AJR AJR Integrative Imaging Dec 2008 articles
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Magnussen, J. S.
Right arrow Articles by Vu, D. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Magnussen, J. S.
Right arrow Articles by Vu, D. H.
Social Bookmarking
 Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?
AJR 2000; 174:1333-1336
© American Roentgen Ray Society


Creation of a Three-Dimensional Model of Human Segmental Lung Anatomy

John S. Magnussen1, Peter Chicco1, Amanda W. Palmer2, Hans Van der Wall1 and Dzung H. Vu3

1 Department of Nuclear Medicine, Concord Hospital, Hospital Rd., Concord, New South Wales 2139, Australia.
2 Department of Radiology, St. George Hospital, Belgrave St., Kogarah 2217, Australia.
3 School of Anatomy, University of New South Wales, Anzac Pde., Kensington 2052, Australia.

Received July 19, 1999; accepted after revision October 21, 1999.

 
Supported by research grants JM000103 and PC000104 from the Nuclear Medicine Research Foundation, Sydney, Australia.

Address correspondence to J. S. Magnussen.


Abstract
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References
 
OBJECTIVE. The investigation of pulmonary embolism using scintigraphic tomography requires a model of the internal architecture of the segments and subsegments in the human lung. Such a model has been developed by the segmentation and subsegmentation of an existing whole-body tissue-segmented phantom.

MATERIALS AND METHODS. By using information from suitably windowed human axial CT scans, combined with the information gained from the injection of color-coded dyes into the segmental bronchi of human cadaveric lungs, the lobar and segmental boundaries were added to the existing phantom. Further refinements were added from reports in the literature regarding the predominant pattern of subsegmental bronchi in a series of human cadavers, enabling the creation of subsegmental boundaries.

RESULTS. A digitized model of the segmental and subsegmental anatomy of the human lung was successfully created. External, or pleural, projections of the complex internal arrangement of the segments closely corresponded with the projections of the best available authorities on the subject.

CONCLUSION. The model provides the opportunity to address several issues germane to scintigraphy and important for diagnosing pulmonary embolic disease. In particular, the model allows the manipulation of three-dimensional data sets to explore issues of importance to tomographic lung scanning.


Introduction
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References
 
In the past, knowledge of the simple surface projections of the segmental anatomy of the lungs has been adequate for the scintigraphic investigation of pulmonary thromboembolic disease [1, 2]. As greater sophistication in imaging has become available, pressure has increased to use tomographic cross-sectional imaging in an attempt to improve the sensitivity and specificity of tomography over those of planar imaging. Therefore, the availability of a coherent three-dimensional model of the arrangement of segments and subsegments in the human lung is of paramount interest. Such a model is needed to identify the segmental anatomy before a disease process such as pulmonary thromboembolism can be scintigraphically investigated in a clinical setting [1, 2].

Voxel-based anthropomorphic phantoms exist that could be used for clinical therapy planning [3]; however, these phantoms cannot be used to address questions regarding pulmonary embolic disease because adjacent segments within the lungs have not been differentiated in the models. In the existing models, organs are treated as homogeneous structures that lack internal subdivisions; these subdivisions are critical when considering the pathophysiology of embolic disease.

Zubal et al. [3, 4] created a tissue-segmented phantom based on transverse slices of CT data from a living human man whose height and weight were similar to the dosimetry standard mathematic phantom [5]. The phantom of Zubal et al. has the potential to transpose and encode pulmonary segmental and subsegmental divisions. Using the phantom as a basis for such delineation, we undertook a study to appropriately subdivide the lungs to create a segmental lung chart in surface views and axial cross-sections.

Instead of trying to represent the large variability in both the external shape and the internal arrangement of segments in human lungs, the intention of our study was to construct a plausible and representative model—that is, a model that can be used to draw general conclusions rather than one that attempts to apply to individuals in a population that is inherently variable.


Materials and Methods
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References
 
After preparation of the data from the anthropomorphic phantom created by Zubal et al. [3], subdivision of the lungs took place in several phases. These included the delineation of lobes using a suitably windowed CT scan from a human, the incorporation of segments using information from lung sections [6, 7], and appropriate anatomic reference texts [8,9,10].

Data Preparation
All data processing and image manipulation was performed on an Intel processor (Intel, Santa Clara, CA) workstation running Windows NT version 3.51 service pack 5 (Microsoft, Redmond, WA).

