Accurate identification of the glomerular ultrastructure is critical for the diagnosis of chronic kidney diseases,but the high cost of acquiring high-quality annotated data limits the application of fully-supervised learning.Therefore,a multi-class semi-supervised semantic segmentation framework based on segment anything model(MC4S-SAM)is proposed.After improving the mask decoder of segment anything model to enable multi-class semantic segmentation without requiring prompt information,the improved model is used to generate and refine pseudo-labels through a self-training strategy,and multi-level consistency regularization constraints are incorporated to enhance the model's performance.Experimental results show that,in the task of segmenting the glomerular mesangial ultrastructure,MC4S-SAM outperformes the fully-supervised model by 11.72%in mean intersection over union(mIoU)and 11.45%in mean Dice similarity coefficient(mDSC)when the labeled data accountes for 1/16 of the total.When the labeled data proportion is 1/4,the mIoU and mDSC reach 68.91%and 78.73%,respectively,demonstrating its significant potential for aiding the diagnosis of chronic kidney diseases.

Accurate identification of the glomerular ultrastructure is critical for the diagnosis of chronic kidney diseases,but the high cost of acquiring high-quality annotated data limits the application of fully-supervised learning.Therefore,a multi-class semi-supervised semantic segmentation framework based on segment anything model(MC4S-SAM)is proposed.After improving the mask decoder of segment anything model to enable multi-class semantic segmentation without requiring prompt information,the improved model is used to generate and refine pseudo-labels through a self-training strategy,and multi-level consistency regularization constraints are incorporated to enhance the model's performance.Experimental results show that,in the task of segmenting the glomerular mesangial ultrastructure,MC4S-SAM outperformes the fully-supervised model by 11.72%in mean intersection over union(mIoU)and 11.45%in mean Dice similarity coefficient(mDSC)when the labeled data accountes for 1/16 of the total.When the labeled data proportion is 1/4,the mIoU and mDSC reach 68.91%and 78.73%,respectively,demonstrating its significant potential for aiding the diagnosis of chronic kidney diseases.

Objective To develop a multi-modal deep learning method for automatic classification of immune-mediated glomerular diseases based on images of optical microscopy(OM),immunofluorescence microscopy(IM),and transmission electron microscopy(TEM).Methods We retrospectively collected the pathological images from 273 patients and constructed a multi-modal multi-instance model for classification of 3 immune-mediated glomerular diseases,namely immunoglobulin A nephropathy(IgAN),membranous nephropathy(MN),and lupus nephritis(LN).This model adopts an instance-level multi-instance learning(I-MIL)method to select the TEM images for multi-modal feature fusion with the OM images and IM images of the same patient.By comparing this model with unimodal and bimodal models,we explored different combinations of the 3 modalities and the optimal methods for modal feature fusion.Results The multi-modal multi-instance model combining OM,IM,and TEM images had a disease classification accuracy of(88.34±2.12)%,superior to that of the optimal unimodal model[(87.08±4.25)%]and that of the optimal bimodal model[(87.92±3.06)%].Conclusion This multi-modal multi-instance model based on OM,IM,and TEM images can achieve automatic classification of immune-mediated glomerular diseases with a good classification accuracy.

