The Successive Projection Algorithm (SPA), an Algorithm with a Spatial Constraint for the Automatic Search of Endmembers in Hyperspectral Data.

Spectral mixing is a problem inherent to remote sensing data and results in fewimage pixel spectra representing "pure" targets. Linear spectral mixture analysis isdesigned to address this problem and it assumes that the pixel-to-pixel variability in ascene results from varying proportions of spectral endmembers. In this paper we present adifferent endmember-search algorithm called the Successive Projection Algorithm (SPA).SPA builds on convex geometry and orthogonal projection common to other endmembersearch algorithms by including a constraint on the spatial adjacency of endmembercandidate pixels. Consequently it can reduce the susceptibility to outlier pixels andgenerates realistic endmembers.This is demonstrated using two case studies (AVIRISCuprite cube and Probe-1 imagery for Baffin Island) where image endmembers can bevalidated with ground truth data. The SPA algorithm extracts endmembers fromhyperspectral data without having to reduce the data dimensionality. It uses the spectralangle (alike IEA) and the spatial adjacency of pixels in the image to constrain the selectionof candidate pixels representing an endmember. We designed SPA based on theobservation that many targets have spatial continuity (e.g. bedrock lithologies) in imageryand thus a spatial constraint would be beneficial in the endmember search. An additionalproduct of the SPA is data describing the change of the simplex volume ratio between successive iterations during the endmember extraction. It illustrates the influence of a newendmember on the data structure, and provides information on the convergence of thealgorithm. It can provide a general guideline to constrain the total number of endmembersin a search.

Radionuclide brain perfusion studies in suspected brain death.

The radionuclide brain perfusion study (RPS) has been suggested as a method of confirming suspected brain death. The hospital records and RPSs of 34 patients referred were reviewed because brain death was suspected. In every case but one the RPS showed absent or drastically reduced cerebral blood flow. No patient survived more than five days, and 25 survived less than 24 hours. These findings are consistent with the results of previous reports of a total of 248 patients; only one of 248 survived and was discharged from the hospital after having had a positive RPS. The RPS is highly accurate in confirming brain death.