Smart identification of psoriasis by images using convolutional neural networks: a case study in China

BACKGROUND

Psoriasis is a chronic inflammatory skin disease, which holds a high incidence in China. However, professional dermatologists who can diagnose psoriasis early and correctly are insufficient in China, especially in the rural areas. A smart approach to identify psoriasis by pictures would be highly adaptable countrywide and could play a useful role in early diagnosis and regular treatment of psoriasis.

OBJECTIVES

Design and evaluation of a smart psoriasis identification system based on clinical images (without relying on a dermatoscope) that works effectively similar to a dermatologist.

METHODS

A set of deep learning models using convolutional neural networks (CNNs) was explored and compared in the system for automatic identification of psoriasis. The work was carried out on a standardized dermatological dataset with 8021 clinical images of 9 common disorders including psoriasis along with full electronic medical records of patients built over the last 9 years in China. A two-stage deep neural network was designed and developed to identify psoriasis. In the first stage, a multilabel classifier was trained to learn the visual patterns for each individual skin disease. In the second stage, the output of the first stage was utilized to distinguish psoriasis from other skin diseases.

RESULTS

The area under the curve (AUC) of the two-stage model reached 0.981 ± 0.015, which outperforms a single-stage model. And, the classifier showed superior performance (missed diagnosis rate: 0.03, misdiagnosis rate: 0.04) than 25 Chinese dermatologists (missed diagnosis rate: 0.19, misdiagnosis rate: 0.10) in the diagnosis of psoriasis on 100 clinical images.

CONCLUSIONS

Using clinical images to identify psoriasis is feasible and effective based on CNNs, which also builds a solid technical base for smart care of skin diseases especially psoriasis using mobile/tablet applications for teledermatology in China.

A microRNA expression signature as a predictor of survival for colon adenocarcinoma

Colon cancer is a major cause of cancer mortality worldwide and most colon cancers are adenocarcinoma. MicroRNA (miRNA) expression signature has been shown to be able to predict progression and prognosis of various cancers. The aim of our study was to explore a novel signature of microRNA expression for predicting survival of colon adenocarcinoma patients. By analyzing the miRNA expression profiles and clinical information of 329 colon adenocarcinoma patients derived from The Cancer Genome Atlas database. 129 miRNAs were identified to be expressed differentially between the cancer and adjacent tissues. Among them, 27 miRNAs were found to be associated with the corresponding clinical characteristics of the patients. Furthermore, 7 miRNAs (let-7a-2, mir-32, mir-181a-1, mir-197, mir-328, mir-505 and mir-652) were found to be significantly correlated with the patient survival. The risk established by the 7-miRNA signature we built was proved be an independent prognostic factor (Hazard ratio [HR] = 2.048; 95% CI = 1.144-3.664; p, 0.016). In summary, our study identified miRNAs correlated with progression and prognosis of colon adenocarcinoma and built a 7-microRNA expression signature for prediction of the survival of the patients with colon adenocarcinoma.

Cytotoxic effect of oxaloacetate on HepG2-human hepatic carcinoma cells via apoptosis and ROS accumulation

Oxaloacetate (OA) is one of the intermediates of the Krebs cycle. In addition to its role in energy production, OA may have other effects on the cell. We report here that OA could have a cell type dependent cytotoxic effect on the human hepatic carcinoma cell line HepG2 through induction of apoptosis and reactive oxygen species (ROS) accumulation. In our study, OA decreased the viability and colony formation of HepG2 cells and induced cell death. Caspase-3 activity was increased, the pro-apoptotic protein Bax was up-regulated, and the anti-apoptotic protein Bcl-2 was down-regulated in OA-treated HepG2 cells indicating that apoptosis through the intrinsic pathway was involved in the cell death. The ROS level in OA-treated HepG2 cells was increased. The anti-oxidant N-acetylcysteine (NAC) and glutathione (GSH) prevented the OA-induced decrease in cell but did not alter the enhanced apoptotic Bax/Bcl-2 mRNA ratio. These results suggest that the OA-induced apoptosis of HepG2 cell is not driven by oxidative damage and at least two distinct mechanisms, one mediated by ROS and one involving apoptosis, result in the cytotoxic effects of OA on HepG2 cells. These studies expand the biological functional repertoire of OA and provide a mechanism by which hepatocellular carcinoma may be targeted by OA.

Effects of climate on soil phosphorus cycle and availability in natural terrestrial ecosystems

Climate is predicted to change over the 21st century. However, little is known about how climate change can affect soil phosphorus (P) cycle and availability in global terrestrial ecosystems, where P is a key limiting nutrient. With a global database of Hedley P fractions and key-associated physiochemical properties of 760 (seminatural) natural soils compiled from 96 published studies, this study evaluated how climate pattern affected soil P cycle and availability in global terrestrial ecosystems. Overall, soil available P, indexed by Hedley labile inorganic P fraction, significantly decreased with increasing mean annual temperature (MAT) and precipitation (MAP). Hypothesis-oriented path model analysis suggests that MAT negatively affected soil available P mainly by decreasing soil organic P and primary mineral P and increasing soil sand content. MAP negatively affected soil available P both directly and indirectly through decreasing soil primary mineral P; however, these negative effects were offset by the positive effects of MAP on soil organic P and fine soil particles, resulting in a relatively minor total MAP effect on soil available P. As aridity degree was mainly determined by MAP, aridity also had a relatively minor total effect on soil available P. These global patterns generally hold true irrespective of soil depth (≤10 cm or >10 cm) or site aridity index (≤1.0 or >1.0), and were also true for the low-sand (≤50%) soils. In contrast, available P of the high-sand (>50%) soils was positively affected by MAT and aridity and negatively affected by MAP. Our results suggest that temperature and precipitation have contrasting effects on soil P availability and can interact with soil particle size to control soil P availability.

