time (mean difference -29.4 min, 95% CI -46.8 to -12.0 min, I-2=62.8%) compared with first-line SR. At 90-day follow-up, patients subjected to first-line CA showed a higher functional independence (modified Rankin Scale score 0-2; OR 1.38, 95% CI 1.01 to 1.87, I-2=23.5%) and a lower mortality (OR 0.71, 95% CI 0.50 to 1.00, p=0.050, I-2=0%) than those subjected to first-line SR. Conclusions This meta-analysis suggests that the first-line CA strategy could achieve better recanalization and clinical outcomes for acute PCS than first-line SR. Limited by the quality of included studies, this conclusion should be drawn with caution.

目的:分析单边双通道内镜(UBE)下腰椎间盘髓核摘除术用于中高度脱垂游离型腰椎间盘突出症(LDH)治疗的效果.方法:对2020年2月-2021年2月中国科学院大学宁波华美医院诊断为腰椎间盘突出中高度游离型后由同一脊柱外科医生进行单边双通道内镜下腰椎间盘髓核摘除术的患者进行回顾性分析.基于标准共纳入50例患者,中度、高度脱垂游离型LDH各有26、24例.记录手术时间、术中出血量、术后住院时间,借助疼痛视觉模拟评分(VAS)、Oswestry功能障碍指数(ODI)与改良Macnab疗效评定标准,对临床效果展开评估.结果:50例患者手术皆成功,平均手术时间为(66.2±8.1)min,术中出血<50 ml.平均术后住院时间(3.5±2.6)d.随访时间为(8.0±2.4)个月.患者术前腰痛VAS评分(5.7±1.2)分,术后1、3个月和末次随访时分别降低至(3.1±1.4)、(2.7±1.6)、(1.4±0.9)分,较术前降低,差异有统计学意义(P<0.05).患者术前ODI评分为(54.8±7.2)分,术后1、3个月和末次随访时分别降至(47.6±5.3)、(28.9±2.6)、(12.6±2.1)分,较术前明显降低(P<0.05).患者末次随访时改良Macnab标准评估结果为,优、良、可、差依次32、15、2、1例,总有效占比94％.结论:在治疗中高度脱垂游离型LDH患者方面,UBE下腰椎间盘髓核摘除术拥有更为宽广的操作视野,器械要求简单,可彻底摘除髓核,临床治疗效果理想.

Meningiomas are the most common benign intracranial tumors and frequently present with a gradual onset of neurological deficits; conversely, their acute presentation with hemorrhagic onset appears to be a rare event. Nonetheless, as early surgical evacuation is the foundation of treatment, a timely diagnosis of this rare type of intracranial hemorrhage is necessary. The purpose of the present single-center study was to investigate the radiological characteristics and propose a new bleeding classification for guiding the diagnosis and treatment. A total of 19 patients consecutively diagnosed with hemorrhagic meningioma were enrolled in this retrospective study. Intracranial extra-axial mass, tumor-associated hemorrhage and peritumoral brain edema were the three main radiological features of the hemorrhagic meningiomas. The site of tumor-associated hemorrhage included the peritumoral space, subarachnoid space, subdural space, brain parenchyma and/or intratumor region. Based on the anatomical relationship between meningioma and hematoma, the spontaneous hemorrhage stemming from meningiomas was further summarized into three bleeding patterns involving purely intratumoral hemorrhage (type I), purely extratumoral hemorrhage (type II) and combined intra/extratumoral hemorrhage (type III); furthermore, the type III hemorrhage usually came from type I bleeding that extended into the surrounding regions. The symptoms in type I patients were generally mild and early surgery was performed following adequate preoperative evaluations. The symptoms in type II patients were mild in certain cases and moderate to severe in others, so early or emergency surgery was chosen according to the clinical status of the patient. Almost all type III patients had moderate to severe symptoms and these patients usually required emergency surgery. In addition, patients with different bleeding types may have different pathological mechanisms underlying the tumor bleeding. Apart from being convenient for diagnosis, this concise and practical bleeding classification may aid in the selection of the treatment strategy and facilitate the understanding of the associated mechanisms.