The anthropomorphic phantom of Zubal et al. [3] consists of a volume array of 246 axial slices (each slice, 128 x 128 voxels). Each voxel contained an integer index value representing the tissue type and a volume of 4 mm3. A subset of the phantom was created for use in later simulations. This subset contained only the thorax and resulted in a smaller data set, 128 x 128 x 64, which was used throughout the rest of the study.

A set of 64 slices, each 128 x 128, was then created with the lung tissue set to a value of one, and all other tissue types set to zero. Photoshop version 3.04 (Adobe, Mountain View, CA) was used for the image processing necessary to segment and subsegment the slices.

Manual interpolation was required to remove slice duplication present in the phantom of Zubal et al. [3]. Care was taken to ensure that the boundaries of the lung tissue did not overlap with other tissue types and that the total lung volume remained constant.

Throughout the segmentation and subsegmentation process, the adopted nomenclature was that of Jackson and Huber [11] for the segments and that of Ikeda [12] for the subsegments, as used in the work of Netter [13]. The nomenclature differs from that used by Boyden [10] for several of the segments, but is used more widely than Boyden's nomenclature [14,15,16,17].

Segmentation
An experienced radiologist marked the interlobar fissures that were visible on a CT scan of the thorax with normal findings, windowed for lung parenchymal tissue. This scan consisted of 26 slices at 10-mm intervals, with a total height of lung tissue of 260 mm, from a man who was 54 years old, 182-cm tall, and 73 kg. The approximate locations of the segmental boundaries were marked on the slices of the CT scan with reference to anatomic texts [13]. One slice from the CT scan was calculated to correspond with two from the phantom of Zubal et al. [3].

The CT scan was used for external landmarks, lobar fissures, and the boundaries transferred to the reduced data set of Zubal et al. [3]. Results obtained from the cadaveric human lung sections [6, 7] were used to match the interlobar fissures and segmental boundaries that were subsequently transferred to the reduced data set of Zubal et al.

The area within the segmental boundaries of each slice was filled with a unique color in Photoshop. Manual interpolation of the lobar and segmental boundaries was performed to ensure each was unique and provided a plausible and smooth transition between areas.

Iterative corrections were made to the segmented data set to more closely resemble the external landmarks described by Netter [13]. A set of six external views of the lungs, anterior, posterior, both lateral, and both medial projections, were created for visual comparison with the work of Netter.

Subsegmentation
Subsegmentation was performed on a segment-by-segment basis, using information from studies of the pattern of subsegments present in cadaveric human lungs [8,9,10]. In each case, the predominant pattern present in the examined lungs was used as the basis for partitioning the segments within the previously established boundaries.

Initially, the number of subsegments in each segment was determined, taking into consideration the overall size of the segment and the predominant anatomic pattern. Thus, larger segments would be divided by the equivalent of a further generation of bronchial subdivisions.

The spatial distribution of the division of each segment was found to be possible in only six different ways, almost completely defined by the anatomic location of the segment, its boundaries, and the predominant pattern of branching of the higher order bronchi and bronchioles (Fig. 1). Rotation of the six different patterns of subdivisions accounted for all the segments.



View larger version (18K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 1. —Diagram shows pattern of segmental subdivision used in creation of reduced Zubal data set [3] for each number of subdivisions. At left is number of subsegments within each segment. To right are possible patterns of division.

 

The subsegmental boundaries were added to the individual slices in the data set of Zubal et al. [3] and the resulting boundaries were filled with unique colors in Photoshop that were chosen to represent the subsegments. Views of the data set were used to iteratively improve the subsegmentation to more closely match the predominant patterns reported in the literature and to improve the consistency of external boundaries, which were otherwise difficult to maintain when manipulating individual slices.


Results
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References
 
External views of the reduced data set of Zubal et al. [3] are displayed in the normal axes with both the lateral and medial views of each lung (Fig. 2). Oblique views were used during the process of iterative correction of segmental and subsegmental surface projections. Figure 2 shows the three stages of the segmentation process, with the inclusion of the segmental and subsegmental boundaries in the lower rows.



View larger version (39K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 2. —Diagram shows anterior, posterior, medial, lateral, and both oblique views for each lung. Also shown are internal views of each lung (medial views), revealing those parts hidden by contralateral lung, and significant when considering shine-through activity. In each case, lung of interest has lobar and segmental boundaries darkened. Note that reduced Zubal data set [3] has been printed at its original resolution, without smoothing, giving ragged appearance to edges.

 

A small selection of axial slices through the reduced data set of Zubal et al. [3] shows the three stages of the segmentation process with segmental and subsegmental boundaries included in the lower rows (Fig. 3). A complex arrangement of segments and subsegments is apparent within the lower lobes in particular, because of the density of segments. This highlights the presence of two radially arranged rows of segments in the right lower lobe, in comparison with the single row in the base of the left lung.



