Learning based image segmentation of post-operative CT-images: A hydrocephalus case study

Venkateswararao Cherukuri, Peter Ssenyonga, Benjamin C. Warf, Abhaya V. Kulkarni, Vishal Monga, Steven Schiff

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

Accurate estimation of volumes for cerebrospinal fluid (CSF) and brain before and after surgery (pre-op and post-op) plays an important role in analyzing treatment for hydrocephalus. This in turn, relies upon segmentation of brain imagery into brain tissue and CSF. Segmentation of preop images is a relatively straightforward problem and has been well researched. However, segmenting post-op CT-scans becomes challenging due to distorted anatomy and subdural hematoma collections pressing on the brain. Most intensity and feature based segmentation methods fail to separate subdurals from brain and CSF as subdural geometry varies greatly across different patients and their intensity varies with time. We combat this problem by a learning approach that treats segmentation as supervised classification at the pixel level, i.e. a training set of CT scans with labeled pixel identities is employed. Inspired by sparsity constrained classification, our central contribution is a dictionary learning framework that learns class (segment) specific dictionaries that can efficiently represent test samples from the same class while poorly represent corresponding samples from other classes. Because discriminating features are discovered automatically, we call our method feature learning for image segmentation (FLIS). Experiments performed on infant CT brain images acquired from CURE children0s hospital of Uganda reveal the success of our method against existing alternatives.

Original languageEnglish (US)
Title of host publication8th International IEEE EMBS Conference on Neural Engineering, NER 2017
PublisherIEEE Computer Society
Pages13-16
Number of pages4
ISBN (Electronic)9781538619162
DOIs
StatePublished - Aug 10 2017
Event8th International IEEE EMBS Conference on Neural Engineering, NER 2017 - Shanghai, China
Duration: May 25 2017May 28 2017

Publication series

NameInternational IEEE/EMBS Conference on Neural Engineering, NER
ISSN (Print)1948-3546
ISSN (Electronic)1948-3554

Other

Other8th International IEEE EMBS Conference on Neural Engineering, NER 2017
CountryChina
CityShanghai
Period5/25/175/28/17

Fingerprint

Image segmentation
Brain
Cerebrospinal fluid
Computerized tomography
Glossaries
Pixels
Surgery
Tissue
Geometry
Experiments

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Mechanical Engineering

Cite this

Cherukuri, V., Ssenyonga, P., Warf, B. C., Kulkarni, A. V., Monga, V., & Schiff, S. (2017). Learning based image segmentation of post-operative CT-images: A hydrocephalus case study. In 8th International IEEE EMBS Conference on Neural Engineering, NER 2017 (pp. 13-16). [8008280] (International IEEE/EMBS Conference on Neural Engineering, NER). IEEE Computer Society. https://doi.org/10.1109/NER.2017.8008280
Cherukuri, Venkateswararao ; Ssenyonga, Peter ; Warf, Benjamin C. ; Kulkarni, Abhaya V. ; Monga, Vishal ; Schiff, Steven. / Learning based image segmentation of post-operative CT-images : A hydrocephalus case study. 8th International IEEE EMBS Conference on Neural Engineering, NER 2017. IEEE Computer Society, 2017. pp. 13-16 (International IEEE/EMBS Conference on Neural Engineering, NER).
@inproceedings{7c407ae9b6674134b083f44faea28874,
title = "Learning based image segmentation of post-operative CT-images: A hydrocephalus case study",
abstract = "Accurate estimation of volumes for cerebrospinal fluid (CSF) and brain before and after surgery (pre-op and post-op) plays an important role in analyzing treatment for hydrocephalus. This in turn, relies upon segmentation of brain imagery into brain tissue and CSF. Segmentation of preop images is a relatively straightforward problem and has been well researched. However, segmenting post-op CT-scans becomes challenging due to distorted anatomy and subdural hematoma collections pressing on the brain. Most intensity and feature based segmentation methods fail to separate subdurals from brain and CSF as subdural geometry varies greatly across different patients and their intensity varies with time. We combat this problem by a learning approach that treats segmentation as supervised classification at the pixel level, i.e. a training set of CT scans with labeled pixel identities is employed. Inspired by sparsity constrained classification, our central contribution is a dictionary learning framework that learns class (segment) specific dictionaries that can efficiently represent test samples from the same class while poorly represent corresponding samples from other classes. Because discriminating features are discovered automatically, we call our method feature learning for image segmentation (FLIS). Experiments performed on infant CT brain images acquired from CURE children0s hospital of Uganda reveal the success of our method against existing alternatives.",
author = "Venkateswararao Cherukuri and Peter Ssenyonga and Warf, {Benjamin C.} and Kulkarni, {Abhaya V.} and Vishal Monga and Steven Schiff",
year = "2017",
month = "8",
day = "10",
doi = "10.1109/NER.2017.8008280",
language = "English (US)",
series = "International IEEE/EMBS Conference on Neural Engineering, NER",
publisher = "IEEE Computer Society",
pages = "13--16",
booktitle = "8th International IEEE EMBS Conference on Neural Engineering, NER 2017",
address = "United States",

