<scp>OP15</scp>.02: Automatic fetal head detection on video clips from a low‐cost portable <scp>USB</scp> ultrasound (<scp>US</scp>) device

M. A. Maraci, R. Napolitano, Aris T. Papageorghiou, J. Alison NobleUltrasound 2013 · DOI

Abstract

The use of ultrasound (US) is less prevalent in low income countries, in part due to lack of availability of trained staff; and because of high costs associated with US devices. Low cost US probes connected to laptop computers using dedicated software are currently available and can be used for real time fetal scans. In this paper we have investigated whether a combined machine learning and data acquisition approach to fetal head detection from a low cost USB probe is comparable to a high end probe. Transabdominal 2D fetal US video clips were acquired from 15 subjects participating in the INTERGROWTH-21st study (www.intergrowth21st.org) at the University of Oxford. Data acquisition was carried out using a mid-range US machine (Philips HD9 with a V7-3 transducer) and a low-cost portable USB US machine (Interson Seemore). We compared 30 images from each machine (2 per fetus) to assess how well the fetal head could be detected using image analysis solutions. To do this, dedicated software was developed with a machine learning approach for detection of head circumference. The Likert-scale grading system with a scale of 1–3 was used to rate the performance and a Wilcoxon-Rank test was performed. Automatic head detection was possible in 23/30 (77%) and 24/30 (80%) of images taken from the Philips and Interson Seemore probes respectively. There was no significant difference between the two machines in the accuracy of head detection using the proposed method (p = 0.20), despite a significant difference between sharpness (p = 1.23*10-4) and visibility (p = 6.69*10-4) of the images prior to being processed. We have described a new machine learning strategy for processing images obtained from a low-cost probe. Although the visibility and sharpness of the low-cost probe was lower than the traditional US machine, the performance of an automatic fetal head detection algorithm was not affected.
BibTeX
@inproceedings{maraci2013scp,
  author = {M. A. Maraci and R. Napolitano and Aris T. Papageorghiou and J. Alison Noble},
  title = {{<scp>OP15</scp>.02: Automatic fetal head detection on video clips from a low‐cost portable <scp>USB</scp> ultrasound (<scp>US</scp>) device}},
  booktitle = {Ultrasound},
  year = {2013},
  doi = {10.1002/uog.12844},
}

Loading related content...

Figure