A reciprocal learning strategy for semisupervised medical image segmentation.

BACKGROUND: Semisupervised strategy has been utilized to alleviate issues from segmentation applications due to challenges in collecting abundant annotated segmentation masks, which is an essential prerequisite for training high-performance 3D convolutional neural networks (CNNs) . PURPOSE: Existing semisupervised segmentation methods are mainly concerned with how to generate the pseudo labels with regularization but not evaluate the quality of the pseudo labels explicitly. To alleviate this problem, we offer a simple yet effective reciprocal learning strategy for semisupervised volumetric medical image segmentation, which generates more reliable pseudo labels for the unannotated data. METHODS: Our proposed reciprocal learning is achieved through a pair of networks, one as a teacher network and the other as a student network. The student network learns from pseudo labels generated by the teacher network. In addition, the teacher network autonomously optimizes its parameters based on the reciprocal feedback signals from the student's performance on the annotated images. The efficacy of the proposed method is evaluated on three medical image data sets, including 82 pancreas computed tomography (CT) scans (training/testing: 62/20), 100 left atrium gadolinium-enhanced magnetic resonance (MR) scans (training/testing: 80/20), and 200 breast cancer MR scans (training/testing: 68/132). The comparison methods include mean teacher (MT) model, uncertainty-aware MT (UA-MT) model, shape-aware adversarial network (SASSNet), and transformation-consistent self-ensembling model (TCSM). The evaluation metrics are Dice similarity coefficient (Dice), Jaccard index (Jaccard), 95% Hausdorff distance (95HD), and average surface distance (ASD). The Wilcoxon signed-rank test is used to conduct the statistical analyses. RESULTS: By utilizing 20% labeled data and 80% unlabeled data for training, our proposed method achieves an average Dice of 84.77%/90.46%/78.53%, Jaccard of 73.71%/82.67%/69.00%, ASD of 1.58/1.90/0.57, and 95HD of 6.24/5.97/4.34 on pancreas/left atrium/breast data sets, respectively. These results outperform several cutting-edge semisupervised approaches, showing the feasibility of our method for the challenging semisupervised segmentation applications. CONCLUSIONS: The proposed reciprocal learning strategy is a general semisupervised solution and has the potential to be applied for other 3D segmentation tasks.

LSPR Sensing of Epithelial Cell Adhesion Molecules through Sphere and Cavity Plasmons of a Composite Ring Array of Poly[2-(dimethylamino)ethyl methacrylate]/Gold Nanoparticles.

Self-organization facilitates the formation of specific structures as a result of constituent interactions. In this study, the bottom of a 600 nm hole array photoresist template, which was deposited with a hydrophobic atom transfer radical polymerization (ATRP) initiator, was wetted by treatment with oxygen plasma. After the removal of the photoresist template, ring patterns of the ATRP initiator were formed at the interface between the hydrophobic and wetting regions. Poly[2-(dimethylamino)ethyl methacrylate] (PDMAEMA) was grafted from the ring array of the initiator to immobilize gold nanoparticles (AuNPs) as a uniform ring array on a silicon substrate via repeated swelling/shrinking cycles. The localized surface plasmon resonance (LSPR) peak of the PDMAEMA-AuNP hybrid ring (PAHR) red-shifted after 12 swelling/shrinking cycles. In comparison to gold nanoparticles, scalable gold nanorings can effectively develop a variety of nanostructures to design LSPR-based sensors and optimize the sensing accuracy and stability. To detect epithelial cell adhesion molecules (EpCAM) during the structural change from a ring to a disk, antiEpCAM was anchored onto the PAHR as a biosensor during swelling/shrinking. The coupling of antiEpCAM and EpCAM led to asymptotical convergence from rings to disks as well as blue shifts of the LSPR peaks. Linear correlation between the blue shift and EpCAM concentration showed a limit of detection of ∼27 pg mL-1 and a linear range of 25-200 pg mL-1 for the detection of EpCAM within 30 min. The simple method of combining lithography and plasma technology provides a versatile platform for developing the scalable ring structure of AuNPs for highly sensitive and selective biosensing.

Preparation and Characterization of Biochars from Eichornia crassipes for Cadmium Removal in Aqueous Solutions.

