Bubble growth analysis during subcooled boiling experiments on-board the international space station: Benchmark image analysis

This work compares four different image processing algorithms for the analysis of image data obtained during the Multiscale Boiling Experiment of ESA, executed on-board the International Space Station. Two separate experimental campaigns have been performed in 2019 and 2020, aiming to investigate boiling phenomena in microgravity, with and without the presence of shear flow and electric field. A heated substrate, at the bottom of the test cell, creates a temperature profile across the liquid bulk above it. A laser beam hits a designated microcavity at the middle of the substrate, to initiate nucleation of a single, isolated bubble. In the presence of shear flow or electric field forces, the bubble slides or detaches respectively, leaving the cavity free for the nucleation and growth of a new bubble. The growth of such a bubble within the prescribed temperature profile is studied for varying experimental conditions (i.e. pressure, heat flux, subcooling temperature) by capturing high speed, black and white video images. The presence of light reflections at random locations around the bubble contour vary with bubble size and population. This, combined with the refraction induced optical distortion of vertical image dimension close to the heater, make the accurate detection of bubbles contour a real challenge. Four research teams, namely the University of Pisa (UNIPI), the Institute of Fluid Mechanics of Toulouse (IMFT), the joint group of Aix Marseille University (AMU) and Kutateladze Institute of Thermophysics (IT), and the joined group of Aristotle University of Thessaloniki (AUTH), Technical University of Darmstadt (TUD) and Foundation of Research and Technology in Crete (FORTH), developed separate specialized algorithms to: a) detect bubble edges and b) use these edges to calculate basic bubble geometrical features, such as contact line diameter, bubble diameter and contact angles. These four different approaches diverge in complexity and concept. In the absence of reference measurements at microgravity conditions, measurements efficiency is evaluated based on the comparison of the estimated bubble geometrical features along with pertinent physical arguments. Results show that the efficiency of each approach varies with the nature of measurement. The studied benchmark dataset is published allowing other research groups to test further their own image processing algorithms.

I2Cnet medical image annotation service

I2Cnet (Image Indexing by Content network) aims to provide services related to the content-based management of images in healthcare over the World-Wide Web. Each I2Cnet server maintains an autonomous repository of medical images and related information. The annotation service of I2Cnet allows specialists to interact with the contents of the repository, adding comments or illustrations to medical images of interest. I2Cnet annotations may be communicated to other users via e-mail or posted to I2Cnet for inclusion in its local repositories. This paper discusses the annotation service of I2Cnet and argues that such services pave the way towards the evolution of active digital medical image libraries.

The I2Cnet service architecture paradigm

The main objective of the Image Indexing by Content network (I2Cnet) is to provide network-transparent content-based access to medical image archives based on a collection of interoperable Internet/intranet added-value services. This paper discusses I2Cnet, focusing on its service architecture paradigm. I2Cnet services such as image annotation, processing, description, and content-based retrieval, as well as the on-line collaboration service are presented. Exemplary user sessions are used to illustrate how virtual workspaces facilitate the interoperation of I2Cnet services, following the "network computer" approach to information management.