Neighbor-Guided Pseudo-Label Generation and Refinement for Single-Frame Supervised Temporal Action Localization.

Due to the sparse single-frame annotations, current Single-Frame Temporal Action Localization (SF-TAL) methods generally employ threshold-based pseudo-label generation strategies. However, these approaches suffer from inefficient data utilization, as only parts of unlabeled frames with confidence scores surpassing a predefined threshold are selected for training. Moreover, the variability of single-frame annotations and unreliable model predictions introduce pseudo-label noise. To address these challenges, we propose two strategies by using the relationship of the video segments with their neighbors': 1) temporal neighbor-guided soft pseudo-label generation (TNPG); and 2) semantic neighbor-guided pseudo-label refinement (SNPR). TNPG utilizes a local-global self-attention mechanism in a transformer encoder to capture temporal neighbor information while focusing on the whole video. Then the generated self-attention map is multiplied by the network predictions to propagate information between labeled and unlabeled frames, and produce soft pseudo-label for all segments. Despite this, label noise persists due to unreliable model predictions. To mitigate this, SNPR refines pseudo-labels based on the assumption that predictions should resemble their semantic nearest neighbors'. Specifically, we search for semantic nearest neighbors of each video segment by cosine similarity in the feature space. Then the refined soft pseudo-labels can be obtained by a weight combination of the original pseudo-label and the semantic nearest neighbors'. Finally, the model can be trained with the refined pseudo-labels, and the performance has been greatly improved. Comprehensive experimental results on different benchmarks show that we achieve state-of-the-art performances on THUMOS14, ActivityNet1.2, and ActivityNet1.3 datasets.

Weakly Supervised Temporal Action Localization With Bidirectional Semantic Consistency Constraint.

Weakly supervised temporal action localization (WTAL) aims to classify and localize temporal boundaries of actions for the video, given only video-level category labels in the training datasets. Due to the lack of boundary information during training, existing approaches formulate WTAL as a classification problem, i.e., generating the temporal class activation map (T-CAM) for localization. However, with only classification loss, the model would be suboptimized, i.e., the action-related scenes are enough to distinguish different class labels. Regarding other actions in the action-related scene (i.e., the scene same as positive actions) as co-scene actions, this suboptimized model would misclassify the co-scene actions as positive actions. To address this misclassification, we propose a simple yet efficient method, named bidirectional semantic consistency constraint (Bi-SCC), to discriminate the positive actions from co-scene actions. The proposed Bi-SCC first adopts a temporal context augmentation to generate an augmented video that breaks the correlation between positive actions and their co-scene actions in the inter-video. Then, a semantic consistency constraint (SCC) is used to enforce the predictions of the original video and augmented video to be consistent, hence suppressing the co-scene actions. However, we find that this augmented video would destroy the original temporal context. Simply applying the consistency constraint would affect the completeness of localized positive actions. Hence, we boost the SCC in a bidirectional way to suppress co-scene actions while ensuring the integrity of positive actions, by cross-supervising the original and augmented videos. Finally, our proposed Bi-SCC can be applied to current WTAL approaches and improve their performance. Experimental results show that our approach outperforms the state-of-the-art methods on THUMOS14 and ActivityNet. The code is available at https://github.com/lgzlIlIlI/BiSCC.

Multi-Hierarchical Category Supervision for Weakly-Supervised Temporal Action Localization.

Weakly Supervised Temporal Action Localization (WTAL) aims to localize action segments in untrimmed videos with only video-level category labels in the training phase. In WTAL, an action generally consists of a series of sub-actions, and different categories of actions may share the common sub-actions. However, to distinguish different categories of actions with only video-level class labels, current WTAL models tend to focus on discriminative sub-actions of the action, while ignoring those common sub-actions shared with different categories of actions. This negligence of common sub-actions would lead to the located action segments incomplete, i.e., only containing discriminative sub-actions. Different from current approaches of designing complex network architectures to explore more complete actions, in this paper, we introduce a novel supervision method named multi-hierarchical category supervision (MHCS) to find more sub-actions rather than only the discriminative ones. Specifically, action categories sharing similar sub-actions will be constructed as super-classes through hierarchical clustering. Hence, training with the new generated super-classes would encourage the model to pay more attention to the common sub-actions, which are ignored training with the original classes. Furthermore, our proposed MHCS is model-agnostic and non-intrusive, which can be directly applied to existing methods without changing their structures. Through extensive experiments, we verify that our supervision method can improve the performance of four state-of-the-art WTAL methods on three public datasets: THUMOS14, ActivityNet1.2, and ActivityNet1.3.