Glioblastoma (GBM) is the most frequent tumor in the central nervous system. Long non-coding RNAs (lncRNAs) have been widely accepted as essential participators in cancer progression. Nonetheless, the specific role and mechanism of lncRNA SRY-box transcription factor 2 overlapping transcript (SOX2-OT) in GBM have not been studied. We evaluated expression levels of SOX2-OT, miR-192-5p and Ras-related protein Rab-2A (RAB2A) in GBM cells via qRT-PCR. To investigate the roles of SOX2-OT in GBM cells, CCK-8, JC-1, EdU, and western blot assays were performed. The connection among SOX2-OT, miR-192-5p and RAB2A in GBM cells was explored through pull down, luciferase reporter, and RIP assays. Western blot and qRT-PCR were employed to analyze the activity of extracellular-signal-regulated kinase (ERK) signaling pathway. SOX2-OT expression was higher in GBM cell lines than in normal cells. SOX2-OT knockdown repressed proliferation and promoted apoptosis of GBM cells. Mechanism assays revealed that SOX2-OT could sponge miR-192-5p. Moreover, RAB2A was certified to be the target gene of miR-192-5p. Overexpression of RAB2A reversed the repressive function of SOX2-OT knockdown on GBM cell growth. Furthermore, SOX2-OT activated ERK signaling pathway in GBM cells. SOX2-OT regulated miR-192-5p/RAB2A axis and ERK pathway to promote GBM cell growth.

A recent study has reported that lumican (LUM) is expressed at a high level in the nucleus pulposus specimens from herniated lumbar disc, without description of the specific mechanism. This study was designed to investigate the function and mechanism of LUM in intervertebral disc degeneration (IDD). In this study, human nucleus pulposus cells (hNPCs) cells were challenged with tumor necrosis factor (TNF)-α to establish the IDD in vitro model. After LUM silencing, cell viability was detected using CCK-8 kit, and the expression of inflammatory factors was evaluated using RT-qPCR and ELISA. Flow cytometry and β-galactosidase staining were used to determine cell cycle and cell senescence. The expression of cycle and senescence-related proteins was evaluated with western blotting. Then, Fas ligand (FasL) was overexpressed and proteins in apoptosis signal regulating kinase 1 (ASK1)/p38 signaling were tested. Finally, GS-4997, an inhibitor of ASK1, was used to explore the regulatory effects of LUM on ASK1/p38 signaling in TNF-α-induced hNPCs. Results indicated that LUM expression was upregulated in TNF-α-challenged hNPCs. LUM gene interference mitigated TNF-α-induced inflammatory response, cell cycle arrest, and senescence of hNPCs. It was then found that LUM silencing could inhibit ASK1/p38 signaling in TNF-α-treated hNPCs, which was reversed by FasL overexpression. Additionally, ASK1/p38 participated in the mediation by LUM of TNF-α-induced inflammation, cell cycle arrest, and senescence of hNPCs. To conclude, interference with LUM effectively mitigated TNF-α-induced inflammatory response, cell cycle arrest, and cell senescence. Further experiments showed the involvement of ASK1/p38 pathway in LUM-mediated NP cell phenotypes through FasL.

BackgroundAbdominal cerebrospinal fluid (CSF) pseudocyst is an uncommon but important complication of ventriculoperitoneal (VP) shunts. While individual articles have reported many cases of abdominal CSF pseudocyst following VP shunts, no case of a hemorrhagic abdominal pseudocyst after VP shunts has been reported so far.Case presentationThis article reports a 68-year-old woman with a 4-month history of progressive abdominal pain and distention. She denied any additional symptoms. A VP shunt was performed 15 years earlier to treat idiopathic normal pressure hydrocephalus and no other abdominal surgery was performed. Physical examination revealed an elastic palpable mass in her right lower abdomen, which was dull to percussion. Abdominal computed tomography (CT) scan indicated a large cystic collection of homogenous iso-density fluid in the right lower abdominal region with clear margins. The distal segment of the peritoneal shunt catheter was located within the cystic mass. Abdominal CSF pseudocyst was highly suspected as a diagnosis. Laparoscopic cyst drainage with removal of the whole cystic mass was performed, 15-cm cyst which found with thick walls and organized chronic hematic content. No responsible vessel for the cyst hemorrhage was identified. No further shunt revision was placed. Histological examination showed that the cyst wall consisted of outer fibrous tissue and inner granulation tissue without epithelial lining, and the cystic content was chronic hematoma. The patient had an uneventful postoperative course and remained asymptomatic for 8-mo follow-up.ConclusionTo the best of our knowledge, this is the first report of hemorrhagic onset in the abdominal pseudocyst following VP shunt. Such special condition can accelerate the appearance of clinical signs of the abdominal pseudocyst after VP shunts, and its mechanisms may be similar to the evolution of subdural effusion into chronic subdural hematoma (CSDH).