Robust-Deep: A Method for Increasing Brain Imaging Datasets to Improve Deep Learning Models’ Performance and Robustness

A small dataset commonly affects generalization, robustness, and overall performance of deep neural networks (DNNs) in medical imaging research. Since gathering large clinical databases is always difficult, we proposed an analytical method for producing a large realistic/diverse dataset. Clinical brain PET/CT/MR images including full-dose (FD), low-dose (LD) corresponding to only 5 % of events acquired in the FD scan, non-attenuated correction (NAC) and CT-based measured attenuation correction (MAC) PET images, CT images and T1 and T2 MR sequences of 35 patients were included. All images were registered to the Montreal Neurological Institute (MNI) template. Laplacian blending was used to make a natural presentation using information in the frequency domain of images from two separate patients, as well as the blending mask. This classical technique from the computer vision and image processing communities is still widely used and unlike modern DNNs, does not require the availability of training data. A modified ResNet DNN was implemented to evaluate four image-to-image translation tasks, including LD to FD, LD+MR to FD, NAC to MAC, and MRI to CT, with and without using the synthesized images. Quantitative analysis using established metrics, including the peak signal-to-noise ratio (PSNR), structural similarity index metric (SSIM), and joint histogram analysis was performed for quantitative evaluation. The quantitative comparison between the registered small dataset containing 35 patients and the large dataset containing 350 synthesized plus 35 real dataset demonstrated improvement of the RMSE and SSIM by 29% and 8% for LD to FD, 40% and 7% for LD+MRI to FD, 16% and 8% for NAC to MAC, and 24% and 11% for MRI to CT mapping task, respectively. The qualitative/quantitative analysis demonstrated that the proposed model improved the performance of all four DNN models through producing images of higher quality and lower quantitative bias and variance compared to reference images.     

Correlation of vitreous chamber depth with ocular biometry in high axial myopia

Purpose:The proportion of axial length (AL) occupied by vitreous chamber depth (VCD), or VCD:AL, consistently correlates to ocular biometry in the general population. Relation of VCD:AL to ocular biometry in high myopia is not known. The purpose of this study is to evaluate the relation of VCD and VCD:AL to ocular biometry of highly myopic eyes.Methods:This was a cross-sectional retrospective study of records of 214 myopic eyes (<−1 D SE, aged 20–40 years) attending the refractive surgery services. High axial myopia was defined as AL >26.5 mm. Eyes with posterior staphyloma and myopic maculopathy were excluded. Records were assessed for measurements of AL, central corneal thickness (CCT), anterior chamber depth (ACD), lens thickness (LT), white to white diameter (WTW), and vitreous chamber depth (VCD). Groups were formed based on increasing AL, while the sum of CCT, ACD, and LT was recorded as anterior segment depth (AS). The main outcome measure was the correlation of VCD and VCD:AL to ocular biometry. A comparison was also performed based on of degree of axial myopia.Results:Mean age of the patients was 27.0 ± 5.2 years. VCD showed a very strong correlation with AL (R = 0.98, P < 0.001) but did not correlate to any anterior parameter. VCD:AL showed moderate negative relation with AS (R = −0.43, P < 0.001) and ACD (R = −0.3, P < 0.001), while it had a weakly negative relation with LT (R = −0.18, P = 0.006). VCD:AL showed strong negative relation (R > ~0.7) with AS in all individual groups of AL. Among anterior parameters, WTW showed the most consistent relation with ocular biometry.Conclusion:VCD:AL is a better correlate of ocular biometry in high myopia as compared to VCD. However, the correlation is weaker than that noted by previous studies done on the general population. Longitudinal studies of VCD:AL in the younger age group is recommended.     

三种充填技术根管封闭能力的研究

目的:利用葡萄糖定量法检测根管微渗漏模型,评价侧向加压充填技术、垂直加压充填技术及固核载体充填技术对根管微渗漏的影响.方法:选取71颗直根管上前牙,根管预备后随机分为7组,A组(14颗)用热牙胶垂直加压加AH Plus充填;B组(8颗)仅用热牙胶垂直加压充填;C组(13颗)用侧方加压加AH Plus充填;D组(8颗)仅用侧方加压充填技术;E组(11颗)用Thermafil加AH Plus充填;F组(7颗)仅用Thermafil充填;G组(10颗)为阴性对照组,用粘蜡充填根管.牙根外表面、根管口及根尖孔处涂二层指甲油.于第7 d、14 d、21 d、28 d用葡萄糖氧化酶法检测从冠方向根方漏出的葡萄糖量.结果:采用重复测量方差分析,发现第7 d、14 d、21 d、28 d A、C组及C、E组微渗漏值有显著性差异(P＜0.001),但A、E组无显著性差异(P＞0.05).各时间点C、D组比较有显著性差异(P＜0.001),但A、B组与E、F组比较无显著性差异(P＞0.05).第7 d、14 d及第21 d、28 d各组内比较无显著性差异(P＞0.05),但各组第14 d与第21 d比较有显著性差异(P＜0.05).结论:侧方加压充填技术封闭根管的能力优于垂直加压充填技术及固核载体充填技术,而后两种充填技术封闭根管的能力相当.     