ADNEX Model-Based Diagnosis of Ovarian Cancer Using MRI Images.

This exploration aims to investigate the important role of magnetic resonance imaging (MRI) in the diagnosis of ovarian cancer under the ADNEX. From March 2017 to December 2019, 84 patients with ovarian cancer confirmed by pathological operation were selected as the research objects. The consistency of ADNEX, MRI, and ADNEX∗MRI in the diagnosis and staging of ovarian cancer was calculated separately. SPSS 26.0 statistical software was used to compare the accuracy, sensitivity, specificity, and diagnostic value of the two diagnostic methods. The results show that the accuracy and sensitivity of ADNEX are 78.6% and 93.2%, respectively. The accuracy and sensitivity of MRI are 81.2% and 89.4%, respectively. There is no significant difference between the two methods (p < 0.05). The overall consistency rates of ADNEX∗MRI, MRI diagnosis, and ADNEX for ovarian cancer staging are 94.2%, 74%, and 65.4%, respectively. There was a significant difference (p < 0.05). ADNEX∗MRI and MRI diagnosis were compared with each stage of ADNEX. There is a significant difference between the second and fourth stages (p < 0.05), and there is also a significant difference in the fourth stage (p < 0.017). It is concluded that MRI diagnosis of ovarian cancer based on ADNEX is superior to ADNEX and MRI examination alone, which provides a certain reference value for clinical staging of ovarian cancer.

Fuzi-Lizhong pill compensates hypothyroid-hypothermia via ghrelin release.

ETHNOPHARMACOLOGICAL RELEVANCE: Fuzi-Lizhong pill (FLZ) is a traditional Chinese medicine for treating patients with Spleen Yang deficient syndrome. Ghrelin, a peptide with 28 amino acid residues, plays multiple roles in thermogenesis. This study aims to explore FLZ regulating ghrelin to compensate hypothermia in rats with hypothyroid and indigestion. MATERIALS AND METHODS: In litter-matched rats, hypothermia was developed with both thyroidectomy at d1 and interscapular brown adipose (IBA) removal at d42, indigestion was induced with both high fat diet and fasting-feeding cycle from d56; the littermates with hypothermia and indigestion were administrated with FLZ from d70. Adaptive thermogenesis, thyroid hormones, metabolites, ghrelin dynamics were measured at d98. RESULTS: The results showed that plasma ghrelin levels were inversely correlated with the gastric ghrelin levels and adaptive thermogenesis in rats undergone both thyroidectomy and IBA removal. Fatty diet and FLZ enhanced the increase of plasma ghrelin of hypothyroid rats. These were supported by the changes of plasma thyroid related hormones, plasma metabolites, gastric ghrelin mRNA and protein, and the effects of fatty diet or FLZ. CONCLUSIONS: Our results suggest that more ghrelin release compensate chronic hypothermia in rats with both hypothyroidism and indigestion. It could explain the mechanisms of FLZ in relieving chronic hypothermia.

Tetramethylpyrazine reduces cellular inflammatory response following permanent focal cerebral ischemia in rats.

