This paper presents a new supervised method for segmentation of blood vessels in retinal photographs. This method uses an ensemble system of bagged and boosted decision trees and utilizes a feature vector based on the orientation analysis of gradient vector field, morphological transformation, line strength measures, and Gabor filter responses. The feature vector encodes information to handle the healthy as well as the pathological retinal image. The method is evaluated on the publicly available DRIVE and STARE databases, frequently used for this purpose and also on a new public retinal vessel reference dataset CHASE_DB1 which is a subset of retinal images of multiethnic children from the Child Heart and Health Study in England (CHASE) dataset. The performance of the ensemble system is evaluated in detail and the incurred accuracy, speed, robustness, and simplicity make the algorithm a suitable tool for automated retinal image analysis.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1109/TBME.2012.2205687 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Angle-controlled strong and weak coupling in photon molecules.
Sci Rep
January 2025
Terahertz Research Center, School of Electronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu, 610054, China.
Strong light-matter coupling occurs when the rate of energy exchange between the electromagnetic mode and the molecular ensemble exceeds the competitive dissipation process. Coupled photon molecules with near-field light-matter interactions may produce new hybridized states when they reach the strong coupling region. Tunable Terahertz (THz) meta materials can be used to design sensors, optical modulators, etc.
View Article and Find Full Text PDFImpacts of climate change on storm event-based flow regime and channel stability of urban headwater streams.
J Environ Manage
January 2025
Tetra Tech, Inc., P.O. Box 14409, Research Triangle Park, NC, 27709, United States. Electronic address:
Due to the recent improved availability of global and regional climate change (CC) models and associated data, the projected impact of CC on urban stormwater management is well documented. However, most studies are based on simplified design storm analysis and unit-area runoff models; evaluations of the long-term, continuous hydrologic response of extensive stormwater control measures (SCM) implementation under future CC scenarios are limited. Moreover, channel stability in response to CC is seldom evaluated due to the input data required to develop a long-term, continuous sediment transport model.
View Article and Find Full Text PDFDigital and Tunable Genetically Encoded Tension Sensors Based on Engineered Coiled-Coils.
Angew Chem Int Ed Engl
January 2025
Emory University, Chemistry, 1515 Dickey Dr., 30322, Atlanta, UNITED STATES OF AMERICA.
Genetically encoded tension sensors (GETSs) allow for quantifying forces experienced by intracellular proteins involved in mechanotransduction. The vast majority of GETSs are comprised of a FRET pair flanking an elastic "spring-like" domain that gradually extends in response to force. Because of ensemble averaging, the FRET signal generated by such analog sensors conceals forces that deviate from the average, and hence it is unknown if a subset of proteins experience greater magnitudes of force.
View Article and Find Full Text PDFEnvironmental factors affecting the BMI of older adults in the Philippines spatially assessed using machine learning.
Heliyon
January 2025
Department of Science and Technology - Food and Nutrition Research Institute, Taguig, Metro Manila, Philippines.
This study aimed to assess the environmental variables affecting the Body Mass Index of older adults at neighborhood levels (1 ha) while mapping probability distributions of normal, overweight-obese, and underweight older adults. We applied a data-driven method that integrates open-access remote sensing products and geospatial data, along with the first nutritional survey in the Philippines with geo-locations conducted in 2021. We used ensemble machine learning of different presence-only and presence-absence models, all subjected to hyperparameter tuning and variable decorrelation.
View Article and Find Full Text PDFEditorial: Attacking Biological Problems Through Single-Molecule Approaches.
BMB Rep
January 2025
Department of Physics, Division of Interdisciplinary Bioscience & Bioengineering, POSTECH, Pohang 37673, Republic of Korea.
In the last few decades, single-molecule techniques have emerged as transformative tools for exploring biological problems. By observing and analyzing individual molecules, these methods make it possible to investigate fundamental dynamics of biomolecular processes deeper. Unlike traditional ensemble methods that average the behavior of populations, single-molecule approaches provide a unique window to observe molecular heterogeneity, transient interactions, and dynamic processes that are otherwise hidden.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!