The results obtained during the project deal with the identification of pixel and image features capable of providing information regarding the collagen structure within capsules surrounding thyroid nodules.
- Pixel-level information provided by fitting experimental PSHG data with theoretical collagen models (single axis molecule model) were extracted in the form of second order susceptibility tensor elements ratios. We assessed the influence of hematoxylin and eosin staining on the fitting procedure.
- Different immunohistochemistry markers were tested with regard to their influence on MPM (TPEF + SHG) imaging. A new use of MPM imaging on IHC-stained tissues sections was proposed: creating image collections acquired on such samples which can facilitate the interpretation of MPM images for untrained pathologists in novel imaging approaches.
- FIJI macros for automatization of pixel-level and image feature extraction on SHG and bright-field microscopy images.
- Classification algorithm with random forest for the classification of thyroid nodule capsule images.
- An extended image set containing thyroid nodule capsule SHG images with additionally generated images starting from theoretical collagen models.