Differences in the annotation between facial images and videos for training an artificial intelligence for skin type determination.

BACKGROUND: The Grand-AID research project, consisting of GRANDEL-The Beautyness Company, the dermatology department of Augsburg University Hospital and the Chair of IT Infrastructure for Translational Medical Research at Augsburg University, is currently researching the development of a digital skin consultation tool that uses artificial intelligence (AI) to analyze the user's skin and ultimately perform a personalized skin analysis and a customized skin care routine. Training the AI requires annotation of various skin features on facial images. The central question is whether videos are better suited than static images for assessing dynamic parameters such as wrinkles and elasticity. For this purpose, a pilot study was carried out in which the annotations on images and videos were compared. MATERIALS AND METHODS: Standardized image sequences as well as a video with facial expressions were taken from 25 healthy volunteers. Four raters with dermatological expertise annotated eight features (wrinkles, redness, shine, pores, pigmentation spots, dark circles, skin sagging, and blemished skin) with a semi-quantitative and a linear scale in a cross-over design to evaluate differences between the image modalities and between the raters. RESULTS: In the videos, most parameters tended to be assessed with higher scores than in the images, and in some cases significantly. Furthermore, there were significant differences between the raters. CONCLUSION: The present study shows significant differences between the two evaluation methods using image or video analysis. In addition, the evaluation of the skin analysis depends on subjective criteria. Therefore, when training the AI, we recommend regular training of the annotating individuals and cross-validation of the annotation.

MISM: A Medical Image Segmentation Metric for Evaluation of Weak Labeled Data.

Performance measures are an important tool for assessing and comparing different medical image segmentation algorithms. Unfortunately, the current measures have their weaknesses when it comes to assessing certain edge cases. These limitations arise when images with a very small region of interest or without a region of interest at all are assessed. As a solution to these limitations, we propose a new medical image segmentation metric: MISm. This metric is a composition of the Dice similarity coefficient and the weighted specificity. MISm was investigated for definition gaps, an appropriate scoring gradient, and different weighting coefficients used to propose a constant value. Furthermore, an evaluation was performed by comparing the popular metrics in the medical image segmentation and MISm using images of magnet resonance tomography from several fictitious prediction scenarios. Our analysis shows that MISm can be applied in a general way and thus also covers the mentioned edge cases, which are not covered by other metrics, in a reasonable way. In order to allow easy access to MISm and therefore widespread application in the community, as well as reproducibility of experimental results, we included MISm in the publicly available evaluation framework MISeval.

The Antarctic ozone hole and the pattern effect on climate sensitivity.

Since about 1980, the tropical Pacific has been anomalously cold, while the broader tropics have warmed. This has caused anomalous weather in midlatitudes as well as a reduction in the apparent sensitivity of the climate associated with enhanced low-cloud abundance over the cooler waters of the eastern tropical Pacific. Recent modeling work has shown that cooler temperatures over the Southern Ocean around Antarctica can lead to cooler temperatures over the eastern tropical Pacific. Here we suggest that surface wind anomalies associated with the Antarctic ozone hole can cause cooler temperatures over the Southern Ocean that extend into the tropics. We use the short-term variability of the Southern Annular Mode of zonal wind variability to show an association between surface zonal wind variations over the Southern Ocean, cooling over the Southern Ocean, and cooling in the eastern tropical Pacific. This suggests that the cooling of the eastern tropical Pacific may be associated with the onset of the Antarctic ozone hole.

Congestus Mode Invigoration by Convective Aggregation in Simulations of Radiative-Convective Equilibrium.

