A deep learning system for detecting diabetic retinopathy across the disease spectrum.

Retinal screening contributes to early detection of diabetic retinopathy and timely treatment. To facilitate the screening process, we develop a deep learning system, named DeepDR, that can detect early-to-late stages of diabetic retinopathy. DeepDR is trained for real-time image quality assessment, lesion detection and grading using 466,247 fundus images from 121,342 patients with diabetes. Evaluation is performed on a local dataset with 200,136 fundus images from 52,004 patients and three external datasets with a total of 209,322 images. The area under the receiver operating characteristic curves for detecting microaneurysms, cotton-wool spots, hard exudates and hemorrhages are 0.901, 0.941, 0.954 and 0.967, respectively. The grading of diabetic retinopathy as mild, moderate, severe and proliferative achieves area under the curves of 0.943, 0.955, 0.960 and 0.972, respectively. In external validations, the area under the curves for grading range from 0.916 to 0.970, which further supports the system is efficient for diabetic retinopathy grading.

Domain-invariant interpretable fundus image quality assessment.

Objective and quantitative assessment of fundus image quality is essential for the diagnosis of retinal diseases. The major factors in fundus image quality assessment are image artifact, clarity, and field definition. Unfortunately, most of existing quality assessment methods focus on the quality of overall image, without interpretable quality feedback for real-time adjustment. Furthermore, these models are often sensitive to the specific imaging devices, and cannot generalize well under different imaging conditions. This paper presents a new multi-task domain adaptation framework to automatically assess fundus image quality. The proposed framework provides interpretable quality assessment with both quantitative scores and quality visualization for potential real-time image recapture with proper adjustment. In particular, the present approach can detect optic disc and fovea structures as landmarks, to assist the assessment through coarse-to-fine feature encoding. The framework also exploit semi-tied adversarial discriminative domain adaptation to make the model generalizable across different data sources. Experimental results demonstrated that the proposed algorithm outperforms different state-of-the-art approaches and achieves an area under the ROC curve of 0.9455 for the overall quality classification.

IDRiD: Diabetic Retinopathy - Segmentation and Grading Challenge.

Diabetic Retinopathy (DR) is the most common cause of avoidable vision loss, predominantly affecting the working-age population across the globe. Screening for DR, coupled with timely consultation and treatment, is a globally trusted policy to avoid vision loss. However, implementation of DR screening programs is challenging due to the scarcity of medical professionals able to screen a growing global diabetic population at risk for DR. Computer-aided disease diagnosis in retinal image analysis could provide a sustainable approach for such large-scale screening effort. The recent scientific advances in computing capacity and machine learning approaches provide an avenue for biomedical scientists to reach this goal. Aiming to advance the state-of-the-art in automatic DR diagnosis, a grand challenge on "Diabetic Retinopathy - Segmentation and Grading" was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI - 2018). In this paper, we report the set-up and results of this challenge that is primarily based on Indian Diabetic Retinopathy Image Dataset (IDRiD). There were three principal sub-challenges: lesion segmentation, disease severity grading, and localization of retinal landmarks and segmentation. These multiple tasks in this challenge allow to test the generalizability of algorithms, and this is what makes it different from existing ones. It received a positive response from the scientific community with 148 submissions from 495 registrations effectively entered in this challenge. This paper outlines the challenge, its organization, the dataset used, evaluation methods and results of top-performing participating solutions. The top-performing approaches utilized a blend of clinical information, data augmentation, and an ensemble of models. These findings have the potential to enable new developments in retinal image analysis and image-based DR screening in particular.

Minimally effective concentration of zoledronic acid to suppress osteoclasts in vitro.

Zoledronic acid is regarded as the most potent bisphosphonate and is widely used in patients with osteoporosis; however, its side effects, including acute-phase reactions, gastrointestinal complaints, renal dysfunction and bisphosphonate-associated osteonecrosis impair the safety and quality of life of patients. The present study was designed to determine the minimal effective concentration of zoledronic acid through testing the dose-dependent effects of zoledronic acid on osteoclast suppression. A primary culture of bone marrow mononuclear cells obtained from C57 mice (age, 6 weeks) was established and induced to form osteoclasts. The number of multinuclear cells was determined by tartrate-resistant acid phosphatase staining and compared among cultured marrow cells treated with different concentrations of zoledronic acid. Furthermore, the cellular properties, including adhesion, migration and bone resorption, were compared at the minimal effective concentration. At a concentration of 1×10-6 mol/l, zoledronic acid significantly inhibited the formation of osteoclasts. This inhibitory effect was further enhanced at the concentration of 1×10-5 mol/l. However, the inhibitory effect of zoledronic acid tapered at the concentration of 1×10-4 mol/l and there was no further dose-dependent increase. In addition, the concentration of 1×10-6 mol/l was sufficient to alter cellular functions, including cell adhesion, migration and bone resorption. In conclusion, zoledronic acid was effective in reducing osteoclast formation and suppressing cellular functions. The minimal effective concentration of zoledronic acid in vitro was 1 µmol/l. Based on these results, a comparable dosage should be explored in clinical applications.

Phthalate exposure and high blood pressure in adults: a cross-sectional study in China.