ObjectiveBased on the experience of traditional quality evaluation， the quality of Atractylodis Macrocephalae Rhizoma（AMR） with different production methods such as direct seeding， transplanting after seedling raising， topping and non-topping， and difference in growth years was compared. MethodVernier caliper was used to measure the trait data of AMR in different production methods. Paraffin sections of AMR with different production methods were made by saffron solid green staining， and the microstructure was observed. The contents of water-soluble and alcohol-soluble extracts in AMR with different production methods were determined according to the 2020 edition of Chinese Pharmacopoeia. The content of water-soluble total polysaccharides in AMR with different production methods was detected by sulfuric acid-anthrone method. Fiber analyzer was used to detect the content of fiber components in AMR with different production methods. The contents of monosaccharides， oligosaccharides and some secondary metabolites in AMR with different production methods were detected by ultra performance liquid chromatography（UPLC）， and the differences of chemical components were compared by multivariate statistical analysis methods such as principal component analysis（PCA） and partial least squares-discriminant analysis（PLS-DA）. ResultIn terms of traits， the 3-year-old AMR with direct seeding and without topping was close to the high-quality AMR with "phoenix-head and crane-neck， strong sweetness and clear aroma" recorded in ancient materia medica， followed by the 3-year-old AMR with topping after transplanting， while the 2-year-old AMR with topping after transplanting with high market circulation rate was generally fat and strong with mild odor. In the microscopic aspect， the arrangement of xylem vessels and fiber bundles in the 3-year-old samples formed two obvious rings. Compared with the 2-year-old samples cultivated in Bozhou and Zhejiang， the 3-year-old samples without topping after transplanting had more wood fibers. In terms of chemical composition， the contents of 70% ethanol extract， fructose， glucose， sucrose， 1-kestose， atractylenolide Ⅰ， chlorogenic acid， neochlorogenic acid， cryptochlorogenic acid and other components in 3-year-old AMR with direct seeding and without topping were significantly higher than those in the other three samples（P<0.05）. The contents of cellulose， 70% ethanol extract， sucrose， atractylenolide Ⅰ， atractylone and other components in 3-year-old AMR with topping after transplanting were significantly higher than those in the 2-year-old AMR with high market circulation rate（P<0.05）， while the contents of water-soluble extract and water-soluble total polysaccharides in 2-year-old samples with topping after transplanting were significantly higher than those in the 3-year-old AMR with topping after transplanting， direct seeding and without topping（P<0.05）. ConclusionUnder the current mainstream production mode， too much manual intervention makes AMR heavily enriched in polysaccharides and increased the yield， but the accumulation of sweet substances， fragrant substances and fiber substances is insufficient， which affects its quality. The current quality standard of AMR has some shortcomings in guiding the high quality production of it， it is suggested to revise the quality standard of AMR， supplement the quantitative analysis of secondary metabolites， and strengthen the production of imitation wild AMR.

Objective To develop a multi-modal deep learning method for automatic classification of immune-mediated glomerular diseases based on images of optical microscopy(OM),immunofluorescence microscopy(IM),and transmission electron microscopy(TEM).Methods We retrospectively collected the pathological images from 273 patients and constructed a multi-modal multi-instance model for classification of 3 immune-mediated glomerular diseases,namely immunoglobulin A nephropathy(IgAN),membranous nephropathy(MN),and lupus nephritis(LN).This model adopts an instance-level multi-instance learning(I-MIL)method to select the TEM images for multi-modal feature fusion with the OM images and IM images of the same patient.By comparing this model with unimodal and bimodal models,we explored different combinations of the 3 modalities and the optimal methods for modal feature fusion.Results The multi-modal multi-instance model combining OM,IM,and TEM images had a disease classification accuracy of(88.34±2.12)%,superior to that of the optimal unimodal model[(87.08±4.25)%]and that of the optimal bimodal model[(87.92±3.06)%].Conclusion This multi-modal multi-instance model based on OM,IM,and TEM images can achieve automatic classification of immune-mediated glomerular diseases with a good classification accuracy.

