Lung and Pancreatic Tumor Characterization in the Deep Learning Era: Novel Supervised and Unsupervised Learning Approaches

Sarfaraz Hussein, Pujan Kandel, Candice W. Bolan, Michael B Wallace, Ulas Bagci

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Risk stratification (characterization) of tumors from radiology images can be more accurate and faster with computer-aided diagnosis (CAD) tools. Tumor characterization through such tools can also enable non-invasive cancer staging, prognosis, and foster personalized treatment planning as a part of precision medicine. In this papet, we propose both supervised and unsupervised machine learning strategies to improve tumor characterization. Our first approach is based on supervised learning for which we demonstrate significant gains with deep learning algorithms, particularly by utilizing a 3D convolutional neural network and transfer learning. Motivated by the radiologists' interpretations of the scans, we then show how to incorporate task-dependent feature representations into a CAD system via a graph-regularized sparse multi-task learning framework. In the second approach, we explore an unsupervised learning algorithm to address the limited availability of labeled training data, a common problem in medical imaging applications. Inspired by learning from label proportion approaches in computer vision, we propose to use proportion-support vector machine for characterizing tumors. We also seek the answer to the fundamental question about the goodness of 'deep features' for unsupervised tumor classification. We evaluate our proposed supervised and unsupervised learning algorithms on two different tumor diagnosis challenges: lung and pancreas with 1018 CT and 171 MRI scans, respectively, and obtain the state-of-the-art sensitivity and specificity results in both problems.

Original languageEnglish (US)
Article number8624570
Pages (from-to)1777-1787
Number of pages11
JournalIEEE transactions on medical imaging
Volume38
Issue number8
DOIs
StatePublished - Aug 1 2019

Fingerprint

Unsupervised learning
Supervised learning
Tumors
Learning
Lung
Learning algorithms
Computer aided diagnosis
Neoplasms
Precision Medicine
Radiology
Neoplasm Staging
Medical imaging
Diagnostic Imaging
Computer vision
Medicine
Support vector machines
Learning systems
Deep learning
Labels
Pancreas

Keywords

  • 3D CNN
  • IPMN
  • lung cancer
  • pancreatic cancer
  • Unsupervised learning

ASJC Scopus subject areas

  • Software
  • Radiological and Ultrasound Technology
  • Computer Science Applications
  • Electrical and Electronic Engineering

Cite this

Lung and Pancreatic Tumor Characterization in the Deep Learning Era : Novel Supervised and Unsupervised Learning Approaches. / Hussein, Sarfaraz; Kandel, Pujan; Bolan, Candice W.; Wallace, Michael B; Bagci, Ulas.

In: IEEE transactions on medical imaging, Vol. 38, No. 8, 8624570, 01.08.2019, p. 1777-1787.

Research output: Contribution to journalArticle

Hussein, Sarfaraz ; Kandel, Pujan ; Bolan, Candice W. ; Wallace, Michael B ; Bagci, Ulas. / Lung and Pancreatic Tumor Characterization in the Deep Learning Era : Novel Supervised and Unsupervised Learning Approaches. In: IEEE transactions on medical imaging. 2019 ; Vol. 38, No. 8. pp. 1777-1787.
@article{e47b617b38b54f3ca7299750ddbcbb7d,
title = "Lung and Pancreatic Tumor Characterization in the Deep Learning Era: Novel Supervised and Unsupervised Learning Approaches",
abstract = "Risk stratification (characterization) of tumors from radiology images can be more accurate and faster with computer-aided diagnosis (CAD) tools. Tumor characterization through such tools can also enable non-invasive cancer staging, prognosis, and foster personalized treatment planning as a part of precision medicine. In this papet, we propose both supervised and unsupervised machine learning strategies to improve tumor characterization. Our first approach is based on supervised learning for which we demonstrate significant gains with deep learning algorithms, particularly by utilizing a 3D convolutional neural network and transfer learning. Motivated by the radiologists' interpretations of the scans, we then show how to incorporate task-dependent feature representations into a CAD system via a graph-regularized sparse multi-task learning framework. In the second approach, we explore an unsupervised learning algorithm to address the limited availability of labeled training data, a common problem in medical imaging applications. Inspired by learning from label proportion approaches in computer vision, we propose to use proportion-support vector machine for characterizing tumors. We also seek the answer to the fundamental question about the goodness of 'deep features' for unsupervised tumor classification. We evaluate our proposed supervised and unsupervised learning algorithms on two different tumor diagnosis challenges: lung and pancreas with 1018 CT and 171 MRI scans, respectively, and obtain the state-of-the-art sensitivity and specificity results in both problems.",
keywords = "3D CNN, IPMN, lung cancer, pancreatic cancer, Unsupervised learning",
author = "Sarfaraz Hussein and Pujan Kandel and Bolan, {Candice W.} and Wallace, {Michael B} and Ulas Bagci",
year = "2019",
month = "8",
day = "1",
doi = "10.1109/TMI.2019.2894349",
language = "English (US)",
volume = "38",
pages = "1777--1787",
journal = "IEEE Transactions on Medical Imaging",
issn = "0278-0062",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "8",