View larger version (30K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 3. —Diagram shows three axial slices (numbered superior to inferior), chosen for depiction of some complexities in arrangement of segments and subsegments, particularly in lower lobes. In top row, lung outlines are darkened and lobar boundaries are shown. In middle row, lobar boundaries are darkened and segmental boundaries added. In bottom row, segmental boundaries are darkened and subsegmental boundaries added.

 

The left lung was found to occupy 52% and the right to occupy 48% of the total volume of lung tissue.


Discussion
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References
 
Our model gives a plausible example of the distribution of the most common variation seen for each segment, which has been transformed to conform to the boundaries imposed by the derivation of the data set from an actual human. The model does not attempt to represent all possible variations on the segmental and subsegmental anatomy of the human lungs.

Visual comparison of the current model with the segmental boundaries presented in the report by Netter [13] shows good correspondence (Fig. 2). The nature of the three-dimensional data set was such that only variations within narrow limits produced segmental boundaries approximating those shown by Netter and other anatomic texts [14,15,16] as the predominant pattern. Iterative corrections rapidly converged on the patterns required.

In no case was the subsegmentation taken past a second generation of airway branching. Consideration of the size of subsegments in contrast to their generation is reasonable because the relationship between the size of the bronchus and the volume of lung tissue supplied remained largely constant. This constancy is despite the imposition of a rigid nomenclature and should not be taken to imply that all subsegments of equal order are necessarily of equal volume.

Examination of the model showed that a number of segments, particularly those in the lower lobes, never extend to the pleural surface of the lungs. Such a situation was implied by the anatomic distribution of segments and their predominant pattern of branching [10].

Unfortunately, the current model represents neither all variations on segmental and subsegmental anatomy nor the gross variations in size and shape of the human thorax and the variations that would be present as a result of different disease states, such as hyperinflation with emphysema or displacement of the lung volume by gross cardiomegaly.

Despite the limitations, our model could be applied in many ways to the interpretation of scintigraphic images and could be a valuable resource for teaching segmental and subsegmental anatomy and the cross-sectional anatomy of the lungs. The use of our model could be further extended to the interpretation of planar and tomographic scintigraphic imaging; currently, no standard for the interpretation of tomographic perfusion images exists. The clinical usefulness of the model is illustrated in Figure 4, where the segments involved by pulmonary embolism are clearly delineated. This model could help define the patterns of segmental and subsegmental perfusion defects.



View larger version (75K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 4. —Clinical example of tomographic perfusion study in 62-year-old woman with multiple pulmonary emboli. Note loss of perfusion to virtually all segments in right lower lobe, as shown in accompanying axial wireframe diagram, and to posterobasal segment of left lower lobe. Perfusion defects are indicated by arrowheads.

 

A number of crucial problems have been identified in both planar and tomographic scintigraphic imaging that could potentially be solved by the judicious use of such a model. These problems include the investigation of the usefulness of the various planar projections [18], planar versus tomographic lung imaging and the nature of attenuation, and scatter and shine-through activity in pulmonary scintigraphic imaging [19].

In conclusion, a methodology was developed for the segmentation and subsegmentation of the lungs within an existing anthropomorphic whole-body phantom. This proceeded in several stages, introducing first the lobar, then the segmental, and finally the subsegmental boundaries.

Reference to and comparison with the literature suggests that this methodology has provided a plausible and consistent example of segmentation and subsegmentation, designed to represent a majority of the population. This model can by no means represent every individual, but will nonetheless prove useful for interpreting planar and tomographic scintigraphic images of the lungs.


