Related Experiment Video
Updated: Mar 27, 2026
13:56
Characterization of Molecular Mechanisms of In vivo UVR Induced Cataract
Published on: November 28, 2012
10.5K
Automatic Cataract Hardness Classification Ex Vivo by Ultrasound Techniques
Miguel Caixinha1, Mário Santos2, Jaime Santos2
1Department of Physics, University of Coimbra, PT-3030-290 Coimbra, Portugal; Department of Electrical and Computer Engineering, University of Coimbra, Coimbra, Portugal.
Ultrasound in Medicine & Biology
|January 9, 2016
Summary
Ultrasound techniques offer a new, non-invasive method for characterizing cataract hardness. This approach accurately classifies cataract severity, aiding in diagnosis and treatment planning.
Area of Science:
- Ophthalmology
- Biomedical Engineering
- Medical Imaging
Background:
- Cataract hardness impacts surgical difficulty and outcomes.
- Accurate characterization of cataract hardness is crucial for surgical planning.
- Current methods for cataract assessment are often subjective or invasive.
Purpose of the Study:
- To establish the feasibility of a novel ultrasound-based methodology for cataract hardness characterization.
- To develop an automatic classification system for different cataract severities using ultrasound data.
- To evaluate the performance of various machine learning classifiers for cataract classification.
Main Methods:
- Induced varying degrees of cataracts in 210 porcine lenses.
- Utilized a 25-MHz ultrasound transducer to acquire acoustical parameters (velocity, attenuation) and backscattering signals.
- Constructed B-Scan and Nakagami images, extracted 97 parameters, and applied Principal Component Analysis.
- Employed Bayes, K-Nearest-Neighbours, Fisher Linear Discriminant, and Support Vector Machine (SVM) classifiers.
Main Results:
- Demonstrated statistically significant increases in velocity, attenuation, B-Scan brightness intensity, and Nakagami m parameter with cataract formation (p < 0.01).
- Achieved high performance (F-measure ≥ 92.68%) for classifying healthy versus cataractous lenses across all four classifiers.
- The SVM classifier exhibited superior performance (90.62%) in differentiating initial from severe cataracts.
Conclusions:
- Ultrasound techniques provide a feasible and non-invasive approach for cataract hardness characterization.
- Automatic classification of cataract severity is achievable using ultrasound parameters and machine learning.
- This methodology holds potential for improving objective assessment and management of cataracts.