Tetramethylpyrazine (TMP) has been used to treat ischemic stroke. However, scientific evidence related to its effectiveness or precise modes of neuroprotective action is largely unclear. This study provides evidence of an alternative target for TMP and sheds light on the mechanism of its physiological benefits. We report a global inhibitory effect of TMP on intracerebral cellular inflammatory response in a rat model of permanent cerebral ischemia. TMP exhibited a neuroprotective effect against ischemic deficits by reduction of behavioral disturbance, brain infarction, and edema. The results of immunohistochemistry, enzymatic assay, Western blot, real-time reverse transcriptase-polymerase chain reaction (RT-PCR), and flow cytometric analysis revealed that TMP reduced the percentages of activated macrophages/microglia and infiltrative lymphocytes, neutrophils, and macrophages and pro-inflammatory cytokine expression after cerebral ischemia. In parallel with these immunosuppressive phenomena, TMP also attenuated the activities of ischemia-induced inflammation-associated signaling molecules and transcription factors. Another finding in this study was that the anti-inflammatory and neuroprotective effects of TMP were accompanied by a further elevated expression of NF-E2-related factor 2 (Nrf2) and heme oxygenase-1 (HO-1) in ipsilateral neurons and macrophages/microglia after cerebral ischemia. Taken together, our results suggest that both the promotion of endogenous defense capacity and the attenuation of the extent and composition percentage of the major cellular inflammatory responses via targeting of macrophages/microglia by elevating Nrf2/HO-1 expression might actively contribute to TMP-mediated neuroprotection against cerebral ischemia.

Underlying mechanism of aconitum lizhong acting on experimental hypothermia with indigestion in rats: role of ghrelin.

This study is aimed to investigate the Aconitum Lizhong pill (ALZ) pharmacological actions on hypothermia with indigestion, especially the ghrelin roles. The littermate-matched rats were randomly divided into four groups. Control did sham operation or standard diet, Model carried out interscapular brown adipose (IBA) removal with standard diet, Fat-diet did IBA removal with fat-diet, and ALZ did IBA removal and fat-diet with 4.536 g/kg/d ALZ. The potency of adaptive thermogenesis, ghrelin levels in plasma or gastric mucosa, thyroid hormones and metabolite in sera, expression of ghrelin mRNA, and protein in gastric mucous membrane were determined. ALZ relieved the hypothermia processes with indigestion, via inhibiting ghrelin expression and increasing ghrelin secretion; the dynamics from the therapy is supported with the energy changes as less body weight loss, less plasma lipid decrease, more plasma T(3) or T(4) increase with TSH decrease, and more compensation of thermogenic AUC decrease. Ghrelin played key roles in the actions of ALZ on the hypothermia with indigestion. The pharmacological mechanisms of ALZ involved the homeostasis of ghrelin expression and secretion.

[Study on the mechanism of electroacupuncture of "Neiguan" (PC 6) in inhibiting ventricular arrhythmia induced by hypothalamic electrostimulation in the rabbit].

OBJECTIVE: To research whether opiate receptors in the hypothalamic arcuate nuclei (ARC) participate in the inhibitory effect of electroacupuncture (EA) of "Neiguan" (PC 6) on ventricular arrhythmia in the rabbit. METHODS: This research consists of three parts. 1) Effect of electrolytic lesion of hypothalamic ARC on EA-induced suppression of heart ventricular extrasystole (HVE): HVE model was established by electric stimulation (0.4-0.8 mA, wave-duration 0.5 ms, 50 Hz, 5-10 s every 5 min) of hypothalamic midline region (R/L: 0.5-1.5 mm; H: -5 - -4.5 mm) in 4 rabbits. Electrolytic lesion (5 mA, 20 s) was induced by delivering a direct current to ARC region (A 1, L/R: 0.5 mm, H: -7 mm) through a stainless steel electrode (anode). EA (0.25 mA, 5 Hz, pulse-duration 0.5 ms) was applied to "Neiguan" (PC 6) for 3 min, 4 times altogether. (2) Effect of microinjection of morphine in ARC area on HVE: After inducing HVE, morphine (2 microg/microl) was injected into hypothalamic ARC region, and changes of HVE number were observed in 12 rabbits (6 for morphine group and the other 6 for normal saline group). (3) Effect of ARC microinjection of naloxone on EA-induced suppression of HVE: After inducing stable HVE, microinjection of naloxone (2 microg/microl) or same dose of normal saline was performed into ARC region in 6 rabbits respectively, followed by EA of PC6. RESULTS: Electric stimulation of hypothalamic midline region could induce continuous HVE. After EA, it decreased significantly in the occurrence rate (P<0.01). Following ARC region lesion, the effect of EA in inhibiting HVE disappeared. After microinjection of morphine into ARC region, the occurrence rate of HVE reduced apparently (P<0.01), while after microinjection of naloxone, the inhibitory effect of EA vanished. No significant changes were found in the occurrence rate and the inhibitory rate of HVE after microinjection of naloxone or normal saline in the ARC region (P>0.05). CONCLUSION: Hypothalamic ARC and it's opium receptors participate in the inhibitory effect of EA on hypothalamic midline region electro-stimulation-induced HVE in the rabbit.

