NSF EAGER: DEEP LEARNING-BASED VIRTUAL HISTOLOGY STAINING OF TISSUE SAMPLES

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

Microscopic imaging of tissue samples is a fundamental tool used for the diagnosis of various diseases and forms the workhorse of pathology and biological sciences. The clinically-established gold standard image of a tissue section is the result of a laborious process. This work will demonstrate the ability to virtually stain label-free tissue sections and will revolutionize the current paradigm for histological analysis. To demonstrate deep learning-based virtual histology staining of label-free human tissue samples this proposal will use salivary gland, thyroid, kidney, liver and lung samples, and will use three commonly used stains: H&E (salivary gland and thyroid), Jones stain (kidney) and Masson's Trichrome (liver and lung). This proposal will determine the staining efficacy of the proposed approach for whole slide images and will blindly evaluate the virtually stained outputs with gold standard stained samples. The output of this proposed system will be validated by a group of pathologists who will compare histopathological features with the virtual staining technique against conventional histology techniques. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.