Dominant Species in Subtropical Forests Could Decrease Photosynthetic N Allocation to Carboxylation and Bioenergetics and Enhance Leaf Construction Costs during Forest Succession

It is important to understand how eco-physiological characteristics shift in forests when elucidating the mechanisms underlying species replacement and the process of succession and stabilization. In this study, the dominant species at three typical successional stages (early-, mid-, and late-succession) in the subtropical forests of China were selected. At each stage, we compared the leaf construction costs (CC), payback time (PBT), leaf area based N content (NA), maximum CO2 assimilation rate (Pmax), specific leaf area (SLA), photosynthetic nitrogen use efficiency (PNUE), and leaf N allocated to carboxylation (NC), and to bioenergetics (NB). The relationships between these leaf functional traits were also determined. The results showed that the early-succession forest is characterized with significantly lower leaf CC, PBT, NA, but higher Pmax, SLA, PNUE, NC, and NB, in relation to the late-succession forest. From the early- to the late-succession forests, the relationship between Pmax and leaf CC strengthened, whereas the relationships between NB, NC, PNUE, and leaf CC weakened. Thus, the dominant species are able to decrease the allocation of the photosynthetic N fraction to carboxylation and bioenergetics during forest succession. The shift in these leaf functional traits and their linkages might represent a fundamental physiological mechanism that occurs during forest succession and stabilization.

Dominant Trees in a Subtropical Forest Respond to Drought Mainly via Adjusting Tissue Soluble Sugar and Proline Content

It is well-known that drought has considerable effects on plant traits from leaf to ecosystem scales; however, little is known about the relative contributions of various traits within or between tree species in determining the plant's sensitivity or the tolerance to drought under field conditions. We conducted a field throughfall exclusion experiment to simulate short-term drought (∼67% throughfall exclusion during the dry season from October to March) and prolonged drought (∼67% throughfall exclusion prolonging the dry season from October to May) and to understand the effects of drought on two dominant tree species (Michelia macclurei and Schima superba) in subtropical forests of southern China. The morphological, physiological, and nutritional responses of the two species to the two types of drought were determined. There were significantly different morphological (leaf max length, max width, leaf mass per area), physiological (leaf proline) and nutritional (P, S, N, K, Ca, Mg) responses by M. macclurei and S. superba to prolonged drought. Comparison between the drought treatments for each species indicated that the trees responded species-specifically to the short-term and prolonged drought, with S. superba exhibiting larger plasticity and higher adaption than M. macclurei. M. macclurei responded more sensitively to prolonged drought in terms of morphology, proline content, and nutritional traits and to short-term drought with regard to soluble sugars content. The differential species-specific responses to drought will allow us to estimate the changes in dominant trees in subtropical forests of China that have experienced a decade's worth of annual seasonal drought.

Modelling and characterisation of a ultrasound-actuated needle for improved visibility in ultrasound-guided regional anaesthesia and tissue biopsy

Clear needle visualisation is recognised as an unmet need for ultrasound guided percutaneous needle procedures including regional anaesthesia and tissue biopsy. With inadequate needle visibility, these procedures may result in serious complications or a failed operation. This paper reports analysis of the modal behaviour of a previously proposed ultrasound-actuated needle configuration, which may overcome this problem by improving needle visibility in colour Doppler imaging. It uses a piezoelectric transducer to actuate longitudinal resonant modes in needles (outer diameter 0.8-1.2mm, length>65mm). The factors that affect the needle's vibration mode are identified, including the needle length, the transducer's resonance frequency and the gripping position. Their effects are investigated using finite element modelling, with the conclusions validated experimentally. The actuated needle was inserted into porcine tissue up to 30mm depth and its visibility was observed under colour Doppler imaging. The piezoelectric transducer is able to generate longitudinal vibration with peak-to-peak amplitude up to 4μm at the needle tip with an actuating voltage of 20Vpp. Actuated in longitudinal vibration modes (distal mode at 27.6kHz and transducer mode at 42.2kHz) with a drive amplitude of 12-14Vpp, a 120mm needle is delineated as a coloured line in colour Doppler images, with both needle tip and shaft visualised. The improved needle visibility is maintained while the needle is advanced into the tissue, thus allowing tracking of the needle position in real time. Moreover, the needle tip is highlighted by strong coloured artefacts around the actuated needle generated by its flexural vibration. A limitation of the technique is that the transducer mode requires needles of specific lengths so that the needle's resonance frequency matches the transducer. This may restrict the choice of needle lengths in clinical applications.