skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Registration of lung nodules using a semi-rigid model: Method and preliminary results

Abstract

The tracking of lung nodules across computed tomography (CT) scans acquired at different times for the same patient is helpful for the determination of malignancy. We are developing a nodule registration system to facilitate this process. We propose to use a semi-rigid method that considers principal structures surrounding the nodule and allows relative movements among the structures. The proposed similarity metric, which evaluates both the image correlation and the degree of elastic deformation amongst the structures, is maximized by a two-layered optimization method, employing a simulated annealing framework. We tested our method by simulating five cases that represent physiological deformation as well as different nodule shape/size changes with time. Each case is made up of a source and target scan, where the source scan consists of a nodule-free patient CT volume into which we inserted ten simulated lung nodules, and the target scan is the result of applying a known, physiologically based nonrigid transformation to the nodule-free source scan, into which we inserted modified versions of the corresponding nodules at the same, known locations. Five different modification strategies were used, one for each of the five cases: (1) nodules maintain size and shape, (2) nodules disappear, (3) nodules shrink uniformlymore » by a factor of 2, (4) nodules grow uniformly by a factor of 2, and (5) nodules grow nonuniformly. We also matched 97 real nodules in pairs of scans (acquired at different times) from 12 patients and compared our registration to a radiologist's visual determination. In the simulation experiments, the mean absolute registration errors were 1.0{+-}0.8 mm (s.d.), 1.1{+-}0.7 mm (s.d.), 1.0{+-}0.7 mm (s.d.), 1.0{+-}0.6 mm (s.d.), and 1.1{+-}0.9 mm (s.d.) for the five cases, respectively. For the 97 nodule pairs in 12 patient scans, the mean absolute registration error was 1.4{+-}0.8 mm (s.d.)« less

Authors:
; ; ; ; ;  [1]
  1. Department of Electrical Engineering and Department of Radiology, Stanford University, Stanford, California 94305 (United States)
Publication Date:
OSTI Identifier:
20951054
Resource Type:
Journal Article
Resource Relation:
Journal Name: Medical Physics; Journal Volume: 34; Journal Issue: 2; Other Information: DOI: 10.1118/1.2432073; (c) 2007 American Association of Physicists in Medicine; Country of input: International Atomic Energy Agency (IAEA)
Country of Publication:
United States
Language:
English
Subject:
62 RADIOLOGY AND NUCLEAR MEDICINE; ANNEALING; CARCINOMAS; COMPUTERIZED TOMOGRAPHY; DEFORMATION; ELASTICITY; ERRORS; IMAGE PROCESSING; IMAGES; LUNGS; MODIFICATIONS; OPTIMIZATION; PATIENTS; SIMULATION

