Thoracic abnormality detection with data adaptive structure estimation

Yang Song, Weidong Cai, Yun Zhou, Dagan Feng

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

7 Scopus citations

Abstract

Automatic detection of lung tumors and abnormal lymph nodes are useful in assisting lung cancer staging. This paper presents a novel detection method, by first identifying all abnormalities, then differentiating between lung tumors and abnormal lymph nodes based on their degree of overlap with the lung field and mediastinum. Regression-based appearance model and graph-based structure labeling are designed to estimate the actual lung field and mediastinum from the pathology-affected thoracic images adaptively. The proposed method is simple, effective and generalizable, and can be potentially applicable to other medical imaging domains as well. Promising results are demonstrated based on our evaluations on clinical PET-CT data sets from lung cancer patients.

Original languageEnglish
Title of host publicationMedical Image Computing and Computer-Assisted Intervention, MICCAI2012 - 15th International Conference, Proceedings
EditorsNicholas Ayache, Herve Delingette, Polina Golland, Kensaku Mori
PublisherSpringer Verlag
Pages74-81
Number of pages8
ISBN (Print)9783642334146
DOIs
StatePublished - 2012
Event15th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2012 - Nice, France
Duration: Oct 1 2012Oct 5 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7510 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2012
Country/TerritoryFrance
CityNice
Period10/1/1210/5/12

Fingerprint

Dive into the research topics of 'Thoracic abnormality detection with data adaptive structure estimation'. Together they form a unique fingerprint.

Cite this