Presented by:
Tsai, Mu-Hung, M.D.
Department of Radiation Oncology, National Cheng Kung University Hospital
2016/09/02
Lung Cancer
 Most prevalent cancer worldwide
 Histopathological classification
 Small cell carcinoma
 Non-small cell carcinoma
 Squamous cell carcinoma
 Adenocarcinoma
 Grade (Gr.1 – Gr.3)
 Low level of inter-observer agreement
Pathology
Squamous cell carcinoma Adenocarcinoma
Aim
 Fully automated microscopic pathology image features
 Predict lung cancer survival
Training
Validation
Methods & Materials
 The Cancer Genome
Atlas (TCGA)
 Stanford Tissue
Microarray (TMA)
Database
Cross-validate
Image: euthman @ Flickr
Results
Identifying Tumor
Adenocarcinoma vs. Normal Lung
SCC vs. Normal Lung
Adenocarcinoma vs. SCC (TCGA)
Adenocarcinoma vs. SCC (TMA)
Results
Predicting Survival
Predicting Adenocarcinoma Prognosis
Stage IB Grade 3 adenocarcinoma
Predicting SCC Prognosis
Discussion
 Automated workflow to analyze whole slide pathology
 Elastic net-Cox proportional hazards models
 Computationally efficient
 Handles right-censored survival data
 Performance not very sensitive to ML method
 Obtain large number of features!

Predicting NSCLC prognosis by automated pathology