}

Cherukuri, V, Ssenyonga, P, Warf, BC, Kulkarni, AV, Monga, V & Schiff, S 2017, Learning based image segmentation of post-operative CT-images: A hydrocephalus case study. in 8th International IEEE EMBS Conference on Neural Engineering, NER 2017., 8008280, International IEEE/EMBS Conference on Neural Engineering, NER, IEEE Computer Society, pp. 13-16, 8th International IEEE EMBS Conference on Neural Engineering, NER 2017, Shanghai, China, 5/25/17. https://doi.org/10.1109/NER.2017.8008280

Learning based image segmentation of post-operative CT-images : A hydrocephalus case study. / Cherukuri, Venkateswararao; Ssenyonga, Peter; Warf, Benjamin C.; Kulkarni, Abhaya V.; Monga, Vishal; Schiff, Steven.

8th International IEEE EMBS Conference on Neural Engineering, NER 2017. IEEE Computer Society, 2017. p. 13-16 8008280 (International IEEE/EMBS Conference on Neural Engineering, NER).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Learning based image segmentation of post-operative CT-images

T2 - A hydrocephalus case study

AU - Cherukuri, Venkateswararao

AU - Ssenyonga, Peter

AU - Warf, Benjamin C.

AU - Kulkarni, Abhaya V.

AU - Monga, Vishal

AU - Schiff, Steven

PY - 2017/8/10

Y1 - 2017/8/10

N2 - Accurate estimation of volumes for cerebrospinal fluid (CSF) and brain before and after surgery (pre-op and post-op) plays an important role in analyzing treatment for hydrocephalus. This in turn, relies upon segmentation of brain imagery into brain tissue and CSF. Segmentation of preop images is a relatively straightforward problem and has been well researched. However, segmenting post-op CT-scans becomes challenging due to distorted anatomy and subdural hematoma collections pressing on the brain. Most intensity and feature based segmentation methods fail to separate subdurals from brain and CSF as subdural geometry varies greatly across different patients and their intensity varies with time. We combat this problem by a learning approach that treats segmentation as supervised classification at the pixel level, i.e. a training set of CT scans with labeled pixel identities is employed. Inspired by sparsity constrained classification, our central contribution is a dictionary learning framework that learns class (segment) specific dictionaries that can efficiently represent test samples from the same class while poorly represent corresponding samples from other classes. Because discriminating features are discovered automatically, we call our method feature learning for image segmentation (FLIS). Experiments performed on infant CT brain images acquired from CURE children0s hospital of Uganda reveal the success of our method against existing alternatives.

AB - Accurate estimation of volumes for cerebrospinal fluid (CSF) and brain before and after surgery (pre-op and post-op) plays an important role in analyzing treatment for hydrocephalus. This in turn, relies upon segmentation of brain imagery into brain tissue and CSF. Segmentation of preop images is a relatively straightforward problem and has been well researched. However, segmenting post-op CT-scans becomes challenging due to distorted anatomy and subdural hematoma collections pressing on the brain. Most intensity and feature based segmentation methods fail to separate subdurals from brain and CSF as subdural geometry varies greatly across different patients and their intensity varies with time. We combat this problem by a learning approach that treats segmentation as supervised classification at the pixel level, i.e. a training set of CT scans with labeled pixel identities is employed. Inspired by sparsity constrained classification, our central contribution is a dictionary learning framework that learns class (segment) specific dictionaries that can efficiently represent test samples from the same class while poorly represent corresponding samples from other classes. Because discriminating features are discovered automatically, we call our method feature learning for image segmentation (FLIS). Experiments performed on infant CT brain images acquired from CURE children0s hospital of Uganda reveal the success of our method against existing alternatives.

UR - http://www.scopus.com/inward/record.url?scp=85028611114&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028611114&partnerID=8YFLogxK

U2 - 10.1109/NER.2017.8008280

DO - 10.1109/NER.2017.8008280

M3 - Conference contribution

AN - SCOPUS:85028611114

T3 - International IEEE/EMBS Conference on Neural Engineering, NER

SP - 13

EP - 16

BT - 8th International IEEE EMBS Conference on Neural Engineering, NER 2017

PB - IEEE Computer Society

ER -

Cherukuri V, Ssenyonga P, Warf BC, Kulkarni AV, Monga V, Schiff S. Learning based image segmentation of post-operative CT-images: A hydrocephalus case study. In 8th International IEEE EMBS Conference on Neural Engineering, NER 2017. IEEE Computer Society. 2017. p. 13-16. 8008280. (International IEEE/EMBS Conference on Neural Engineering, NER). https://doi.org/10.1109/NER.2017.8008280