Citation Formats

Sun Shaohua, Rubin, Geoffrey D., Paik, David, Steiner, Robert M., Zhuge Feng, and Napel, Sandy. Registration of lung nodules using a semi-rigid model: Method and preliminary results. United States: N. p., 2007. Web. doi:10.1118/1.2432073.
Sun Shaohua, Rubin, Geoffrey D., Paik, David, Steiner, Robert M., Zhuge Feng, & Napel, Sandy. Registration of lung nodules using a semi-rigid model: Method and preliminary results. United States. doi:10.1118/1.2432073.
Sun Shaohua, Rubin, Geoffrey D., Paik, David, Steiner, Robert M., Zhuge Feng, and Napel, Sandy. Thu . "Registration of lung nodules using a semi-rigid model: Method and preliminary results". United States. doi:10.1118/1.2432073.
@article{osti_20951054,
title = {Registration of lung nodules using a semi-rigid model: Method and preliminary results},
author = {Sun Shaohua and Rubin, Geoffrey D. and Paik, David and Steiner, Robert M. and Zhuge Feng and Napel, Sandy},
abstractNote = {The tracking of lung nodules across computed tomography (CT) scans acquired at different times for the same patient is helpful for the determination of malignancy. We are developing a nodule registration system to facilitate this process. We propose to use a semi-rigid method that considers principal structures surrounding the nodule and allows relative movements among the structures. The proposed similarity metric, which evaluates both the image correlation and the degree of elastic deformation amongst the structures, is maximized by a two-layered optimization method, employing a simulated annealing framework. We tested our method by simulating five cases that represent physiological deformation as well as different nodule shape/size changes with time. Each case is made up of a source and target scan, where the source scan consists of a nodule-free patient CT volume into which we inserted ten simulated lung nodules, and the target scan is the result of applying a known, physiologically based nonrigid transformation to the nodule-free source scan, into which we inserted modified versions of the corresponding nodules at the same, known locations. Five different modification strategies were used, one for each of the five cases: (1) nodules maintain size and shape, (2) nodules disappear, (3) nodules shrink uniformly by a factor of 2, (4) nodules grow uniformly by a factor of 2, and (5) nodules grow nonuniformly. We also matched 97 real nodules in pairs of scans (acquired at different times) from 12 patients and compared our registration to a radiologist's visual determination. In the simulation experiments, the mean absolute registration errors were 1.0{+-}0.8 mm (s.d.), 1.1{+-}0.7 mm (s.d.), 1.0{+-}0.7 mm (s.d.), 1.0{+-}0.6 mm (s.d.), and 1.1{+-}0.9 mm (s.d.) for the five cases, respectively. For the 97 nodule pairs in 12 patient scans, the mean absolute registration error was 1.4{+-}0.8 mm (s.d.)},
doi = {10.1118/1.2432073},
journal = {Medical Physics},
number = 2,
volume = 34,
place = {United States},
year = {Thu Feb 15 00:00:00 EST 2007},
month = {Thu Feb 15 00:00:00 EST 2007}
}
  • Purpose: To evaluate localization accuracy resulting from rigid registration of locally-advanced lung cancer targets using fully automatic and semi-automatic protocols for image-guided radiation therapy. Methods: Seventeen lung cancer patients, fourteen also presenting with involved lymph nodes, received computed tomography (CT) scans once per week throughout treatment under active breathing control. A physician contoured both lung and lymph node targets for all weekly scans. Various automatic and semi-automatic rigid registration techniques were then performed for both individual and simultaneous alignments of the primary gross tumor volume (GTV{sub P}) and involved lymph nodes (GTV{sub LN}) to simulate the localization process in image-guidedmore » radiation therapy. Techniques included ''standard'' (direct registration of weekly images to a planning CT), ''seeded'' (manual prealignment of targets to guide standard registration), ''transitive-based'' (alignment of pretreatment and planning CTs through one or more intermediate images), and ''rereferenced'' (designation of a new reference image for registration). Localization error (LE) was assessed as the residual centroid and border distances between targets from planning and weekly CTs after registration. Results: Initial bony alignment resulted in centroid LE of 7.3 {+-} 5.4 mm and 5.4 {+-} 3.4 mm for the GTV{sub P} and GTV{sub LN}, respectively. Compared to bony alignment, transitive-based and seeded registrations significantly reduced GTV{sub P} centroid LE to 4.7 {+-} 3.7 mm (p = 0.011) and 4.3 {+-} 2.5 mm (p < 1 x 10{sup -3}), respectively, but the smallest GTV{sub P} LE of 2.4 {+-} 2.1 mm was provided by rereferenced registration (p < 1 x 10{sup -6}). Standard registration significantly reduced GTV{sub LN} centroid LE to 3.2 {+-} 2.5 mm (p < 1 x 10{sup -3}) compared to bony alignment, with little additional gain offered by the other registration techniques. For simultaneous target alignment, centroid LE as low as 3.9 {+-} 2.7 mm and 3.8 {+-} 2.3 mm were achieved for the GTV{sub P} and GTV{sub LN}, respectively, using rereferenced registration. Conclusions: Target shape, volume, and configuration changes during radiation therapy limited the accuracy of standard rigid registration for image-guided localization in locally-advanced lung cancer. Significant error reductions were possible using other rigid registration techniques, with LE approaching the lower limit imposed by interfraction target variability throughout treatment.« less