References
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References
 

  1. Gottschalk A, Juni JE, Sostman HD, et al. Ventilation-perfusion scintigraphy in the PIOPED study. I. Data collection and tabulation. J Nucl Med 1993;34:1109 -1118[Abstract/Free Full Text]
  2. Gottschalk A, Sostman HD, Coleman RE, et al. Ventilation-perfusion scintigraphy in the PIOPED study. II. Evaluation of the scintigraphic criteria and interpretations. J Nucl Med 1993;34:1119 -1126[Abstract/Free Full Text]
  3. Zubal IG, Harrell CR, Smith EO, Rattner Z, Gindi G, Hoffer PB. Computerized three-dimensional segmented human anatomy. Med Phys 1994;21:299 -302[Medline]
  4. Zubal IG, Gindi G, Lee M, Harrell H, Smith E. High resolution anthropomorphic phantom for monte carlo analysis of internal radiation sources. In: Proceedings of the 3rd annual IEEE symposium on computer-based medical systems. Chapel Hill, NC: IEEE Computer Society Press, 1990:540 -546
  5. Snyder WS, Ford MR, Warner GG. Estimates of specific absorbed fractions for photon sources uniformly distributed in various organs of a heterogeneous phantom. MIRD (Medical Internal Radiation Dose) pamphlet number 5, revised. New York: Society of Nuclear Medicine 1978: 5
  6. Magnussen JS, Chicco P, Palmer AW, Mackey DJ, Van der Wall H, Murray IPC. Creation of a digital phantom for the monte carlo simulation of the scintigraphic segmental anatomy of the lungs (abstr). J Nucl Med 1997;38:211P
  7. Magnussen J, Chicco P, Vu D, Pandey P, D'Mello A. A study of sectional segmental anatomy of the human lung using gelatin embedded sections and CT imaging techniques. In: Proceedings of the 1996 conference of the Anatomical Society of Australia and New Zealand. Sydney, Australia: Anatomical Society of Australia and New Zealand, 1996: 22
  8. Berg RM, Boyden EA, Smith FR. An analysis of variations of the segmental bronchi of the left lower lobe of fifty dissected, and ten injected, lungs. J Thorac Surg 1949;18:216 -236[Medline]
  9. Smith FR, Boyden EA. An analysis of variations of the segmental bronchi of the right lower lobe of fifty injected lungs. J Thorac Surg 1949; 18:195 -215[Medline]
  10. Boyden EA. Segmental anatomy of the lungs: a study of the patterns of the segmental bronchi and related pulmonary vessels. New York: McGraw-Hill, 1955
  11. Jackson C, Huber JF. Correlated applied anatomy of the bronchial tree and lungs with a system of nomenclature. Dis Chest 1943;9:319 -326
  12. Ikeda S. Atlas of flexible bronchofiberoscopy. Baltimore: University Park, 1974
  13. Netter FH. Thorax. In: Colacino S, ed. Atlas of human anatomy. Summit, NJ: Ciba-Geigy, 1995:167 -230
  14. Last RJ. Thorax. In: McMinn RMH, ed. Last's anatomy: regional and applied, 9th ed. Edinburgh: Churchill Livingstone, 1995: 241-294
  15. Moore KL. The thorax. In: Clinically oriented anatomy, 2nd ed. Baltimore: Williams & Wilkins, 1985: 49-148
  16. Parker R, Joseph J, Dussek J, Dyson M, Gabella G. Respiratory system. In: Williams PL, Bannister LH, Berry MM, et al., eds. Gray's anatomy, 38th ed. New York: Churchill Livingstone, 1999:1627 -1682
  17. Agur AMR, Lee MJ. The thorax. In: Grant's atlas of anatomy, 9th ed. Baltimore: Williams & Wilkins, 1991: 1-76
  18. Magnussen JS, Chicco P, Palmer AW, et al. Optimization of the scintigraphic segmental anatomy of the lungs. J Nucl Med 1997;38:1987 -1991[Abstract/Free Full Text]
  19. Magnussen JS, Chicco P, Mackey DW, Murray IP, Van der Wall H. Simulation of the stripe sign in a scintigraphic model of the lungs. Nucl Med Commun 1997;18:648 -654[Medline]

Add to CiteULike CiteULike   Add to Complore Complore   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us   Add to Digg Digg   Add to Reddit Reddit   Add to Technorati Technorati    What's this?


This article has been cited by other articles:


Home page
Am. J. Roentgenol.Home page
A. Kluge, W. Luboldt, and G. Bachmann
Acute pulmonary embolism to the subsegmental level: diagnostic accuracy of three MRI techniques compared with 16-MDCT.
Am. J. Roentgenol., July 1, 2006; 187(1): W7 - 14.
[Abstract] [Full Text] [PDF]


Home page
Am. J. Roentgenol.Home page
D. T. Boll, R. C. Gilkeson, T. R. Fleiter, K. A. Blackham, J. L. Duerk, and J. S. Lewin
Volumetric Assessment of Pulmonary Nodules with ECG-Gated MDCT
Am. J. Roentgenol., November 1, 2004; 183(5): 1217 - 1223.
[Abstract] [Full Text] [PDF]


This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Magnussen, J. S.
Right arrow Articles by Vu, D. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Magnussen, J. S.
Right arrow Articles by Vu, D. H.
Social Bookmarking
 Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS