Fig. 1
The identification and analysis of mitochondrial state in glioma under electron microscope by Machine Learning. a Analysis flow chart of a total of 7141 mitochondria of 54 glioma patients. b Mitochondrial state was valued in four aspects: a, Mito-density; b - d, Mito-size; e, Mito-shape; f, MRC state. c Mask R-CNN model for mitochondrial identification. d Electron microscopic representation of mitochondria in different grade glioma. Scale bars: 0.5 µm. e Statistical analysis of mitochondrial size, shape and density. Mitochondrial number were: Grade 1, n = 505; Grade 2, n = 1048; Grade 3, n = 2128; Grade 4, n = 3460. f Comparison of mitochondrial state between LGG and HGG by Radar map. g Electron microscopic representation of 3 types MRC state in different grade glioma. Green marked Type 1 healthy MRC, blue marked Type 2 unhealthy MRC, and red marked Type 3 damaged MRC. h Statistical analysis of 3 types MRC in different grade glioma. Mitochondrial number were: Grade 1, n = 72; Grade 2, n = 82; Grade 3, n = 128; Grade 4, n = 270. The data were presented as means ± SD. P were calculated by Kruskal-Wallis test (e) and ordinary one-way ANOVA (h)