A Novel Semi-Automated Method of Tracking Fetal Movements

S. Bhaskar, C. Surlea, F. Mone, F. Kurugollu, P. Milligan, S. Ong

Research output: Contribution to journalArticlepeer-review


We have designed software that can â€â€™look’’ at recorded ultrasound sequences. We analyzed fifteen video sequences representing recorded ultrasound scans of nine fetuses. Our method requires a small amount of user labelled pixels for processing the first frame. These initialize GrowCut 1 , a background removal algorithm, which was used for separating the fetus from its surrounding environment (segmentation). For each subsequent frame, user input is no longer necessary as some of the pixels will inherit labels from the previously processed frame. This results in our software’s ability to track movement. Two sonographers rated the results of our computer’s â€vision’ on a scale from 1 (poor fit) to 10 (excellent fit). They assessed tracking accuracy for the entire video as well as segmentation accuracy (the ability to identify fetus from non-fetus) for every 100th processed frame. There was no appreciable deterioration in the software’s ability to track the fetus over time. I
Original languageEnglish
Number of pages2
JournalIrish Medical Journal
Issue number8
Publication statusPublished - Sep 2014

Fingerprint Dive into the research topics of 'A Novel Semi-Automated Method of Tracking Fetal Movements'. Together they form a unique fingerprint.

Cite this