Principal component analysis graphs of the comparisons of different control and pathological samples. PCA graphs for the comparisons of muscle images from different sources using different groups of characteristics. The green dots represent the control images (quadricep biopsies from children in a, c, and d and adult biceps in b). The black star represents the centroid for the control dataset in each graph. The red and blue dots are the MD and NA images, respectively. (a) Control versus MD comparison using characteristics 15, 18 and 19. (b) Control versus NA comparison using characteristics 12, 20, 21 and 22. (c) Same comparison as in (a), showing the degree of pathology as evaluated by the pathologist and the correlation coefficient with distances to the centroid. (d) Control versus MD comparison using characteristics 13, 15, 18, 27 and 41 and showing the degree of pathology and the correlation with distances to the centroid. Three images (dots) are highlighted with an orange, light blue or violet circle. These dots correspond to the images in e, f and g respectively. (e) Detail of a representative control image (the one closer to the centroid, QC60-1). (f) Detail of the QD54-2 image. This image shows a small increase in the amount of collagen between the fibers and in the heterogeneity of sizes and shapes of them. (g) Detail of the QD58-2 image. This image presents a clear increase of the endomysium and higher heterogeneity in sizes and shapes than in e and f. PCA, principal component analysis.
Sáez et al. BMC Medicine 2013 11:77 doi:10.1186/1741-7015-11-77