PROCEDURE FOR PROCESSING HISTOPATHOLOGICAL IMAGES TO CREATE SINGLE-CELL IMAGE DATA WITH MULTI-FEATURE LABELS FOR TRAINING DEEP LEARNING MODELS

Abstract

The invention relates to the field of information technology and biomedical engineering, specifically concerning the processing of histopathological images and the construction of a dataset of single-cell images with multi-feature labeling. This solution focuses on multi-scale spatial synchronization, pixel-level morphological signal optimization, and color invariance preservation, for the purpose of training, evaluating, and deploying deep learning models in supporting biomedical diagnosis and research.