ABSTRACT
This paper describes an enhanced process able to attain righteous classification of morphological malformation in foetal head ultrasound images. These anomalies can be detected approximately 20–22 weeks. In effect, the experts rely on manual analysis by extracting typical biometric measures from head region to interpret pregnancy evolution. The contribution of this work presents a totally computerised method of cerebral defect-recognition based on conventional neural network (CNN) in order to supply quantitative appraisal of hydrocephalus (HD) or healthy (HL) topics. The obtained results have achieved an important classification (Accuracy = 97.20%, sensitivity = 97.95% and specificity = 98.23%) when applying the CNN. The proposed methodology enables us to assess the anomalous cases in premature period within a reduced processing time. Experimental results prove the success of the proposed rating system for a suitable diagnostic of foetal head hydrocephalus when compared to previous work.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Additional information
Notes on contributors
Hanene Sahli
Hanene Sahli is associate professor in image and signal processing. She is a member of the research group in Signal Image and Energy Mastery laboratory (SIME) at the university of Tunis, ENSIT. Her research interests include Machine learning methods, pattern recognition, biomedical Video- Image analysis and biomedical Signal Processing.
Mounir Sayadi
Mounir Sayadiis professor at ENSIT- University of Tunis and head of of the research group in Signal Image and Energy Mastery laboratory (SIME) at the University of Tunis. His research interests are focused on adaptive signal processing and filtering, medical image and texture classification and segmentation.
Radhouane Rachdi
Radhouane Rachdi is a professor of gynaecology in Military hospital, Tunis Tunisia. His research interests include andrology, pediatrics and gynaecology.