}

TY - JOUR

T1 - Lung and Pancreatic Tumor Characterization in the Deep Learning Era

T2 - Novel Supervised and Unsupervised Learning Approaches

AU - Hussein, Sarfaraz

AU - Kandel, Pujan

AU - Bolan, Candice W.

AU - Wallace, Michael B

AU - Bagci, Ulas

PY - 2019/8/1

Y1 - 2019/8/1

N2 - Risk stratification (characterization) of tumors from radiology images can be more accurate and faster with computer-aided diagnosis (CAD) tools. Tumor characterization through such tools can also enable non-invasive cancer staging, prognosis, and foster personalized treatment planning as a part of precision medicine. In this papet, we propose both supervised and unsupervised machine learning strategies to improve tumor characterization. Our first approach is based on supervised learning for which we demonstrate significant gains with deep learning algorithms, particularly by utilizing a 3D convolutional neural network and transfer learning. Motivated by the radiologists' interpretations of the scans, we then show how to incorporate task-dependent feature representations into a CAD system via a graph-regularized sparse multi-task learning framework. In the second approach, we explore an unsupervised learning algorithm to address the limited availability of labeled training data, a common problem in medical imaging applications. Inspired by learning from label proportion approaches in computer vision, we propose to use proportion-support vector machine for characterizing tumors. We also seek the answer to the fundamental question about the goodness of 'deep features' for unsupervised tumor classification. We evaluate our proposed supervised and unsupervised learning algorithms on two different tumor diagnosis challenges: lung and pancreas with 1018 CT and 171 MRI scans, respectively, and obtain the state-of-the-art sensitivity and specificity results in both problems.

AB - Risk stratification (characterization) of tumors from radiology images can be more accurate and faster with computer-aided diagnosis (CAD) tools. Tumor characterization through such tools can also enable non-invasive cancer staging, prognosis, and foster personalized treatment planning as a part of precision medicine. In this papet, we propose both supervised and unsupervised machine learning strategies to improve tumor characterization. Our first approach is based on supervised learning for which we demonstrate significant gains with deep learning algorithms, particularly by utilizing a 3D convolutional neural network and transfer learning. Motivated by the radiologists' interpretations of the scans, we then show how to incorporate task-dependent feature representations into a CAD system via a graph-regularized sparse multi-task learning framework. In the second approach, we explore an unsupervised learning algorithm to address the limited availability of labeled training data, a common problem in medical imaging applications. Inspired by learning from label proportion approaches in computer vision, we propose to use proportion-support vector machine for characterizing tumors. We also seek the answer to the fundamental question about the goodness of 'deep features' for unsupervised tumor classification. We evaluate our proposed supervised and unsupervised learning algorithms on two different tumor diagnosis challenges: lung and pancreas with 1018 CT and 171 MRI scans, respectively, and obtain the state-of-the-art sensitivity and specificity results in both problems.

KW - 3D CNN

KW - IPMN

KW - lung cancer

KW - pancreatic cancer

KW - Unsupervised learning

UR - http://www.scopus.com/inward/record.url?scp=85064203279&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064203279&partnerID=8YFLogxK

U2 - 10.1109/TMI.2019.2894349

DO - 10.1109/TMI.2019.2894349

M3 - Article

C2 - 30676950

AN - SCOPUS:85064203279

VL - 38

SP - 1777

EP - 1787

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

SN - 0278-0062

IS - 8

M1 - 8624570

ER -