OBJECTIVE: We propose a novel region- level self-supervised contrastive learning method USRegCon (ultrastructural region contrast) based on the semantic similarity of ultrastructures to improve the performance of the model for glomerular ultrastructure segmentation on electron microscope images. METHODS: USRegCon used a large amount of unlabeled data for pre- training of the model in 3 steps: (1) The model encoded and decoded the ultrastructural information in the image and adaptively divided the image into multiple regions based on the semantic similarity of the ultrastructures; (2) Based on the divided regions, the first-order grayscale region representations and deep semantic region representations of each region were extracted by region pooling operation; (3) For the first-order grayscale region representations, a grayscale loss function was proposed to minimize the grayscale difference within regions and maximize the difference between regions. For deep semantic region representations, a semantic loss function was introduced to maximize the similarity of positive region pairs and the difference of negative region pairs in the representation space. These two loss functions were jointly used for pre-training of the model. RESULTS: In the segmentation task for 3 ultrastructures of the glomerular filtration barrier based on the private dataset GlomEM, USRegCon achieved promising segmentation results for basement membrane, endothelial cells, and podocytes, with Dice coefficients of (85.69 ± 0.13)%, (74.59 ± 0.13)%, and (78.57 ± 0.16)%, respectively, demonstrating a good performance of the model superior to many existing image-level, pixel-level, and region-level self-supervised contrastive learning methods and close to the fully- supervised pre-training method based on the large- scale labeled dataset ImageNet. CONCLUSION USRegCon facilitates the model to learn beneficial region representations from large amounts of unlabeled data to overcome the scarcity of labeled data and improves the deep model performance for glomerular ultrastructure recognition and boundary segmentation.
Humans ; Electrons ; Endothelial Cells ; Learning ; Podocytes ; Kidney Diseases

Objective:To evaluate the behaviors and eye movement data of the crew in different interface layouts, so as to provide a reference for the interface design of full ocean depth manned submersible.Methods:To simulate the visual stipulations of 7 interface areas in full ocean depth manned submersible by E-prime software, then to record and analyze the correct rates and reaction time of the subjects, and continuously record their eye movement data, i. e. visual dwell time, visual intake frequency, average time of visual intake, average pupil diameter, saccade amplitude, average saccade acceleration, and average speed of saccade.Results:The interface areas were ranked according to the correct rates from the highest to the lowest: 3#, 6#, 7#, 2#, 1#, 4#, 5#. The interface areas were ordered according to the reaction time from the shortest to the longest: 3#, 2#, 4#, 6#, 7#, 1#, 5#. The visual dwell time of 3# and 4# areas were significantly longer than others with statistical significance ( P<0.05 or P<0.01). The visual intake frequency of 3# and 4# areas were significantly higher than others with statistical significance ( P<0.05 or P<0.01). The average time of visual intake of 3# area was significantly longer than others with statistical significance ( P<0.05 or P<0.01). The average pupil diameter of 6# and 7# areas were significantly smaller than others, except 3# area, with statistical significance ( P<0.05 or P<0.01). There was no statistical difference in the comparison of saccade amplitude, average saccade acceleration, and average speed of saccade among all the areas ( P>0.05). Conclusion:When designing the man-machine interface in full ocean depth manned submersible, the key information should be displayed in 3# and 4# areas, primary information should be displayed in 1#, 2#, and 5# areas, and the secondary information should be displayed in 6# and 7# areas.

Objective:To discuss the cabin design of full ocean depth manned submersible by analyzing the changes of blood flow and surface electromyogram of lower limbs of divers at different crew positions after maintaining irregular posture for long time.Methods:This study focused on the sitting postures of main driver, co-driver, and observer. For all sitting postures of different divers, scheme 1 has baffle support, while scheme 2 has no baffle support. The duration of single test on main driver sitting posture was 2 hours, while that on co-driver and observer sitting posture was 1 hour. Lower limb blood flow and the surface electromyography of trapezius, erector spine and rectus femoris were measured in different divers respectively.Results:Compared with those of main driver in resting posture, the lower limb blood flow of both two schemes showed increasing tendency; and the lower limb blood flow of the scheme 2 showed decreasing tendency as compared with that of the scheme 1. Compared with those of co-driver in resting posture, the lower limb blood flow of both two schemes of co-driver sitting posture and observer sitting posture showed significant increase ( P<0.01). The mean frequency of erector spine muscle of the main driver in sitting posture of scheme 2 at 2 h was significantly higher than that of the scheme 1 ( P<0.05). Compared with those of co-driver in resting posture, the mean frequencies and mean power of trapezius, erector spine, and rectus femoris of the co-driver and observer in sitting position of both two schemes showed significant increase ( P<0.05 or P<0.01). Compared with those of the co-driver and observer in sitting posture of the scheme 1, the mean frequencies and mean power of rectus femoris of the schemes 2 showed increasing tendency. Conclusion:If retractable movable baffles could be installed in the full ocean depth manned submersible cabin, they could meet the needs of the physiological characteristics of the co-driver and observer so as to reduce body fatigue and improve operational performance.

