University of Southern California

Daniel Lee Ruderman

David B Agus

Rishi Raghav Rawat

Paul Thomas Macklin

Fei Sha

Darryl Shibata

Histologic classification of pathology specimens through machine learning is a nascent field which offers tremendous potential to improve cancer medicine. Its utility has been limited, in part because of differences in tissue preparation and the relative paucity of well-annotated images. We introduce tissue recognition, an unsupervised learning problem analogous to human face recognition, in which the goal is to identify individual tumors using a learned set of histologic features. This feature set is the 'tissue fingerprint.' Because only specimen identities are matched to fingerprints, constructing an algorithm for producing them is a self-learning task that does not need image metadata annotations. Here, we provide an algorithm for self-learning tissue fingerprints, that, in conjunction with color normalization, can match hematoxylin and eosin stained tissues to one of 104 patients with 93% accuracy. We applied this identification network's internal representation as a tissue fingerprint for use in predicting the molecular status of an individual tumor (breast cancer clinical estrogen receptor (ER) status). We describe a fingerprint-based classifier that predicts ER status from whole-slides with high accuracy (AUROC=0.90), and is an improvement over traditional transfer learning approaches. The use of tissue fingerprinting for digital pathology as a concise but meaningful histopathologic image representation enables a new range of machine learning algorithms leading to increased information for clinical decision making in patient management.

Histopathology classification through machine self-learning of 'tissue fingerprints'

A method assessing tissue morphology using machine learning includes a step of training a machine learnable device to predict the status of a diagnostic feature in stained tissue samples. The machine learnable device is trained with a characterized set of digital images of stained tissue samples. Each digital image of the characterized set has a known status for the diagnostic feature and an extracted feature map provides values for a extracted feature over an associated 2-dimensional grid of spatial locations. A step of inputting the set of extracted feature maps is inputted into the machine learnable device to form associations therein between the set of extracted feature maps to and the known status for the diagnostic feature to form a trained machine learnable device. The status for the diagnostic feature of a stained tissue sample of unknown status for the diagnostic feature is predicted from the trained machine learnable device.

Growing community of inventors

Los Angeles, CA, United States of America

Daniel Lee Ruderman