Weakly Supervised Deep Learning for Tooth-Marked Tongue Recognition.

The recognition of tooth-marked tongues has important value for clinical diagnosis of traditional Chinese medicine. Tooth-marked tongue is often related to spleen deficiency, cold dampness, sputum, effusion, and blood stasis. The clinical manifestations of patients with tooth-marked tongue include loss of appetite, borborygmus, gastric distention, and loose stool. Traditional clinical tooth-marked tongue recognition is conducted subjectively based on the doctor's visual observation, and its performance is affected by the doctor's subjectivity, experience, and environmental lighting changes. In addition, the tooth marks typically have various shapes and colors on the tongue, which make it very challenging for doctors to identify tooth marks. The existing methods based on deep learning have made great progress for tooth-marked tongue recognition, but there are still shortcomings such as requiring a large amount of manual labeling of tooth marks, inability to detect and locate the tooth marks, and not conducive to clinical diagnosis and interpretation. In this study, we propose an end-to-end deep neural network for tooth-marked tongue recognition based on weakly supervised learning. Note that the deep neural network only requires image-level annotations of tooth-marked or non-tooth marked tongues. In this method, a deep neural network is trained to classify tooth-marked tongues with the image-level annotations. Then, a weakly supervised tooth-mark detection network (WSTDN) as an architecture variant of the pre-trained deep neural network is proposed for the tooth-marked region detection. Finally, the WSTDN is re-trained and fine-tuned using only the image-level annotations to simultaneously realize the classification of the tooth-marked tongue and the positioning of the tooth-marked region. Experimental results of clinical tongue images demonstrate the superiority of the proposed method compared with previously reported deep learning methods for tooth-marked tongue recognition. The proposed tooth-marked tongue recognition model may provide important syndrome diagnosis and efficacy evaluation methods, and contribute to the understanding of ethnopharmacological mechanisms.

Artificial intelligence in tongue diagnosis: Using deep convolutional neural network for recognizing unhealthy tongue with tooth-mark.

Tongue diagnosis plays a pivotal role in traditional Chinese medicine (TCM) for thousands of years. As one of the most important tongue characteristics, tooth-marked tongue is related to spleen deficiency and can greatly contribute to the symptoms differentiation and treatment selection. Yet, the tooth-marked tongue recognition for TCM practitioners is subjective and challenging. Most of the previous studies have concentrated on subjectively selected features of the tooth-marked region and gained accuracy under 80%. In the present study, we proposed an artificial intelligence framework using deep convolutional neural network (CNN) for the recognition of tooth-marked tongue. First, we constructed relatively large datasets with 1548 tongue images captured by different equipments. Then, we used ResNet34 CNN architecture to extract features and perform classifications. The overall accuracy of the models was over 90%. Interestingly, the models can be successfully generalized to images captured by other devices with different illuminations. The good effectiveness and generalization of our framework may provide objective and convenient computer-aided tongue diagnostic method on tracking disease progression and evaluating pharmacological effect from a informatics perspective.

A diterpene derivative enhanced insulin signaling induced by high glucose level in HepG2 cells.

The predominant feature of type 2 diabetes is insulin resistance. Identifying a drug able to reduce insulin resistance is an urgent requirement. ent-3α-Formylabieta-8(14),13(15)-dien-16,12ß-olide had been identified as a new diterpene derivative which showed anticancer activity. This study explores the hypoglycemic effect of ent-3α-formylabieta-8(14),13(15)-dien-16,12ß-olide and studied its mechanism. The insulin response of HepG2 cells following ent-3α-formylabieta-8(14),13(15)-dien-16,12ß-olide treatment, as a model for liver cancer cells, was assessed. The results demonstrated that hyperglycemia resulted in a significant increase in the levels of insulin receptor substrate-1 (IRS-1) serine phosphorylation and decrease in Akt phosphorylation. High glucose also inhibited the phosphorylation of insulin-dependent GSK3ß. ent-3α-Formylabieta-8(14),13(15)-dien-16,12ß-olide treatment improved the effect of insulin on the phosphorylation of IRS-1 Ser307. In addition, this study demonstrated that the effect of ent-3α-formylabieta-8(14),13(15)-dien-16,12ß-olide was dependent on the activation of AMP-activated protein kinase. Collectively, experimental data indicated an association between insulin resistance and hyperglycemia in HepG2 cells, and that ent-3α-formylabieta-8(14),13(15)-dien-16,12ß-olide reduces IRS-1 Ser307 phosphorylation via activating AMPK, thereby decreasing the insulin signaling blockade.

[Effects of low-level laser irradiation on rat mesenteric microcirculatory disturbance during early stage of endotoxemia].

OBJECTIVE: To observe the effects of low-level laser irradiation on mesenteric microcirculation of rats in vivo in the early stage of endotoxemia (ETM). METHODS: The experimental model of ETM was reproduced by injection of lipopolysaccharide (LPS). Sixty healthy male Sprague-Dawley (SD) rats were divided into three groups used random number table: control group, LPS group and low-level laser irradiation group, each group included 20 rats which were subdivided into four temporal subgroups (1, 2, 4, 6 hours, respectively). In low-level laser irradiation group, the rats were irradiated by type SLT semiconductor laser (650 nm, 5 mW) on unilateral femoral artery and vein, and blood vessel of the ear concurrently for 30 minutes. The interference course was vertical irradiation taken at 30 minutes after the injection of LPS. At 1, 2, 4, 6 hours after the injection of LPS, changes in mesenteric microcirculation and microcirculatory blood flow were recorded with the laser Doppler flowmeter, the velocity of red blood cells in venules was observed, and the number of open capillaries and adherent leukocytes were recorded. RESULTS: The blood flow velocity (mm/s) of the mesenteric microcirculation in LPS group was accelerated at 1 hour and 2 hours after LPS injection (1 hour: 0.190+/-0.007 vs. 0.174+/-0.009, 2 hours: 0.200+/-0.010 vs. 0.172+/-0.015, both P<0.05, respectively), but decelerated at 6 hours (0.116+/-0.015 vs. 0.164+/-0.011, P<0.05). The blood flow volume in the mesenteric vessels and the number of open capillaries did not show any significant change at that time. Significant increase in number of adherent leukocytes was observed at 2, 4, 6 hours after injury (2 hours: 2.60+/-1.14 vs. 0.40+/-0.55, 4 hours: 5.40+/-0.89 vs. 0.40+/-0.55, 6 hours : 5.40+/-1.52 vs. 0.60+/-0.90, all P<0.05, respectively). The state of blood flow in the microcirculation became abnormal. After irradiated with laser in low dose, the blood flow velocity was smooth and stable (mm/s, 1 hour: 0.174+/-0.011, 2 hours: 0.180+/-0.023, 4 hours: 0.168+/-0.013, 6 hours: 0.162+/-0.023), and the number of adherent leukocytes was reduced significantly at 4 hours and 6 hours than that in LPS group (4 hours: 2.00+/-0.71 vs. 5.40+/-0.89, 6 hours: 2.60+/-1.52 vs. 5.40+/-1.52, both P<0.05) and the microcirculatory flow state was improved obviously. CONCLUSION: Low-level laser irradiation may ameliorate the local mesenteric microcirculation, alleviate the microcirculatory disorder in early stage of ETM.