Fetal skull segmentation in 3D ultrasound via structured geodesic random forest

Juan J. Cerrolaza*, Ozan Oktay, Alberto Gomez Herrero, Jackie Matthew, Caroline Knight, Bernhard Kainz, Daniel Rueckert

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

12 Citations (Scopus)

Abstract

Ultrasound is the primary imaging method for prenatal screening and diagnosis of fetal anomalies. Thanks to its non-invasive and non-ionizing properties, ultrasound allows quick, safe and detailed evaluation of the unborn baby, including the estimation of the gestational age, brain and cranium development. However, the accuracy of traditional 2D fetal biometrics is dependent on operator expertise and subjectivity in 2D plane finding and manual marking. 3D ultrasound has the potential to reduce the operator dependence. In this paper, we propose a new random forest-based segmentation framework for fetal 3D ultrasound volumes, able to efficiently integrate semantic and structural information in the classification process. We introduce a new semantic features space able to encode spatial context via generalized geodesic distance transform. Unlike alternative auto-context approaches, this new set of features is efficiently integrated into the same forest using contextual trees. Finally, we use a new structured labels space as alternative to the traditional atomic class labels, able to capture morphological variability of the target organ. Here, we show the potential of this new general framework segmenting the skull in 3D fetal ultrasound volumes, significantly outperforming alternative random forest-based approaches.

Original languageEnglish
Title of host publicationFetal, Infant and Ophthalmic Medical Image Analysis
Subtitle of host publicationInternational Workshop, FIFI 2017 and 4th International Workshop, OMIA 2017 Held in Conjunction with MICCAI 2017, Proceedings
PublisherSpringer Verlag
Pages25-32
ISBN (Print)9783319675602
DOIs
Publication statusPublished - 9 Sept 2017
EventInternational Workshop on Fetal and Infant Image Analysis, FIFI 2017 and 4th International Workshop on Ophthalmic Medical Image Analysis, OMIA 2017 held in Conjunction with 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017 - Quebec City, Canada
Duration: 14 Sept 201714 Sept 2017

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10554 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Conference

ConferenceInternational Workshop on Fetal and Infant Image Analysis, FIFI 2017 and 4th International Workshop on Ophthalmic Medical Image Analysis, OMIA 2017 held in Conjunction with 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017
Country/TerritoryCanada
CityQuebec City
Period14/09/201714/09/2017

Keywords

  • Generalized geodesic distance
  • Random forest
  • Structured class

Fingerprint

Dive into the research topics of 'Fetal skull segmentation in 3D ultrasound via structured geodesic random forest'. Together they form a unique fingerprint.

Cite this