Optimized 2-methoxyestradiol invasomes fortified with apamin: a promising approach for suppression of A549 lung cancer cells

Certain anticancer agents selectively target the nucleus of cancer cells. One such drug is 2-methoxyestradiol (2ME), which is used for treating lung cancer. To improve the therapeutic effectiveness of these agents, many new methods have been devised. 2ME was entrapped into the core of hydrophobic invasomes (INVA) covered with Phospholipon 90G and apamin (APA). The Box–Behnken statistical design was implemented to enhance the composition. Using Design-Expert software (Stat-Ease Inc., Minneapolis, MN), the INVA component quantities were optimized to obtain spherical particles with the smallest size, that is, a diameter of 167.8 nm. 2ME-INVA-APA significantly inhibited A549 cells and exhibited IC₅₀ of 1.15 ± 0.04 µg/mL, which is lower than raw 2ME (IC₅₀ 5.6 ± 0.2 µg/mL). Post 2ME-INVA-APA administration, a significant rise in cell death and necrosis was seen among the A549 cells compared to those treated with plain formula or 2ME alone. This effect was indicated by increased Bax expression and reduced Bcl-2 expression, as well as mitochondrial membrane potential loss. Moreover, the cell cycle analysis showed that 2ME-INVA-APA arrests the G2-M phase of the A549 cells. Additionally, it was observed that the micellar formulation of the drug increased the cell count in pre-G1, thereby exhibiting phenomenal apoptotic potential. Furthermore, it up-regulates caspase-9 and p53 and downregulates TNF-α and NF-κβ. Collectively, these findings showed that our optimized 2ME-INVA-APA could easily seep through the cell membrane and induce apoptosis in relatively low doses.     

传统与欠矫手术治疗部分调节性内斜视的比较

目的：评估传统与欠矫手术治疗部分调节性内斜视患者的效果。 方法：将25例具有部分调节性内斜视和正常AC／A的患者随机分为两组，其中13例患者接受传统手术治疗，其余12例患者接受欠矫手术治疗。传统手术是基于远距远视完全矫正所测量的偏差上，而欠矫手术比传统手术标准低20％。所有患者均进行对称性双眼内直肌后退术，术后随访6mo。 结果：术后6mo，传统组与欠矫组手术成功率分别为46％和91％（成功标准定义为通过近距和远距远视的完全矫正斜视≤8^△）。传统组和欠矫组的手术过矫率分别为54％和9％，两组间差异有统计学意义（P〈0．05）。没有残余内斜视。手术成功率或欠矫率与患者的年龄、术前等效球镜均值、术前眼偏斜量无相关性。 结论：对于具有正常AC／A的部分调节性内斜视患者，欠矫的双眼内直肌后退术成功率较高，过矫率较低。     

Superior quality chemically reduced graphene oxide for high performance EMI shielding materials

The chemical reduction process of graphene oxide combined with a mild and controllable thermal treatment under vacuum at 200 °C for 4 hours provided a cost-effective, scalable, and high-yield route for Reduced Graphene Oxide (RGO) industrial production and became a potential candidate for producing electromagnetic interference (EMI) shielding. We investigated graphite, and RGO using l-ascorbic acid and Sodium borohydride before and after thermal treatment by carefully evaluating the chemical and morphological structures. The thermally treated l-ascorbic Acid reduction route (TCRGOL) conductivity was 2.14 × 10³ S m⁻¹ and total shielding efficiency (SET) based on mass loadings per area of shielding was 94 dB with about one-tenth less graphite weight and surpassing other graphene reduction mechanisms in the frequency range of 8.2–12.4 GHz, i.e., X-band, at room temperature while being tested using the waveguide line technique. The developed treatment represents valuable progress in the path to chemical reduction using a safe reducing agent and offering superior quality RGO rarely achieved with the top-down technique, providing a high EMI shielding performance.

The developed two-step protocol offers a superior reduced graphene oxide TCRGOL quality (7 layers), and its SET was 94 dB over the X-band.     

Brief Report: Adenosine A2A receptor blockade prevents cisplatin-induced impairments in neurogenesis and cognitive function

Chemotherapy-induced cognitive impairment (CICI) has emerged as a significant medical problem without therapeutic options. Using the platinum-based chemotherapy cisplatin to model CICI, we revealed robust elevations in the adenosine A_2A receptor (A_2AR) and its downstream effectors, cAMP and CREB, by cisplatin in the adult mouse hippocampus, a critical brain structure for learning and memory. Notably, A_2AR inhibition by the Food and Drug Administration–approved A_2AR antagonist KW-6002 prevented cisplatin-induced impairments in neural progenitor proliferation and dendrite morphogenesis of adult-born neurons, while improving memory and anxiety-like behavior, without affecting tumor growth or cisplatin’s antitumor activity. Collectively, our study identifies A_2AR signaling as a key pathway that can be therapeutically targeted to prevent cisplatin-induced cognitive impairments.