Differentiation of the Follicular Neoplasm on the Gray-Scale US by Image Selection Subsampling along with the Marginal Outline Using Convolutional Neural Network

Jeong Kweon Seo, Young Jae Kim, Kwang Gi Kim, Ilah Shin, Jung Hee Shin, Jin Young Kwak

Research output: Contribution to journalArticle

Abstract

We conducted differentiations between thyroid follicular adenoma and carcinoma for 8-bit bitmap ultrasonography (US) images utilizing a deep-learning approach. For the data sets, we gathered small-boxed selected images adjacent to the marginal outline of nodules and applied a convolutional neural network (CNN) to have differentiation, based on a statistical aggregation, that is, a decision by majority. From the implementation of the method, introducing a newly devised, scalable, parameterized normalization treatment, we observed meaningful aspects in various experiments, collecting evidence regarding the existence of features retained on the margin of thyroid nodules, such as 89.51% of the overall differentiation accuracy for the test data, with 93.19% of accuracy for benign adenoma and 71.05% for carcinoma, from 230 benign adenoma and 77 carcinoma US images, where we used only 39 benign adenomas and 39 carcinomas to train the CNN model, and, with these extremely small training data sets and their model, we tested 191 benign adenomas and 38 carcinomas. We present numerical results including area under receiver operating characteristic (AUROC).

Original languageEnglish
Article number3098293
JournalBioMed Research International
Volume2017
DOIs
Publication statusPublished - 2017 Jan 1

Fingerprint

Ultrasonography
Adenoma
Neural networks
Carcinoma
Neoplasms
Thyroid Neoplasms
Agglomeration
Thyroid Nodule
Neural Networks (Computer)
ROC Curve
Experiments
Learning
Deep learning

All Science Journal Classification (ASJC) codes

  • Biochemistry, Genetics and Molecular Biology(all)
  • Immunology and Microbiology(all)

Cite this

@article{5c6e89e6052d411780d6a5b996cf5bd1,
title = "Differentiation of the Follicular Neoplasm on the Gray-Scale US by Image Selection Subsampling along with the Marginal Outline Using Convolutional Neural Network",
abstract = "We conducted differentiations between thyroid follicular adenoma and carcinoma for 8-bit bitmap ultrasonography (US) images utilizing a deep-learning approach. For the data sets, we gathered small-boxed selected images adjacent to the marginal outline of nodules and applied a convolutional neural network (CNN) to have differentiation, based on a statistical aggregation, that is, a decision by majority. From the implementation of the method, introducing a newly devised, scalable, parameterized normalization treatment, we observed meaningful aspects in various experiments, collecting evidence regarding the existence of features retained on the margin of thyroid nodules, such as 89.51{\%} of the overall differentiation accuracy for the test data, with 93.19{\%} of accuracy for benign adenoma and 71.05{\%} for carcinoma, from 230 benign adenoma and 77 carcinoma US images, where we used only 39 benign adenomas and 39 carcinomas to train the CNN model, and, with these extremely small training data sets and their model, we tested 191 benign adenomas and 38 carcinomas. We present numerical results including area under receiver operating characteristic (AUROC).",
author = "Seo, {Jeong Kweon} and Kim, {Young Jae} and Kim, {Kwang Gi} and Ilah Shin and Shin, {Jung Hee} and Kwak, {Jin Young}",
year = "2017",
month = "1",
day = "1",
doi = "10.1155/2017/3098293",
language = "English",
volume = "2017",
journal = "BioMed Research International",
issn = "2314-6133",
publisher = "Hindawi Publishing Corporation",

}

Differentiation of the Follicular Neoplasm on the Gray-Scale US by Image Selection Subsampling along with the Marginal Outline Using Convolutional Neural Network. / Seo, Jeong Kweon; Kim, Young Jae; Kim, Kwang Gi; Shin, Ilah; Shin, Jung Hee; Kwak, Jin Young.

In: BioMed Research International, Vol. 2017, 3098293, 01.01.2017.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Differentiation of the Follicular Neoplasm on the Gray-Scale US by Image Selection Subsampling along with the Marginal Outline Using Convolutional Neural Network

AU - Seo, Jeong Kweon

AU - Kim, Young Jae

AU - Kim, Kwang Gi

AU - Shin, Ilah

AU - Shin, Jung Hee

AU - Kwak, Jin Young

PY - 2017/1/1

Y1 - 2017/1/1

N2 - We conducted differentiations between thyroid follicular adenoma and carcinoma for 8-bit bitmap ultrasonography (US) images utilizing a deep-learning approach. For the data sets, we gathered small-boxed selected images adjacent to the marginal outline of nodules and applied a convolutional neural network (CNN) to have differentiation, based on a statistical aggregation, that is, a decision by majority. From the implementation of the method, introducing a newly devised, scalable, parameterized normalization treatment, we observed meaningful aspects in various experiments, collecting evidence regarding the existence of features retained on the margin of thyroid nodules, such as 89.51% of the overall differentiation accuracy for the test data, with 93.19% of accuracy for benign adenoma and 71.05% for carcinoma, from 230 benign adenoma and 77 carcinoma US images, where we used only 39 benign adenomas and 39 carcinomas to train the CNN model, and, with these extremely small training data sets and their model, we tested 191 benign adenomas and 38 carcinomas. We present numerical results including area under receiver operating characteristic (AUROC).

AB - We conducted differentiations between thyroid follicular adenoma and carcinoma for 8-bit bitmap ultrasonography (US) images utilizing a deep-learning approach. For the data sets, we gathered small-boxed selected images adjacent to the marginal outline of nodules and applied a convolutional neural network (CNN) to have differentiation, based on a statistical aggregation, that is, a decision by majority. From the implementation of the method, introducing a newly devised, scalable, parameterized normalization treatment, we observed meaningful aspects in various experiments, collecting evidence regarding the existence of features retained on the margin of thyroid nodules, such as 89.51% of the overall differentiation accuracy for the test data, with 93.19% of accuracy for benign adenoma and 71.05% for carcinoma, from 230 benign adenoma and 77 carcinoma US images, where we used only 39 benign adenomas and 39 carcinomas to train the CNN model, and, with these extremely small training data sets and their model, we tested 191 benign adenomas and 38 carcinomas. We present numerical results including area under receiver operating characteristic (AUROC).

UR - http://www.scopus.com/inward/record.url?scp=85042155230&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85042155230&partnerID=8YFLogxK

U2 - 10.1155/2017/3098293

DO - 10.1155/2017/3098293

M3 - Article

C2 - 29527533

AN - SCOPUS:85042155230

VL - 2017

JO - BioMed Research International

JF - BioMed Research International

SN - 2314-6133

M1 - 3098293

ER -