Classification of Medical Ultrasound Transducer Using Neural Network | ||||
| Al-Azhar University Journal of Virus Researches and Studies | ||||
| Article 1, Volume 2, Issue 1, 2020, Page 1-10 PDF (584.07 K) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/AUJV.2020.106704 | ||||
 View on SCiNiTO | ||||
| Author | ||||
| Laila M. Aboughazala | ||||
| Center for Virus Research and Studies, Al-Azhar University, Cairo, Egypt | ||||
| Abstract | ||||
| Diagnostic medical ultrasound makes a significant contribution to patient care and is increasingly used in a variety of clinical settings by many different professionals with varying technical backgrounds (e.g. hepatocellular carcinoma). Therefore, the importance of ultrasound quality control is not only necessary for patient and operator safety but is also essential for maintaining the performance of the equipment to the highest-level achievable and it is required by various regulatory and accrediting agencies. Ultrasound image degradation originates primarily from transducer defects and potentially undermines reliable image interpretation. The ultrasound probe in-air reverberation pattern is used in routine quality assurance. We produce a quantitative quality control based on in-air reverberation images. They are easily generated for any probe independent of the level of expertise of the operator. The results are available to the sonographer prior to clinical use and transducer status can be remotely monitored with trend analysis over time. The method presents a scheme for the classification of normal functioning and defect transducers Region of Interest selected "ROIs" of 65 probes based on texture analysis that automatically detects in-air reverberation regions and recognizes them as normal functioning and defect transducers. However, feature selection is done by Twin Support Vector Machine. The accuracy of these features in distinguishing normal functioning and defect transducers has been evaluated by artificial neural network, and linear support vector machine algorithms classifiers. From the analysis of results, it was found that artificial neural network classifier gave an overall classification accuracy of 100% with 100% sensitivity. The results show that it is feasible to identify defect transducers based on texture features extracted from in-air reverberation ultrasound images. This method is shown to be useful for increased accuracy and increased speed for classification of functioning and defect transducers for improving the quality assurance of ultrasound. | ||||
| Keywords | ||||
| Quality assurance; in-air reverberation ultrasound images; Twin Support Vector Machine (TWSVM); artificial neural network (ANN); support vector machine (SVM) | ||||
|
Statistics Article View: 172 PDF Download: 260 |
||||
