DeepMulticut: Deep Learning of Multicut Problem for Neuron Segmentation from Electron Microscopy Volume.

Superpixel aggregation is a powerful tool for automated neuron segmentation from electron microscopy (EM) volume. However, existing graph partitioning methods for superpixel aggregation still involve two separate stages-model estimation and model solving, and therefore model error is inherent. To address this issue, we integrate the two stages and propose an end-to-end aggregation framework based on deep learning of the minimum cost multicut problem called DeepMulticut. The core challenge lies in differentiating the NPhard multicut problem, whose constraint number is exponential in the problem size. With this in mind, we resort to relaxing the combinatorial solver-the greedy additive edge contraction (GAEC)-to a continuous Soft-GAEC algorithm, whose limit is shown to be the vanilla GAEC. Such relaxation thus allows the DeepMulticut to integrate edge cost estimators, Edge-CNNs, into a differentiable multicut optimization system and allows a decision-oriented loss to feed decision quality back to the Edge-CNNs for adaptive discriminative feature learning. Hence, the model estimators, Edge-CNNs, can be trained to improve partitioning decisions directly while beyond the NP-hardness. Also, we explain the rationale behind the DeepMulticut framework from the perspective of bi-level optimization. Extensive experiments on three public EM datasets demonstrate the effectiveness of the proposed DeepMulticut.

[Clinical significance of negative fluid balance and infection management in the treatment of acute respiratory distress syndrome caused by severe novel coronavirus infection].

OBJECTIVE: To explore the clinical significance of negative fluid balance and infection management in the treatment of acute respiratory distress syndrome (ARDS) caused by severe novel coronavirus infection. METHODS: A retrospective survey was conducted. Patients with ARDS caused by severe novel coronavirus infection who were hospitalized in the department of critical care medicine of the Third Affiliated Hospital of Gansu University of Chinese Medicine and received non-invasive ventilator assisted ventilation were selected as the research objects. The fluid intake and output of all patients were accurately counted every day, and the fluid intake of the next day was adjusted according to the output of the previous day. According to the fluid negative balance, and whether the hospital infection management measures were complied with during the treatment and inspection of the patients, 45 patients with a negative fluid balance of more than 200 mL/d and strict management of nosocomial infection were taken as the observation group, and 48 patients with a negative fluid balance of less than 200 mL/d and no strict management of nosocomial infection were taken as the control group. The general data, weaning success rate, endotracheal intubation rate, mortality, as well as laboratory indicators such as white blood cell count (WBC), procalcitonin (PCT), C-reactive protein (CRP) after treatment were compared between the two groups. RESULTS: There were no significant differences in gender (male: 51.1% vs. 52.1%), age (years old: 66.31±15.92 vs. 67.50±13.59), acute physiology and chronic health evaluation II (APACHE II: 18.98±4.81 vs. 18.54±4.35) between the observation group and the control group (all P > 0.05), indicating that the baseline data were balanced and comparable. Compared with the control group, the weaning success rate of the observation group significantly increased [53.3% (24/45) vs. 31.2% (15/48), P = 0.031], endotracheal intubation rate significantly decreased [22.2% (10/45) vs. 43.8% (21/48), P = 0.028], mortality significantly reduced [20.0% (9/45) vs. 41.7% (20/48), P = 0.024], laboratory indicators WBC, PCT and CRP levels were significantly reduced [WBC (×109/L): 8.085±4.136 vs. 16.898±7.733, CRP (mg/L): 82.827±52.680 vs. 150.679±74.625, PCT (µg/L): 3.142±2.323 vs. 7.539±5.939, all P < 0.01]. CONCLUSIONS: Fluid negative balance and infection management have significant clinical significance in the treatment of severe novel coronavirus infection with ARDS.

Fear memory-associated synaptic and mitochondrial changes revealed by deep learning-based processing of electron microscopy data.

Serial section electron microscopy (ssEM) can provide comprehensive 3D ultrastructural information of the brain with exceptional computational cost. Targeted reconstruction of subcellular structures from ssEM datasets is less computationally demanding but still highly informative. We thus developed a region-CNN-based deep learning method to identify, segment, and reconstruct synapses and mitochondria to explore the structural plasticity of synapses and mitochondria in the auditory cortex of mice subjected to fear conditioning. Upon reconstructing over 135,000 mitochondria and 160,000 synapses, we find that fear conditioning significantly increases the number of mitochondria but decreases their size and promotes formation of multi-contact synapses, comprising a single axonal bouton and multiple postsynaptic sites from different dendrites. Modeling indicates that such multi-contact configuration increases the information storage capacity of new synapses by over 50%. With high accuracy and speed in reconstruction, our method yields structural and functional insight into cellular plasticity associated with fear learning.

[Application of self-made protective clothing in tracheal intubation for patients with severe respiratory infectious diseases].

