Lung Segmentation in 4D CT Volumes Based on Robust Active Shape Model Matching.

Gurman Gill,Reinhard R Beichel

doi:10.1155/2015/125648

Abstract

Dynamic and longitudinal lung CT imaging produce 4D lung image data sets, enabling applications like radiation treatment planning or assessment of response to treatment of lung diseases. In this paper, we present a 4D lung segmentation method that mutually utilizes all individual CT volumes to derive segmentations for each CT data set. Our approach is based on a 3D robust active shape model and extends it to fully utilize 4D lung image data sets. This yields an initial segmentation for the 4D volume, which is then refined by using a 4D optimal surface finding algorithm. The approach was evaluated on a diverse set of 152 CT scans of normal and diseased lungs, consisting of total lung capacity and functional residual capacity scan pairs. In addition, a comparison to a 3D segmentation method and a registration based 4D lung segmentation approach was performed. The proposed 4D method obtained an average Dice coefficient of 0.9773 ± 0.0254, which was statistically significantly better (p value ≪0.001) than the 3D method (0.9659 ± 0.0517). Compared to the registration based 4D method, our method obtained better or similar performance, but was 58.6% faster. Also, the method can be easily expanded to process 4D CT data sets consisting of several volumes.

Highlights

Applications like lung cancer radiotherapy planning [1], assessment of lung diseases like COPD [2], or dynamic lung ventilation studies [3] require the acquisition and subsequent analysis of 4D lung CT scans
3D segmentation methods have been developed to deal with this issue, including approaches that utilize an atlasbased segmentation-by-registration scheme [11], an errorcorrecting hybrid system [12], a shape “break-and-repair” strategy [13], and a 3D robust active shape model (RASM) [14, 15]
None of these approaches takes advantage of 4D lung CT scans and requires lungs to be segmented individually. This can be problematic, especially when segmenting pairs of total lung capacity (TLC) and functional residual capacity (FRC) lung scans, because lungs at FRC are typically more difficult to segment than lungs imaged at TLC

Summary

Introduction

Applications like lung cancer radiotherapy planning [1], assessment of lung diseases like COPD [2], or dynamic lung ventilation studies [3] require the acquisition and subsequent analysis of 4D lung CT scans (e.g., two lung scans at different respiratory states). In order to achieve accurate results and reduce computation time, registration is typically only performed within a lung mask For such approaches, the segmentation of each lung CT volume acquired is a prerequisite. 3D segmentation methods have been developed to deal with this issue, including approaches that utilize an atlasbased segmentation-by-registration scheme [11], an errorcorrecting hybrid system [12], a shape “break-and-repair” strategy [13], and a 3D robust active shape model (RASM) [14, 15] None of these approaches takes advantage of 4D lung CT scans and requires lungs to be segmented individually. Algorithms that simultaneously segment lungs in all available CT volumes are more promising

Methods

Results

Discussion

Conclusion