Detection of Time-Varying Structures by Large Deformation Diffeomorphic Metric Mapping to Aid Reading of High-Resolution CT Images of the Lung

Ryo Sakamoto,Kaori Togashi,Michael I Miller,Susumu Mori,Tomohisa Okada,Zhaohua Ding

doi:10.1371/journal.pone.0085580

Abstract

ObjectivesTo evaluate the accuracy of advanced non-linear registration of serial lung Computed Tomography (CT) images using Large Deformation Diffeomorphic Metric Mapping (LDDMM).MethodsFifteen cases of lung cancer with serial lung CT images (interval: 62.2±26.9 days) were used. After affine transformation, three dimensional, non-linear volume registration was conducted using LDDMM with or without cascading elasticity control. Registration accuracy was evaluated by measuring the displacement of landmarks placed on vessel bifurcations for each lung segment. Subtraction images and Jacobian color maps, calculated from the transformation matrix derived from image warping, were generated, which were used to evaluate time-course changes of the tumors.ResultsThe average displacement of landmarks was 0.02±0.16 mm and 0.12±0.60 mm for proximal and distal landmarks after LDDMM transformation with cascading elasticity control, which was significantly smaller than 3.11±2.47 mm and 3.99±3.05 mm, respectively, after affine transformation. Emerged or vanished nodules were visualized on subtraction images, and enlarging or shrinking nodules were displayed on Jacobian maps enabled by highly accurate registration of the nodules using LDDMM. However, some residual misalignments were observed, even with non-linear transformation when substantial changes existed between the image pairs.ConclusionsLDDMM provides accurate registration of serial lung CT images, and temporal subtraction images with Jacobian maps help radiologists to find changes in pulmonary nodules.

Highlights

The recent advent of radiological imaging devices has led to an overwhelming amount of anatomical information, which often exceeds the ability of radiologists to inspect within a reasonable reading time
The average displacement was 3.1162.47 mm and 3.9963.05 mm, respectively, for proximal and distal landmarks on the affine-transformed second time point images, which decreased to 0.1560.46 mm and 0.6261.27 mm, respectively, after single Large Deformation Diffeomorphic Metric Mapping (LDDMM) transformation, and 0.0260.16 mm and 0.1260.60 mm, respectively, after cascading LDDMM transformation
Almost none of the landmarks required repositioning after cascading LDDMM transformation

Summary

Introduction

The recent advent of radiological imaging devices has led to an overwhelming amount of anatomical information, which often exceeds the ability of radiologists to inspect within a reasonable reading time. For detection and monitoring of a tumor, MDCT is often repeated, which further multiplies the amount of anatomical information. This ample information has not been well exploited. Computer-aided detection (CAD) and quantification of time-dependent anatomical changes are highly desirable. The automated detection of tissue shape change is conceptually straightforward; images from two time points are three-dimensionally registered and a subtraction image is generated [2,3]. Among the organs in the human torso areas, the lung is one of the simplest, and the most researched organs for such automated detection of lesions and their anatomical changes. Precise registration of the lung remains an elusive goal [4,5]

Objectives

Methods

Results

Conclusion