  • We are developing a computer-aided detection system to assist radiologists in the detection of lung nodules on thoracic computed tomography (CT) images. The purpose of this study was to improve the false-positive (FP) reduction stage of our algorithm by developing features that extract three-dimensional (3D) shape information from volumes of interest identified in the prescreening stage. We formulated 3D gradient field descriptors, and derived 19 gradient field features from their statistics. Six ellipsoid features were obtained by computing the lengths and the length ratios of the principal axes of an ellipsoid fitted to a segmented object. Both the gradient fieldmore » features and the ellipsoid features were designed to distinguish spherical objects such as lung nodules from elongated objects such as vessels. The FP reduction performance in this new 25-dimensional feature space was compared to the performance in a 19-dimensional space that consisted of features extracted using previously developed methods. The performance in the 44-dimensional combined feature space was also evaluated. Linear discriminant analysis with stepwise feature selection was used for classification. The parameters used for feature selection were optimized using the simplex algorithm. Training and testing were performed using a leave-one-patient-out scheme. The FP reduction performances in different feature spaces were evaluated by using the area A{sub z} under the receiver operating characteristic curve and the number of FPs per CT section at a given sensitivity as accuracy measures. Our data set consisted of 82 CT scans (3551 axial sections) from 56 patients with section thickness ranging from 1.0 to 2.5 mm. Our prescreening algorithm detected 111 of the 116 solid nodules (nodule size: 3.0-30.6 mm) marked by experienced thoracic radiologists. The test A{sub z} values were 0.95{+-}0.01, 0.88{+-}0.02, and 0.94{+-}0.01 in the new, previous, and combined feature spaces, respectively. The number of FPs per section at 80% sensitivity in these three feature spaces were 0.37, 1.61, and 0.34, respectively. The improvement in the test A{sub z} with the 25 new features was statistically significant (p<0.0001) compared to that with the previous 19 features alone.« less
  • Purpose: To investigate image-registration errors when using fiducial markers with a manual method and the point-based rigid-body registration (PRBR) algorithm in accelerated partial breast irradiation (APBI) patients, with accompanying fiducial deviations. Methods: Twenty-two consecutive patients were enrolled in a prospective trial examining 10-fraction APBI. Titanium clips were implanted intraoperatively around the seroma in all patients. For image-registration, the positions of the clips in daily kV x-ray images were matched to those in the planning digitally reconstructed radiographs. Fiducial and gravity registration errors (FREs and GREs, respectively), representing resulting misalignments of the edge and center of the target, respectively, were comparedmore » between the manual and algorithm-based methods. Results: In total, 218 fractions were evaluated. Although the mean FRE/GRE values for the manual and algorithm-based methods were within 3 mm (2.3/1.7 and 1.3/0.4 mm, respectively), the percentages of fractions where FRE/GRE exceeded 3 mm using the manual and algorithm-based methods were 18.8%/7.3% and 0%/0%, respectively. Manual registration resulted in 18.6% of patients with fractions of FRE/GRE exceeding 5 mm. The patients with larger clip deviation had significantly more fractions showing large FRE/GRE using manual registration. Conclusions: For image-registration using fiducial markers in APBI, the manual registration results in more fractions with considerable registration error due to loss of fiducial objectivity resulting from their deviation. The authors recommend the PRBR algorithm as a safe and effective strategy for accurate, image-guided registration and PTV margin reduction.« less
  • Objective: Irreversible electroporation (IRE) uses direct electrical pulses to create permanent 'pores' in cell membranes to cause cell death. In contrast to conventional modalities, IRE has a nonthermal mechanism of action. Our objective was to study the histopathological and imaging features of IRE in normal swine lung. Materials and Methods: Eleven female swine were studied for hyperacute (8 h), acute (24 h), subacute (96 h), and chronic (3 week) effects of IRE ablation in lung. Paired unipolar IRE applicators were placed under computed tomography (CT) guidance. Some applicators were deliberately positioned near bronchovascular structures. IRE pulse delivery was synchronized withmore » the cardiac rhythm only when ablation was performed within 2 cm of the heart. Contrast-enhanced CT scan was performed immediately before and after IRE and at 1 and 3 weeks after IRE ablation. Representative tissue was stained with hematoxylin and eosin for histopathology. Results: Twenty-five ablations were created: ten hyperacute, four acute, and three subacute ablations showed alveolar edema and necrosis with necrosis of bronchial, bronchiolar, and vascular epithelium. Bronchovascular architecture was maintained. Chronic ablations showed bronchiolitis obliterans and alveolar interstitial fibrosis. Immediate post-procedure CT images showed linear or patchy density along the applicator tract. At 1 week, there was consolidation that resolved partially or completely by 3 weeks. Pneumothorax requiring chest tube developed in two animals; no significant cardiac arrhythmias were noted. Conclusion: Our preliminary porcine study demonstrates the nonthermal and extracellular matrix sparing mechanism of action of IRE. IRE is a potential alternative to thermal ablative modalities.« less