Visualisation of Mapped Lymphoscintigraphy Data

Get Complete Project Material File(s) Now! »

Head and neck

Melanoma sites on the head and neck have been recorded on four out of the six body maps (as shown in Figure 1.14). Anterior and posterior views are given on the torso body maps 1 and 2, while lateral views are drawn on maps 5 and 6. The method utilised to map these melanoma sites onto the skin model is similar to that used for the torso and leg sites (as described in Section 3.1.1). The head and neck of each appropriate body map has initially been host-mesh fitted to the outline of the skin model, and then a direct orthogonal projection has been carried out. As mentioned previously, the torso melanoma sites had been projected onto the skin model before a head and neck model was available. Therefore the head and neck outlines on the anterior and posterior torso body maps have first been separated from the rest of the body before fitting.

Mapping the sentinel nodes

In addition to mapping primary melanoma sites onto the skin model, each patient’s SN fields have been mapped onto the lymph node model created in Section 2.6. The location of each patient’s SNs have been recorded by SMU clinicians as detailed in Section 1.6.2. In the absence of exact locations for each SN, the detailed lymph node model shown in Figure 2.20 has been reduced to one representative node in each field, onto which the SN field locations could be mapped. Figure 3.13 shows this reduced model, where node fields have been labelled according to their codes as listed in Appendix A. Node fields in the head and neck can be seen in Figures 3.13(a) and 3.13(b). The original model of these node fields were based on VH data and were therefore not in the correct space relative to the head and neck of the skin model.

Discussion

There are a number of errors that were created by the mapping process, which are important to be aware of during subsequent data analysis. Error produced by interpolating coordinates from 2D to 3D have been described above (and schematically demonstrated in Figure 3.5). In addition to this error, an aliasing error was also inherent in the grid maps, due to using grid squares to record melanoma sites rather than a continuous field. Melanoma (X, Y ) coordinates were located at the centre of each grid unit, however if these sites had been positioned in one of the four grid unit corners instead, the resultant mapped coordinates could be significantly different. The centralised coordinates however, represent an average location for each melanoma site and were therefore the most accurate position to use for projecting onto the skin model.

Visualisation Methodology

The previous chapter detailed mapping methods used to transform the SMU’s full LS database from 2D images onto the 3D anatomical model created in Chapter 2. This mapped model now provides the framework for improving visualisation of skin lymphatic drainage in 3D. This chapter outlines the mapping methods implemented to display data density using the SMU’s random coordinate approach described in Section 1.6.1. The limitations of this approach are outlined, and serve as a motivation for the new data display methods proposed. Field fitting is introduced as a technique to display lymphatic drainage patterns from any area of skin to specified node fields. Sobolev smoothing is discussed as a constraint which aids the fitting process. Finally, development of an interactive skin selection tool is presented, which allows predictions of SN fields from any element on the skin model.

ndom melanoma coordinates

Random melanoma sites (randX, randY ) located on the torso, lower and upper limb, head and neck have each been mapped with the same methods used for the original melanoma sites (X, Y ), as described in Sections 3.1.1, 3.1.2 and 3.1.4. During mapping some of these sites did not project correctly onto the 3D skin model however, because they were located just outside the body outlines drawn on the body maps. This occurred at coordinates which were located within grid squares containing the body outline. Although the original coordinate was located within the body outline, an offset random coordinate was not necessarily within this boundary. This can be clearly seen for a number of sites on the top of the head in Figure 1.15(b). To ensure that these random coordinates were adequately mapped onto the skin model, they have been moved manually within the body map outlines before projection, while still ensuring that they remained within the same grid unit.

Contents :

• Abstract
• Dedication
• Acknowledgements
• List of Figures
• List of Tables
• Glossary of Symbols
• Glossary of Acronyms
• 1 Introduction
  • 1.1 Motivation
  • 1.2 Cutaneous melanoma
  • 1.2.1 Staging
  • 1.2.2 Treatment
  • 1.3 Lymphatic anatomy
  • 1.3.1 Lymphatic vessels
  • 1.3.2 Lymph nodes
  • 1.4 Previous lymphatic anatomy studies
  • 1.5 Lymphatic mapping and sentinel node biopsy
  • 1.6 Sydney Melanoma Unit’s lymphoscintigraphy database
  • 1.6.1 Recording the primary melanoma site
  • 1.6.2 Recording the sentinel nodes
  • 1.6.3 Additional data recorded
  • 1.6.4 Data visualisation
  • 1.7 Anatomical modelling
  • 1.7.1 Finite element models
  • 1.7.2 Software
  • 1.8 Thesis overview
• 2 Anatomical Geometry
  • 2.1 Finite element basis functions
  • 2.2 Visible human data
  • 2.3 Creating a finite element model
  • 2.3.1 Data digitisation
  • 2.3.2 Initial linear mesh
  • 2.3.3 Fitting
  • 2.4 Customising finite element models
  • 2.5 Skin model construction
  • 2.5.1 Torso and limbs
  • 2.5.2 Head and neck
  • 2.6 Lymph node model construction
  • 2.7 Summary
• 3 Mapping Lymphoscintigraphy Data onto Anatomical Geometry
  • 3.1 Mapping the primary melanoma site
  • 3.1.1 Torso and lower limbs
  • 3.1.2 Upper limbs
  • 3.1.3 Hands and feet
  • 3.1.4 Head and neck
  • 3.2 Mapping the sentinel nodes
  • 3.3 Discussion
  • 3.4 Summary
• 4 Visualisation Methodology
  • 4.1 Introduction
  • 4.2 Random melanoma coordinates
  • 4.3 Spatial heat maps
  • 4.3.1 Field fitting
  • 4.3.2 Methodology
  • 4.4 Interactive skin selection tool
  • 4.4.1 Methodology
• 5 Visualisation of Mapped Lymphoscintigraphy Data
  • 5.1 Spatial heat maps
  • 5.1.1 Torso and limb node fields
  • 5.1.2 Head and neck node fields
  • 5.1.3 Number of draining node fields
  • 5.2 Interactive skin selection tool
  • 5.2.1 Results
  • 5.3 Discussion
  • 5.4 Summary
• 6 Statistical Analysis Methodology
  • 6.1 Introduction
  • 6.2 Analysing Sappey’s Lines
  • 6.2.1 Methodology
  • 6.3 Symmetry Testing
  • 6.3.1 Multinomial modelling approach
  • 6.3.2 Reflection method
  • 6.4 Cluster Analysis
  • 6.4.1 Theory
  • 6.5 Confidence intervals
  • 6.6 Summary
• 7 Statistical Analysis of Mapped Lymphoscintigraphy Data
  • 7.1 Analysing Sappey’s Lines
  • 7.1.1 Results
  • 7.2 Symmetry Testing
  • 7.2.1 Torso
  • 7.2.2 Upper and lower limbs
  • 7.2.3 Head and neck
  • 7.3 Cluster Analysis
  • 7.3.1 Results
  • 7.4 Discussion
  • 7.5 Summary
• 8 Conclusions and Future Work

GET THE COMPLETE PROJECT
3D Visualisation and Analysis of Skin Lymphatic Drainage Patterns in Melanoma