Prediction of Standard-Dose PET Image by Low-Dose PET and MRI Images

Jiayin Kang, Yaozong Gao, Yao Wu, Guangkai Ma, Feng Shi, Weili Lin, Dinggang Shen

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Positron emission tomography (PET) is a nuclear medical imaging technology that produces 3D images of tissue metabolic activity in human body. PET has been used in various clinical applications, such as diagnosis of tumors and diffuse brain disorders. High quality PET image plays an essential role in diagnosing diseases/disorders and assessing the response to therapy. In practice, in order to obtain the high quality PET images, standard-dose radionuclide (tracer) needs to be used and injected into the living body. As a result, it will inevitably increase the risk of radiation. In this paper, we propose a regression forest (RF) based framework for predicting standard-dose PET images using low-dose PET and corresponding magnetic resonance imaging (MRI) images instead of injecting the standard-dose radionuclide into the body. The proposed approach has been evaluated on a dataset consisting of 7 subjects using leave-one-out cross-validation. Moreover, we compare the prediction performance between sparse representation (SR) based method and our proposed method. Both qualitative and quantitative results illustrate the practicability of our proposed method.

Original languageEnglish
Pages (from-to)280-288
Number of pages9
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8679
Publication statusPublished - 2014 Jan 1
Externally publishedYes

Fingerprint

Positron Emission Tomography
Positron emission tomography
Magnetic Resonance Imaging
Magnetic resonance
Dose
Imaging techniques
Prediction
Radioisotopes
Disorder
Radioactive tracers
Sparse Representation
Medical Imaging
Medical imaging
Performance Prediction
3D Image
Cross-validation
Therapy
Standards
Tumors
Tumor

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Prediction of Standard-Dose PET Image by Low-Dose PET and MRI Images. / Kang, Jiayin; Gao, Yaozong; Wu, Yao; Ma, Guangkai; Shi, Feng; Lin, Weili; Shen, Dinggang.

In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 8679, 01.01.2014, p. 280-288.

Research output: Contribution to journalArticle

@article{a81a529e370e4e7c95d0ce738d0c3a1b,
title = "Prediction of Standard-Dose PET Image by Low-Dose PET and MRI Images",
abstract = "Positron emission tomography (PET) is a nuclear medical imaging technology that produces 3D images of tissue metabolic activity in human body. PET has been used in various clinical applications, such as diagnosis of tumors and diffuse brain disorders. High quality PET image plays an essential role in diagnosing diseases/disorders and assessing the response to therapy. In practice, in order to obtain the high quality PET images, standard-dose radionuclide (tracer) needs to be used and injected into the living body. As a result, it will inevitably increase the risk of radiation. In this paper, we propose a regression forest (RF) based framework for predicting standard-dose PET images using low-dose PET and corresponding magnetic resonance imaging (MRI) images instead of injecting the standard-dose radionuclide into the body. The proposed approach has been evaluated on a dataset consisting of 7 subjects using leave-one-out cross-validation. Moreover, we compare the prediction performance between sparse representation (SR) based method and our proposed method. Both qualitative and quantitative results illustrate the practicability of our proposed method.",
author = "Jiayin Kang and Yaozong Gao and Yao Wu and Guangkai Ma and Feng Shi and Weili Lin and Dinggang Shen",
year = "2014",
month = "1",
day = "1",
language = "English",
volume = "8679",
pages = "280--288",
journal = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
issn = "0302-9743",
publisher = "Springer Verlag",

}

TY - JOUR

T1 - Prediction of Standard-Dose PET Image by Low-Dose PET and MRI Images

AU - Kang, Jiayin

AU - Gao, Yaozong

AU - Wu, Yao

AU - Ma, Guangkai

AU - Shi, Feng

AU - Lin, Weili

AU - Shen, Dinggang

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Positron emission tomography (PET) is a nuclear medical imaging technology that produces 3D images of tissue metabolic activity in human body. PET has been used in various clinical applications, such as diagnosis of tumors and diffuse brain disorders. High quality PET image plays an essential role in diagnosing diseases/disorders and assessing the response to therapy. In practice, in order to obtain the high quality PET images, standard-dose radionuclide (tracer) needs to be used and injected into the living body. As a result, it will inevitably increase the risk of radiation. In this paper, we propose a regression forest (RF) based framework for predicting standard-dose PET images using low-dose PET and corresponding magnetic resonance imaging (MRI) images instead of injecting the standard-dose radionuclide into the body. The proposed approach has been evaluated on a dataset consisting of 7 subjects using leave-one-out cross-validation. Moreover, we compare the prediction performance between sparse representation (SR) based method and our proposed method. Both qualitative and quantitative results illustrate the practicability of our proposed method.

AB - Positron emission tomography (PET) is a nuclear medical imaging technology that produces 3D images of tissue metabolic activity in human body. PET has been used in various clinical applications, such as diagnosis of tumors and diffuse brain disorders. High quality PET image plays an essential role in diagnosing diseases/disorders and assessing the response to therapy. In practice, in order to obtain the high quality PET images, standard-dose radionuclide (tracer) needs to be used and injected into the living body. As a result, it will inevitably increase the risk of radiation. In this paper, we propose a regression forest (RF) based framework for predicting standard-dose PET images using low-dose PET and corresponding magnetic resonance imaging (MRI) images instead of injecting the standard-dose radionuclide into the body. The proposed approach has been evaluated on a dataset consisting of 7 subjects using leave-one-out cross-validation. Moreover, we compare the prediction performance between sparse representation (SR) based method and our proposed method. Both qualitative and quantitative results illustrate the practicability of our proposed method.

UR - http://www.scopus.com/inward/record.url?scp=84921747098&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84921747098&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:84921747098

VL - 8679

SP - 280

EP - 288

JO - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

JF - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SN - 0302-9743

ER -