[Study on effect of Tiaomaiyin injection on experimental arrhythmia].

OBJECTIVE: To study the effect of Tiaomaiyin injection on the experimental arrhythmia for analyzing its underlying mechanism in the treatment of cardiovascular disease. METHOD: Experimental animals anesthetized with 20% urethane (6 mL x kg(-1)) were evenly randomized into control group, positive control group, low-dose and high-dose Tiaomaiyin group. The rate of ventricular fibrillation (VF) chloroform-induced in mice, and the epoch of ventricular extrasystole (VE), ventricular tachycardia (VT),VF and cardiac arrest (CA), actonitine-induced in rats (1.0 microg x mL(-1) x min(-1)), and vabain-induced in guinea pigs (10 microg x mL(-1) x min(-1)), were detected respectively. The result loas converted into cumulative dosage of actonitine or vabain. In ischemia-reperfusion model in rats, the duration of arrhythmia and activity of superoxide dismutase (SOD) and malondialdehyde (MDA) were detected. RESULT: After venous injection of Tiaomaiyin, VF in mice was lower significantly (P < 0.01), VE, VT, VF in rats and VF in guinea pigs were lowered considerably (P <0.05). The duration of arrhythmia in ischemia-reperfusion model was reduced considerably (P < 0.05), and the activity of myocardial SOD was raised significantly (P <0.01). CONCLUSION: Tiaomaiyin shows the reduction of experimental arrhythmia and protect effect to ischemia-reperfusion injury of the heart, which indicates that the effect mechanism may have the relationship with inhabition of lipid peroxidation and damnification of the free radical.

[Effects of meteorological factors and mineral elements on the content of resveratrol in Polygonum cuspidatum rootstalk].

By determining monthly the content of resveratrol and eight mineral elements (Ca, Cu, Fe, K, Mg, Na, Zn, Mn) in the rootstalk of Polygonum cuspidatum, by considering the dynamic meteorological factors and by analyzing the principal components, we are investigating the dynamic relationship laws among mineral elements, meteorological factors and the content of resveratrol in Polygonum cuspidatum rootstalk in varied seasons. The results indicated that the correlation coefficient between Cu and Fe; between Mg and K; between Mg and Zn; between K and Zn; between Ca and average sunlight hours per month are 0.812, 0.871, 0.793, 0.729, 0.602. Their correlation is markedly in the positive 0.01 levels. The correlation coefficient between Cu and average air temperature in a month; between Mn and average air temperature in a month; between Fe and K are -0.738, -0.712, -0.766. Their correlation is markedly in negative 0.01 levels. The correlation between Cu and the content of resveratrol is markedly in negative 0.05 levels. Of the five principal factors affecting the content of resveratrol, two were found to be principal: "mineral element factors of promoting enzyme reacts" and "the meteorological factors of high temperature and strong shines". The percentage results of these two principal are 33.539% and 33.358%, making a total of 68.897%. The two factors above are also the main reasons for the variation of Polygonum cuspidatum resveratrol. A regression equation was set up through correlation and regression analysis between the changing content of resveratrol and the principal component.