眼底图像半自动标注系统

近年来,由于人工智能在医疗领域的高速发展,科研人员对医学图像的需求量与日俱增.这些医学图像往往需要经过精细地标注之后才能够被投入使用.与自然图像相比,医学图像的数据标注工作更具专业性、复杂性.因此,医学图像面临着标注速率低、标注成本高等问题,从而导致带标签样本稀缺的困境.眼底图像作为一种重要的医学图像,能够实现绝大多数的眼科疾病筛查与初诊工作,如糖尿病视网膜病变、青光眼等,但也同样面临着标注困难的问题.针对这样的现状,本文设计并开发了一种高效的眼底图像半自动标注系统,该系统的创新点是能够对多种眼病进行半自动标注.针对眼底图像进行多种疾病的预测,预测结果的类型包括疾病分级和病灶分割,标注人员只需对生成的预测结果进行审核并修改,这一过程可以大大降低标注人员的工作量.此外,该系统包括用户管理、项目管理、图像管理、算法模型管理4个模块.通过这4个模块可以实现团队标注中的任务分配,标注进度数据可视化,标注结果快速导出等人性化功能.该系统极大提升了标注人员的标注效率和标注体验.     

Reduction of Escherichia coli O157, Listeria monocytogenes and Salmonella enterica sv. Typhimurium populations on fresh-cut bell pepper using gaseous ozone

Bacterial contamination is the main cause of food poisoning which can lead to diarrhoea, abdominal cramp, vomiting as well as death. Bacterial contamination can potentially be controlled by application of gaseous ozone due to its antibacterial activity. Therefore, the objective of this study was to investigate the efficacy of gaseous ozone for reducing food-borne pathogens such as Escherichia coli O157, Listeria monocytogenes and Salmonella enterica sv. Typhimurium on fresh-cut bell pepper. Efficacy of gaseous ozone to reduce bacterial populations was investigated in vitro and in vivo. Results showed that optimum effect of ozone on reducing bacterial populations was achieved with short term exposures of less than 6 h. Ozone reduced the bacterial population by disrupting bacterial cell structure, which lead to cellular death. Results also showed that bacterial cells have different resistance to ozone where ozone was more effective against L. monocytogenes, followed by E. coli O157 and Salmonella Typhimurium. Optimal reduction of the bacterial population on fresh-cut bell pepper was achieved with exposure to 9 ppm ozone for 6 h. This treatment reduced 2.89, 2.56 and 3.06 log of E. coli O157, Salmonella Typhimurium and L. monocytogenes populations, respectively. In conclusion, exposure to 9 ppm ozone for 6 h helps in reducing food-borne pathogen on fresh-cut bell pepper and has high potential to be an alternative sanitization treatment to reduce pathogen population on fruit.     

High-mobility organic semiconducting crystal for direct X-ray detection

Tissue-equivalent pure organic semiconducting crystals (OSCs) have unique advantages in direct X-ray detectors (DXDs), especially for biological tissue fluoroscopy, but their low carrier mobility and inherent weak absorption restrict the limit-of-detection (LoD) and sensitivity of DXDs. High-mobility OSCs theoretically facilitate charge collection and performance leaps, however, they are rare and suffer from poor solubility due to high conjugation, leading to nearly impossible crystal growth from solution and subsequent device applications. Here we report an ingenious solution-processed crystal-growth and high-performance DXD using 2,6-diphenylanthracene (2,6-DPA), a high-mobility OSC we developed recently. In contrast to previous OSCs, high-mobility 2,6-DPA exhibits low dark current density and large photoconductive gain, resulting in record-breaking LoD of <85 nGy_air s⁻¹ and sensitivity of >10³ μC Gy_air⁻¹ cm⁻². Compared with semiconductors containing high-Z atoms (Z is atomic number), the accuracy of 2,6-DPA based DXDs is not affected by near-edge absorption effects. Moreover, the non-linear relationship between irradiation current and dose rate is confirmed as a high injection effect. High mobility and ingenious crystal growth strategy make 2,6-DPA an ideal active material for DXDs and also provide the possibility for more optoelectronic applications.