11 Electron Tomography
279
Fig. 11.10 a EEL spectral phantom used on the simulation of clustering techniques. b Results of
the application of HAC on the raw data, showing the segmentation under three different threshold
counts: 5000, 2500, and 1500. c Results of applying PCA to the shell region after HAC (green in
b). d Result of applying HAC to the PCA scores of the phantom
n = 128 · 128 = 16384 elements with only p = 3 possible spectral components
(instead of the 1024 initial channels).
11.3.2.2 Example. Magnetic Fe 3 O 4 –Mn 3 O 4 /MnO Nanoparticle
The experimental dataset analyzed corresponds to a core-shell structure for a
magnetic nanoparticle composed by a mixture of Fe 3 O 4 and Mn 3 O 4 /MnO [51].
Once again, raw data clustering segmentation successfully separated regions of
different composition. The core composition was identified as a majority of iron
oxide with minor contributions of manganese oxide (presumably arising from the
279
Fig. 11.10 a EEL spectral phantom used on the simulation of clustering techniques. b Results of
the application of HAC on the raw data, showing the segmentation under three different threshold
counts: 5000, 2500, and 1500. c Results of applying PCA to the shell region after HAC (green in
b). d Result of applying HAC to the PCA scores of the phantom
n = 128 · 128 = 16384 elements with only p = 3 possible spectral components
(instead of the 1024 initial channels).
11.3.2.2 Example. Magnetic Fe 3 O 4 –Mn 3 O 4 /MnO Nanoparticle
The experimental dataset analyzed corresponds to a core-shell structure for a
magnetic nanoparticle composed by a mixture of Fe 3 O 4 and Mn 3 O 4 /MnO [51].
Once again, raw data clustering segmentation successfully separated regions of
different composition. The core composition was identified as a majority of iron
oxide with minor contributions of manganese oxide (presumably arising from the