RESUMEN
Accurate segmentation of retinal images can assist ophthalmologists to determine the degree of retinopathy and diagnose other systemic diseases. However, the structure of the retina is complex, and different anatomical structures often affect the segmentation of fundus lesions. In this paper, a new segmentation strategy called a dual stream segmentation network embedded into a conditional generative adversarial network is proposed to improve the accuracy of retinal lesion segmentation. First, a dual stream encoder is proposed to utilize the capabilities of two different networks and extract more feature information. Second, a multiple level fuse block is proposed to decode the richer and more effective features from the two different parallel encoders. Third, the proposed network is further trained in a semi-supervised adversarial manner to leverage from labeled images and unlabeled images with high confident pseudo labels, which are selected by the dual stream Bayesian segmentation network. An annotation discriminator is further proposed to reduce the negativity that prediction tends to become increasingly similar to the inaccurate predictions of unlabeled images. The proposed method is cross-validated in 384 clinical fundus fluorescein angiography images and 1040 optical coherence tomography images. Compared to state-of-the-art methods, the proposed method can achieve better segmentation of retinal capillary non-perfusion region and choroidal neovascularization.
Asunto(s)
Retina , Enfermedades de la Retina , Humanos , Teorema de Bayes , Fondo de Ojo , Retina/diagnóstico por imagen , Tomografía de Coherencia ÓpticaRESUMEN
Large amounts of sewage sludge is produced from the treatment of wastewater by biological processes, which is usually treated by anaerobic digestion to produce methane gas. Acetogenesis and hydrogen are an intermediate phase during the anaerobic digestion. Batch tests of fermentative hydrogen production under different initial pH (3.0 - 12.5) were compared using the raw sludge and alkaline pretreated sludge. The influences of the characteristics and concentration of sludge were also examined thereafter. Results show that the optimal initial pH for biohydrogen production from sewage sludge was around 11.0. Under this optimal condition, the biohydrogen yield of raw sludge was 8.1 mL/g, and it would reach to 16.9 mL/g when the sludge was pretreated by alkali. Furthermore, there is no methane generation during the biohydrogen fermentation of the alkaline pretreatment sludge in 4 days and the hydrogen consumption is also slowed down. In addition, a low VSS/SS rate will reduce the hydrogen yield, while the concentrations of sludge have no obvious compact on it.
Asunto(s)
Bacterias Anaerobias/metabolismo , Fermentación , Hidrógeno/análisis , Aguas del Alcantarillado/análisis , Eliminación de Residuos Líquidos/métodos , Biodegradación Ambiental , Hidrógeno/metabolismo , Concentración de Iones de HidrógenoRESUMEN
Batch tests were carried out to analyze influences of the alkaline pretreatment and initial pH value on biohydrogen production from sewage sludge. Experimental results of the impact of different initial pH on biohydrogen production showed that both the maximal hydrogen yield occurred and that no methane was detected in the tests of at the initial pH of 11.0. The final pH decreased at the initial pH of 7.0-12.5 but increased atthe initial pH of 3.0-6.0, probably due to the combination of solubilized protein from sludge and the formation of volatile fatty acids (VFAs) and ammonia during biohydrogen fermentation. The performance of biohydrogen production using the raw sludge and the alkaline pretreated sludge was then compared in batch fermentation tests atthe initial pH of 11.0. The hydrogen yield was increased from 9.1 mL of H2/g of dry solids (DS) of the raw sludge to 16.6 mL of H2/g of DS of the alkaline pretreated sludge. No methane and less carbon dioxide (0.8% of control) were present in the biohydrogen production from the alkaline pretreated sludge. These results clearly showed that biohydrogen production could be enhanced and maintained stable by the combination of the high initial pH and alkaline pretreatment. The mechanism of biohydrogen production from sewage sludge at high initial pH was therefore investigated because the results of this study were differentfrom previous studies of biohydrogen production. Results showed that protein was the major substrate for biohydrogen production from sewage sludge and that Eubacterium multiforme and Paenibacillus polymyxa were the dominant bacteria in biohydrogen production from alkaline pretreated sludge at an initial pH of 11.0. The combination of alkaline pretreatment and high initial pH could not only maintain a suitable pH range for the growth of dominant hydrogen-producing anaerobes but also inhibit the growth of hydrogen-consuming anaerobes. In addition, the changes in pH value, oxidation-reduction potential, VFAs and soluble COD during hydrogen fermentation were also discussed.