TY - GEN
T1 - Automatic Measurement of Fetal Head Circumference from 2-Dimensional Ultrasound
AU - Aji, Cahya Perbawa
AU - Fatoni, Muhammad Hilman
AU - Sardjono, Tri Arief
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/11
Y1 - 2019/11
N2 - Two-dimensional (2D) medical ultrasound is the primary imaging modality for the anatomical and functional surveillance of foetus due to its low cost, abundant availability, real-time capability, and the absence of radiation hazards. Head Circumference (HC) is one of the most important foetal biometrics in assessing foetal development during ultrasound examinations. Owing to its low signal-to-noise ratio, clinicians often have difficulty recognizing the foetal plane correctly from ultrasound 2D image. Moreover, clinicians often find difficulty to make the closest ellipse with only three minor and major parameter points provided by the ultrasound machine. The process of measuring HC manually by the clinician is quite an expensive procedure. Research on the automatic measurement of HC has become an active research area. In this study, an automatic measurement system for HC was proposed. The Convolutional Neural Network (CNN) is proposed to semantically segment foetal head from maternal and other foetal tissue. From this result it is expected to be easier to make an elliptical approach to the foetal plane because only the pixels belong to the head plane of the foetal are fed as input. According to the experimental result, in the process of the ellipse approach and its measurement, from thirteen test images the average semantic segmentation accuracy was 0.76 and the average error percentage of ellipse circumference measurement was 14.96%.
AB - Two-dimensional (2D) medical ultrasound is the primary imaging modality for the anatomical and functional surveillance of foetus due to its low cost, abundant availability, real-time capability, and the absence of radiation hazards. Head Circumference (HC) is one of the most important foetal biometrics in assessing foetal development during ultrasound examinations. Owing to its low signal-to-noise ratio, clinicians often have difficulty recognizing the foetal plane correctly from ultrasound 2D image. Moreover, clinicians often find difficulty to make the closest ellipse with only three minor and major parameter points provided by the ultrasound machine. The process of measuring HC manually by the clinician is quite an expensive procedure. Research on the automatic measurement of HC has become an active research area. In this study, an automatic measurement system for HC was proposed. The Convolutional Neural Network (CNN) is proposed to semantically segment foetal head from maternal and other foetal tissue. From this result it is expected to be easier to make an elliptical approach to the foetal plane because only the pixels belong to the head plane of the foetal are fed as input. According to the experimental result, in the process of the ellipse approach and its measurement, from thirteen test images the average semantic segmentation accuracy was 0.76 and the average error percentage of ellipse circumference measurement was 14.96%.
KW - convolutional neural network
KW - ellipse fitting
KW - foetal head circumference
KW - semantic segmentation
UR - http://www.scopus.com/inward/record.url?scp=85084412544&partnerID=8YFLogxK
U2 - 10.1109/CENIM48368.2019.8973258
DO - 10.1109/CENIM48368.2019.8973258
M3 - Conference contribution
AN - SCOPUS:85084412544
T3 - 2019 International Conference on Computer Engineering, Network, and Intelligent Multimedia, CENIM 2019 - Proceeding
BT - 2019 International Conference on Computer Engineering, Network, and Intelligent Multimedia, CENIM 2019 - Proceeding
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2nd International Conference on Computer Engineering, Network, and Intelligent Multimedia, CENIM 2019
Y2 - 19 November 2019 through 20 November 2019
ER -