The scarcity of labeled data often limits the application of supervised deep learning techniques for medical image segmentation. This has motivated the development of semi-supervised techniques that learn from a mixture of labeled and unlabeled images. In this paper, we propose a novel semi-supervised method that, in addition to supervised learning on labeled training images, learns to predict segmentations consistent under a given class of transformations on both labeled and unlabeled images. More specifically, in this work we explore learning equivariance to elastic deformations. We implement this through: 1) a Siamese architecture with two identical branches, each of which receives a differently transformed image, and 2) a composite loss function with a supervised segmentation loss term and an unsupervised term that encourages segmentation consistency between the predictions of the two branches. We evaluate the method on a public dataset of chest radiographs with segmentations of anatomical structures using 5-fold crossvalidation. The proposed method reaches significantly higher segmentation accuracy compared to supervised learning. This is due to learning transformation consistency on both labeled and unlabeled images, with the latter contributing the most. We achieve the performance comparable to state-of-the-art chest X-ray segmentation methods while using substantially fewer labeled images.
A potentially important feature in a divertor design for a high-power tokamak is an extended and expanded divertor leg. The upgrade to MAST will allow a wide range of such divertor leg geometries to be produced, and hence will allow the roles of greatly increased connection length and flux expansion to be experimentally tested. This will include testing the potential of the Super-X configuration [1]. The design process for the upgrade has required analysis of producing and controlling the magnetic configurations, and has included consideration of the roles that divertor closure and increasing magnetic connection length will play.
We present a real-time approach for traversable surface detection using a low-cost monocular camera mounted on an autonomous vehicle. The proposed methodology extracts colour and texture information from various channels of the HSL, YCbCr and LAB colourspaces by temporal analysis in order to create a traversability map. On this map lighting and water artifacts are eliminated including shadows, reections and water prints. Additionally, camera vibration is compensated by temporal ltering leading to robust path edge detection in blurry images.The performance of this approach is extensively evaluated over varying terrain and environmental conditions and the eect of colourspace fusion on the system's precision is analysed. The results show a mean accuracy of 97% over this comprehensive test set.
This study aims to measure the effect of toxic aqueous solutions of metals on the mobility of Artemia salina nauplii by using digital image processing. The instrument consists of a camera with a macro lens, a dark chamber, a light source and a laptop computer. Four nauplii were inserted into a macro cuvette, which contained copper, cadmium, iron and zinc ions at various concentrations. The nauplii were then filmed inside the dark chamber for two minutes and the video sequence was processed by a motion tracking algorithm that estimated their mobility. The results obtained by this system were compared to the mortality assay of the Artemia salina nauplii. Despite the small number of tested organisms, this system demonstrates great sensitivity in quantifying the mobility of the nauplii, which leads to significantly lower EC50 values than those of the mortality assay. Furthermore, concentrations of parts per trillion of toxic compounds could be detected for some of the metals. The main novelty of this instrument relies in the sub-pixel accuracy of the tracking algorithm that enables robust measurement of the deterioration of the mobility of Artemia salina even at very low concentrations of toxic metals.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.