Widespread phthalate exposure has been recently documented and is hypothesized to increase blood pressure (BP) in humans. However, current studies have provided inconclusive evidence for an association between phthalate exposure and BP. Human epidemiologic studies on the topic remain lacking. Therefore, this study aims to examine the association between serum phthalate concentrations and BP in a Chinese population. We measured several parameters of BP (systolic BP, diastolic BP, total cholesterol, and triglyceride) and the concentrations of 16 phthalates (dimethyl phthalate (DMP), diethyl phthalate, diisobutyl phthalate, dibutyl phthalate (DBP), bis (2-methoxyethyl) phthalate, bis (4-methyl-2-pentyl) phthalate, bis (2-ethoxyethyl) phthalate, diamyl phthalate, dihexyl phthalate, benzyl butyl phthalate, bis (2-nbutoxyethyl) phthalate (DBEP), dicyclohexyl phthalate, bis (2-ethyl hexyl) phthalate (DEHP), diisononyl phthalate, diphenyl phthalate, and di-n-octyl phthalate) in the serum of 474 adults recruited from a primary health care clinic. The relationship between serum phthalate concentrations and BP parameters was assessed with multivariate linear regressions. DBP was the most ubiquitous and dominant contaminant in the study population. The systolic BP of subjects in the median-exposure DEHP group significantly increased by 2.96 mmHg (p < 0.05) relative to that of subjects in the low-exposure group. Significant positive dose-related associations of DMP and DBEP with the levels of total cholesterol in serum (p for trend < 0.05) were also found. These associations persisted even when considering exposure to multiple phthalates. Our results suggested that phthalate exposure might increase BP in adults. However, our findings warrant further studies in a larger and more general population.

[Experimental study on inhibitory effect of zoledronic acid on the action style of the osteoclast].

This study is to investigate the inhibitory effect of different concentrations of zoledronic acid on the activity of osteoclasts, to obtain characteristics on inhibitory effect and to find the lowest effective concentration of zoledronic acid. Marrow cells of C57 mice（6 weeks） were cultured in vitro. Osteoclasts were induced by single nuclear cells. According to the concentration of zoledronic acid, we set up the experimental group with five different concentrations, i.e. 1×10­8 mol/L, 1×10­7 mol/L, 1×10­6 mol/L, 1×10­5 mol/L, and 1×10­4 mol/L. The control group did not contain any bisphosphonate. By tartrate resistant acid phosphatase staining, the number of multinuclear cells, cells through the filter and bone resorption lacune were counted. Five days after the cultivation, the number of multinuclear cells in the experimental group decreased with the increase of concentration of zoledronic acid. Inhibition on the formation of osteoclasts in vitro was effective at 1×10­6 mol/L. At the concentration of 1×10­5 mol/L, the effect of inhibition on migration of osteoclast and bone resorption was more obvious. The effect was further enhanced at concentration of 1×10­4 mol/L. However, the concentration and inhibition curves were gradually mild. The inhibitory effect on different concentrations of zoledronic acid on the activity of osteoclasts was different. The inhibition effect was obvious at 1×10­6 mol/L. We should pay attention to administrate appropriate concentration of zoledronic acid in the clinical applications.

Uniportal video-assisted thoracic surgery for left upper lobe: single-direction lobectomy with systematic lymphadenectomy.

A 62-year-old female was admitted to our hospital after computed tomography (CT) revealed a 2.5 cm × 2.1 cm mass in the left upper lobe. PET/CT scan diagnosed as malignant lesion with no signs of metastasis. Under general anesthesia, the patient was placed in right lateral decubitus position. A 4cm incision was made in the 4th intercostal space with plastic protector. The camera was placed in the upper part of the incision and the instruments were inserted below the camera. Left upper lobectomy along with systematic lymphadenectomy was performed. Total surgical time was 135 min and estimated blood loss was 70 mL. The chest tube was removed on the 2nd postoperative day and the patient was discharged on the 3rd postoperative day with no complication. This uniportal VATS single-direction lobectomy for the left upper lobe is feasible and amplifies the concept of thoracotomy-like minimally invasive surgery.

[EXPERIMENTAL STUDY ON EFFECT OF THREE DIFFERENT OPERATIVE WAYS OF ANNULUS FIBROSUS INCISION ON INTERVERTEBRAL DISC BIOMECHANICAL STRENGTH].

OBJECTIVE: To discuss the effect of three different ways of annulus fibrosus incision on the biomechanical strength of intervertebral disc. METHODS: A total of 30 goats underwent intervertebral disc nucleus pulposus extraction at L3,4 and 45 by the working channel in group A (n=10), by circular incision in group B (n=10), and by square incision in group C (n=10). The body weight, male and female ratio, age, intraoperative blood loss, and wound healing time were recorded and compared among 3 groups. The survival rate and wound healing situation were observed after operation. At 24 weeks after operation, the goats were sacrificed, MRI images were taken to observe the signal intensity of nucleus pulposus. The disc height of L(3,4) and L(4,5) was measured to calculate the loss of disc height; biomechanical test was used to assess the strength of the disc and anulus. Histological staining was also conducted to observe the repair effect at L(4,5). RESULTS: There was no significant difference in body weight, male to female ratio, age, intraoperative blood loss, and wound healing time among groups (P>0.05). All goats survived to the end of the experiment. MRI examination showed decreased signal intensity in 3 groups, indicating intervertebral disc degeneration. According to modified Thompson classification method, the degree of intervertebral disc degeneration of group A was significantly higher than that of groups B and C (P<0.05), but no significant difference was found between groups B and C (P>0.05). Difference was not significant in intervertebral space height before operation among 3 groups (P>0.05). But after 24 weeks, the intervertebral space height in group A was significantly higher than that in groups B and C (P<0.05), and the intervertebral space height loss in group A was significantly lower than that in groups B and C (P<0.05). The biomechanical strength in group A was also significantly higher than that in groups B and C (P<0.05), but no significant difference was found between group B and group C (P>0.05). HE and Masson staining showed good continuity of annulus fibrosus and clear layers in group A; poor continuity of annulus fibrosus and obvious scar tissues were observed in groups B and C. CONCLUSION: Application of working channel may have less destruction of annulus fibrosus, it plays a positive role in the maintenance of biomechanical strength and repair of annulus fibrosus.