Segmentation and Classification for Hematoxylin Eosin Images
SciPathJ (Segmentation and Classification of Images - Pipeline for the Analysis of Tissue Histopathology) is a Java software that automates the analysis of H&E-stained microscopy images. It processes individual images or folders full of images by segmenting cells, nuclei, cytoplasm, and vessels. It then saves comprehensive measurements including shape, size, and stain color information to CSV format (more than 150+ features). Users can define custom cell classes and train XGBoost models to classify all cells in the segmented dataset. The program is fully functional by itself, but it is also compatible with ImageJ/Fiji. Detailed information about configuration options and performance metrics is available in the documentation.
My name is Sebastian Micu, a Medical Resident in Clinical Genetics at Sapienza University of Rome with a passion for Programming and Machine Learning. I developed SciPathJ as a follow up of my thesis research, with support from the Centro de Biología Molecular Severo Ochoa in Madrid. I presented this software at the 2025 Maker Faire in Rome. This project combines my interests in medical imaging and computational analysis to create accessible tools for histology research. SciPathj represents my contribution to open-source scientific software, designed to help researchers analyze tissue samples more efficiently.