OBJECTIVE: To explore the effect of self-made protective clothing in tracheal intubation for the patients with respiratory infectious diseases. METHODS: Self-made protective clothing were made by adult model plastic raincoat with sleeve lets and goggles. A prospective randomized controlled study was conducted. Patients with severe respiratory infectious diseases who needed tracheal intubation admitted to the department of intensive medicine of the Third Affiliated Hospital of Gansu University of Chinese Medicine from January 1st 2018 to March 31st 2020 were enrolled. According to the random number table method, they were divided into two groups. The control group was wearing standard protective clothing, while the test group was wearing self-made protective clothing for endotracheal intubation. The wearing time, infection rate of operators and costs of protective clothing were compared between the two groups. The patients were sprayed with trypanosome blue diluent before tracheal intubation, and the whole body of the operator was photographed with fluorescence before wearing self-made protective clothing and after doing tracheal intubation to take off the self-made protective clothing, in order to evaluate the permeability resistance of self-made protective clothing. RESULTS: A total of 86 patients were enrolled. There were 46 cases in the test group, included 28 cases of influenza A (H1N1) virus infection, 11 cases of influenza B virus infection and 7 cases of adenovirus infection. There were 40 cases in the control group, included 15 cases of H1N1 virus infection, 10 cases of influenza B virus infection, 10 cases of adenovirus infection and 5 cases of unknown pathogen. There was no significant difference in respiratory etiology between the two groups (χ2 = 3.789, P = 0.435). The wearing protective clothing time of the control group was 11.6 times than that of the test group (minutes: 22.23±1.45 vs. 1.86±0.24, χ2 = 19.023, P < 0.001). The cost of standard protective clothing was 12.5 times than that of self-made protective clothing (Yuan/set: 500 vs. 40). Fluorescent photography showed that the whole body of the operator was not stained after tracheal intubation, indicating that the protective clothing had good anti permeability and achieved the protective effect. There was no operator infection in the test group and the control group. CONCLUSIONS: Self-made protective clothing has short wearing time, low cost and equivalent isolation effect compared with standard protective clothing, which is worthy of clinical promotion.

[Effect of long-term and short-term mild hypothermia in severe traumatic brain injury: a comparative study].

OBJECTIVE: To explore the effects of different mild hypothermia therapy time on the efficacy and complications of patients with severe traumatic brain injury (STBI). METHODS: A retrospective research method was used. 132 patients with STBI given mild hypothermia therapy admitted to the Third Affiliated Hospital of Gansu University of Traditional Chinese Medicine from January 2010 to December 2018 were enrolled. According to the days of mild hypothermia therapy, the patients were divided into 2-day mild hypothermia treatment group, 5-day mild hypothermia treatment group and 14-day mild hypothermia treatment group. Glasgow coma score (GCS) after treatment of 10 days and 30 days, 30-day disability rate and mortality, coma time, prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen (Fib), D-dimer, and the incidence of complications such as decreased blood pressure, decreased gastrointestinal motility, infection, nonunion of scalp, cerebrospinal fluid leakage, etc. were compared among three groups. RESULTS: Among the 132 patients with STBI, there were 44 cases in the 2-day mild hypothermia treatment group, 45 cases in 5-day mild hypothermia treatment group, and 43 cases in 14-day mild hypothermia treatment group. There was no significant difference in gender, age, GCS score before treatment or time from injury to admission among three groups. Compared with 2-day mild hypothermia treatment group, the GCS score 10 days and 30 days after treatment in 5-day mild hypothermia treatment group and 14-day mild hypothermia treatment group were significantly higher (11.61±2.23, 10.17±2.03 vs. 6.79±1.49; 13.15±2.53, 11.24±2.24 vs. 8.79±1.59), the coma time were shorten (days: 5.79±1.89, 5.45±1.72 vs. 13.65±2.73), and 30-day disability rate and mortality were significantly decreased [13.33% (6/45), 11.63% (5/43) vs. 22.73% (10/44); 17.78% (8/45), 16.28% (7/43) vs. 31.82% (14/44)], PT and APTT were reduced obviously (s: 20.14±4.12, 22.54±3.56 vs. 30.67±5.19; 35.14±12.41, 38.59±13.54 vs. 56.67±10.62), Fib rose obviously (g/L: 1.84±0.25, 1.98±0.27 vs. 0.67±0.12), and D-dimer reduced obviously (mg/L: 53.10±19.84, 49.20±20.13 vs. 102.60±20.13), with statistically significant differences (all P < 0.05). But there was no significant difference in above indicators between the 5-day mild hypothermia treatment group and 14-day mild hypothermia treatment group (all P > 0.05). The incidence of complications in 14-day mild hypothermia group was significantly higher than those in 2-day mild hypothermia group and 5-day mild hypothermia group [decrease of blood pressure: 55.56% (20/36) vs. 36.67% (11/30), 35.14% (13/37); weakening of stomach intestinemotive power: 72.22% (26/36) vs. 46.67% (14/30), 45.95% (17/37); urethral infection: 52.78% (19/36) vs. 36.67% (11/30), 35.14% (13/37); lungs infection: 47.22% (17/36) vs. 36.67% (11/30), 37.84% (14/37); disunion of scalp: 5.56% (2/36) vs. 0% (0/30), 0% (0/37); leak of cerebrospinal fluid: 5.56% (2/36) vs. 0% (0/30), 0% (0/37), all P < 0.05], but there was no significant difference between the 2-day mild hypothermia treatment group and 5-day mild hypothermia treatment group (all P > 0.05). CONCLUSIONS: The optimal time frame for mild hypothermia treatment in patients with STBI is 5-day, which shortens the coma time, and reduces the mortality and the disability rate. The shorter mild temperature time cannot effectively prevent secondary brain injury. However, the prolonged period of mild temperature will affect the repair of the patient's injury tissue, which is not conducive to the recovery of patient's mechanical function and is prone to complications.