Fig. 6: Analysis of the H&E Cryo dataset based on the U-Netsynth predictions.
From: Deep learning-based image analysis in muscle histopathology using photo-realistic synthetic data
a Extracted fiber features from the H&E Cryo dataset. b Scatter plot of the fiber diameter distribution showing the mean fiber diameter (in µm) in relation to the fiber diameter standard deviation (std-dev.) (in µm). Each scatter point represents either a WT (healthy, wild-type sibling, in green) or a DKO Hom (abnormal/myopathic, homozygous desmin knock-out, in orange) section. Fiber diameters were analyzed considering either the whole section (circles), the M. soleus only (triangles) or the M. gastrocnemius only (crosses). c Scatter plots of the connective tissue local thickness distribution for each tissue section considering solely the M. soleus (right) or the M. gastrocnemius (left) parts. d Abnormal/myopathic vs. normal muscle fiber distribution and connective tissue based on the abnormality analysis.