This study examines how the congestus mode of tropical convection is expressed in numerical simulations of radiative-convective equilibrium (RCE). We draw insights from the ensemble of cloud-resolving models participating in the RCE Model Intercomparison Project (RCEMIP) and from a new ensemble of two-dimensional RCE simulations. About half of the RCEMIP models produce a congestus circulation that is distinct from the deep and shallow modes. In both ensembles, the congestus circulation strengthens with large-scale convective aggregation, and in the 2D ensemble this comes at the expense of the shallow circulation centered at the top of the boundary layer. Congestus invigoration occurs because aggregation dries out the upper troposphere, which allows moist congestus outflow to undergo strong radiative cooling. The cooling generates divergence that promotes continued congestus overturning (a positive feedback). This mechanism is fundamentally similar to the driving of shallow circulations by radiative cooling at the top of the surface boundary layer. Aggregation and congestus invigoration are also associated with enhanced static stability throughout the troposphere, but a modeling experiment shows that enhanced stability is not necessary for congestus invigoration; rather, invigoration itself contributes to the stability increase via its impact on the vertical profile of radiative cooling. Changes in entrainment cooling are also found to play an important role in stability enhancement, as has been suggested previously. When present, congestus circulations have a large impact on the mean RCE atmospheric state; for this reason, their inconsistent representation in models and their impact on the real tropical atmosphere warrant further scrutiny.

MISeval: A Metric Library for Medical Image Segmentation Evaluation.

Correct performance assessment is crucial for evaluating modern artificial intelligence algorithms in medicine like deep-learning based medical image segmentation models. However, there is no universal metric library in Python for standardized and reproducible evaluation. Thus, we propose our open-source publicly available Python package MISeval: a metric library for Medical Image Segmentation Evaluation. The implemented metrics can be intuitively used and easily integrated into any performance assessment pipeline. The package utilizes modern DevOps strategies to ensure functionality and stability. MISeval is available from PyPI (miseval) and GitHub: https://github.com/frankkramer-lab/miseval.

The lifecycle of anvil clouds and the top-of-atmosphere radiation balance over the tropical west Pacific.

Observations from a geostationary satellite are used to describe the lifecycle of mesoscale convective systems (MCS), their associated anvil clouds, and their effects on the radiation balance over the warm pool of the tropical west Pacific Ocean. In their developing stages, MCS primarily consist of clouds that are optically thick and have a negative net cloud radiative effect (CRE). As MCS age, ice crystals in the anvil become larger, the cloud top lowers somewhat, and clouds with neutral and positive net CRE become more common. Shading from anvils causes cool anomalies in the underlying sea surface temperature (SST) of up to -0.6 °C. MCS often occur in clusters that are embedded within large westward-propagating disturbances, so shading from anvils can cool SSTs over regions spanning hundreds of kilometers. Triggering of convection is more likely to follow a warm SST anomaly than a cold SST anomaly on timescales of several days. This information is used to test hypotheses on why, over the warm pool, the average shortwave and longwave CRE are individually large but nearly cancel. The results are consistent with the hypothesis that the cancelation in CRE is caused by feedbacks between cloud albedo, large-scale circulation, and SST.

Natural aerosols explain seasonal and spatial patterns of Southern Ocean cloud albedo.

Atmospheric aerosols, suspended solid and liquid particles, act as nucleation sites for cloud drop formation, affecting clouds and cloud properties-ultimately influencing the cloud dynamics, lifetime, water path, and areal extent that determine the reflectivity (albedo) of clouds. The concentration N d of droplets in clouds that influences planetary albedo is sensitive to the availability of aerosol particles on which the droplets form. Natural aerosol concentrations affect not only cloud properties themselves but also modulate the sensitivity of clouds to changes in anthropogenic aerosols. It is shown that modeled natural aerosols, principally marine biogenic primary and secondary aerosol sources, explain more than half of the spatiotemporal variability in satellite-observed N d. Enhanced N d is spatially correlated with regions of high chlorophyll a, and the spatiotemporal variability in N d is found to be driven primarily by high concentrations of sulfate aerosol at lower Southern Ocean latitudes (35(o) to 45(o)S) and by organic matter in sea spray aerosol at higher latitudes (45(o) to 55(o)S). Biogenic sources are estimated to increase the summertime mean reflected solar radiation in excess of 10 W m(-2) over parts of the Southern Ocean, which is comparable to the annual mean increases expected from anthropogenic aerosols over heavily polluted regions of the Northern Hemisphere.