The study investigated the preparation and characterization of biochars from water hyacinth at 300°C to 700°C for cadmium (Cd) removal from aqueous solutions. The adsorption process was dominated by oxygen-containing functional groups with irregular surfaces via esterification reactions. Furthermore, the mineral components in the biochars also contributed to Cd absorption through precipitation. Parameters such as the effects of solution pH, contact time, and initial concentration were studied. The optimum pH value was observed at 5.0, in which nearly 90% of Cd was removed. The maximum Cd adsorption capacities based on the Langmuir isotherm were calculated at 49.837, 36.899, and 25.826 mg g(-1). The adsorption processes of the biochars followed the pseudo-second-order kinetics, with the equilibrium achieved around 5 h. The biochar from E. crassipes is a promising adsorbent for the treatment of wastewater, which can in turn convert one environmental problem to a new cleaning Technology.

Clinical outcomes of amniotic membrane loaded with 5-FU PLGA nanoparticles in experimental trabeculectomy.

AIM: To evaluate the effect of amniotic membrane loaded with 5-fluorouracil poly (lactic-co-glycolic acid) (PLGA) nanoparticles (5-FU-NPs) in the surgical outcomes of experimental trabeculectomy in rabbits. METHODS: Thirty-two New Zealand white rabbits were randomly categorized into four groups with 8 rabbits in each group. Group 1, the control group, performed traditional trabeculectomy without adjuvant treatment. While the experimental groups performed compound trabeculectomy with different implantations including amniotic membrane (group 2), 5-FU-NPs (group 3) and amniotic membrane loaded with 5-FU-NPs (group 4). Clinical evaluations including IOP measurement and filtration bleb analysis were performed in all groups postoperatively. RESULTS: There is no significant difference of mean IOP in all groups at first 7d after surgery. While at P14, mean IOPs of experimental group 2 (9.8±2.1 mm Hg), groups 3 (8.9±2.8 mm Hg) and group 4 (7.6±2.3 mm Hg) were significantly reduced compared to control group (12.4±2.6 mm Hg; n=8, P<0.05). At P21, mean IOPs of groups 3 (11.7±3.2 mm Hg) and group 4 (9.9±1.6 mm Hg) were significantly decreased compare to control group (17.9±1.6 mm Hg) and group 2 (16.6±2.8 mm Hg; n=8, P<0.05). At P28, mean IOPs of groups 3 (13.8±3.3 mm Hg) and group 4 (10.6±2.0 mm Hg) were also significantly reduced compare to control group (19.4±2.3 mm Hg) and group 2 (18.5±2.4 mm Hg; n=8, P<0.05). Meanwhile mean IOP of group 4 is significantly decreased compared to group 3 at P28 (n=8, P<0.05). Survival analysis of functional filtration blub in all groups revealed the longest survival time in group 4 (24.9±5.1d) compared to that in group 3 (20.6±4.3d), group 2 (15.0±5.2d) and control group (10.1±5.7d). CONCLUSION: Amniotic membrane loaded with 5-Fu-NPs may function as an effective anti-scarring implant and provides improved long-term surgical outcomes for experimental trabeculectomy in rabbits.

[The clinical analysis of shallow anterior chamber after complex trabeculectomy surgery].

PURPOSE: To investigate the clinical cause, therapy and prognosis of shallow anterior chamber after complex trabeculectomy surgery of glaucoma. METHODS: A retrospective study of 267 patients (302 eyes) diagnosed with glaucoma at our hospital were reviewed. RESULTS: Forty-three among 302 eyes presented shallow anterior chamber（14.24%）.The main causes of shallow anterior chamber after surgery included excessive filtering function, conjunctival exudation, choroidal detachment and ciliary circle block. These patients recovered after receiving conservative or surgical treatment. CONCLUSION: The shallow anterior chamber is a complication frequently occurring following complex trabeculectomy surgery. The occurrence of this complication may be decreased by controlling filterableness and moderately applying mitomycin C intraoperatively; The prognosis may be significantly enhanced by detecting main measurements including anterior chamber, intraocular tension, filtration condition,and by interfering in shallow anterior chamber early after complex trabeculectomy surgery.