Online Program Home
My Program

Abstract Details

Activity Number: 59 - Deep Learning in Statistics: Really?!
Type: Topic Contributed
Date/Time: Sunday, July 28, 2019 : 4:00 PM to 5:50 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #304432
Title: Deep Learning in Pathological Image Analysis
Author(s): Guanghua Xiao* and Shidan Wang
Companies: UT Southwestern Medical Center and UT Southwestern Medical Center
Keywords: convolutional neural network; computer vision; artificial intellengece; tissue slides; pathology images; nuclei segmentation

Tumor microenvironment dissection from pathological images is informative in lung adenocarcinoma (ADC) diagnosis. However, manually recognize and localize each cell in gigapixel pathological slides are almost impossible. Here we describe our development of the first automated cell type classification pipeline involving nuclei segmentation, convolutional neural network based tumor/stroma/lymphocytes classification, and feature extraction for lung ADC pathological images. The overall classification accuracy is 92.9% and 90.1% in training and independent testing datasets, separately. The extracted 6 simple features describe spatial distributions of different cell types and are prognostic for lung ADC. An image feature based prognosis model is developed and independently validated. The predicted risk group serves as an independent prognostic factor, after adjusting for clinical variables including age, gender, smoking status, and stage.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program