Objective:To evaluate the behaviors and eye movement data of the crew in different interface layouts, so as to provide a reference for the interface design of full ocean depth manned submersible.Methods:To simulate the visual stipulations of 7 interface areas in full ocean depth manned submersible by E-prime software, then to record and analyze the correct rates and reaction time of the subjects, and continuously record their eye movement data, i. e. visual dwell time, visual intake frequency, average time of visual intake, average pupil diameter, saccade amplitude, average saccade acceleration, and average speed of saccade.Results:The interface areas were ranked according to the correct rates from the highest to the lowest: 3#, 6#, 7#, 2#, 1#, 4#, 5#. The interface areas were ordered according to the reaction time from the shortest to the longest: 3#, 2#, 4#, 6#, 7#, 1#, 5#. The visual dwell time of 3# and 4# areas were significantly longer than others with statistical significance ( P<0.05 or P<0.01). The visual intake frequency of 3# and 4# areas were significantly higher than others with statistical significance ( P<0.05 or P<0.01). The average time of visual intake of 3# area was significantly longer than others with statistical significance ( P<0.05 or P<0.01). The average pupil diameter of 6# and 7# areas were significantly smaller than others, except 3# area, with statistical significance ( P<0.05 or P<0.01). There was no statistical difference in the comparison of saccade amplitude, average saccade acceleration, and average speed of saccade among all the areas ( P>0.05). Conclusion:When designing the man-machine interface in full ocean depth manned submersible, the key information should be displayed in 3# and 4# areas, primary information should be displayed in 1#, 2#, and 5# areas, and the secondary information should be displayed in 6# and 7# areas.

Objective:To discuss the cabin design of full ocean depth manned submersible by analyzing the changes of blood flow and surface electromyogram of lower limbs of divers at different crew positions after maintaining irregular posture for long time.Methods:This study focused on the sitting postures of main driver, co-driver, and observer. For all sitting postures of different divers, scheme 1 has baffle support, while scheme 2 has no baffle support. The duration of single test on main driver sitting posture was 2 hours, while that on co-driver and observer sitting posture was 1 hour. Lower limb blood flow and the surface electromyography of trapezius, erector spine and rectus femoris were measured in different divers respectively.Results:Compared with those of main driver in resting posture, the lower limb blood flow of both two schemes showed increasing tendency; and the lower limb blood flow of the scheme 2 showed decreasing tendency as compared with that of the scheme 1. Compared with those of co-driver in resting posture, the lower limb blood flow of both two schemes of co-driver sitting posture and observer sitting posture showed significant increase ( P<0.01). The mean frequency of erector spine muscle of the main driver in sitting posture of scheme 2 at 2 h was significantly higher than that of the scheme 1 ( P<0.05). Compared with those of co-driver in resting posture, the mean frequencies and mean power of trapezius, erector spine, and rectus femoris of the co-driver and observer in sitting position of both two schemes showed significant increase ( P<0.05 or P<0.01). Compared with those of the co-driver and observer in sitting posture of the scheme 1, the mean frequencies and mean power of rectus femoris of the schemes 2 showed increasing tendency. Conclusion:If retractable movable baffles could be installed in the full ocean depth manned submersible cabin, they could meet the needs of the physiological characteristics of the co-driver and observer so as to reduce body fatigue and improve operational performance.