Mapping melanoma lymphoscintigraphy data onto a 3D anatomically based model

Abstract

This study describes three-dimensional (3D) visualization of two-dimensional (2D) melanoma lymphatic mapping data, to provide a framework for analysis of melanoma spread patterns and a platform for recording new lymphoscintigraphy (LS) data more accurately in 3D. Specifically, the Sydney Melanoma Unit's LS database of over 5000 patients' primary cutaneous melanoma sites and sentinel lymph nodes have been mapped from 2D images onto a 3D anatomically based model. Anatomically accurate model geometries were created using the Visible Human dataset, giving a bicubic finite element skin mesh and discrete sentinel lymph node model. The full dataset of 2D melanoma site coordinates, excluding the head and neck, has been transformed onto this 3D skin mesh via free-form deformation and projection techniques. Sentinel lymph nodes were mapped onto the generic lymph node model for each patient. Preliminary spatial analysis indicates that a patient with a primary melanoma on the torso around the waist (on the standardized 3D model this region is 180 mm above and 130 mm below the umbilicus) with lymphatic drainage to the left axilla or left groin, will have a 17.7% probability of dual drainage to both node fields, with 95% confidence limits between 14.5 and 21.0%.