Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images: A Japanese multicenter study

Kenichi Nakajima, Takashi Kudo, Tomoaki Nakata, Keisuke Kiso, Tokuo Kasai, Yasuyo Taniguchi, Shinro Matsuo, Mitsuru Momose, Masayasu Nakagawa, Masayoshi Sarai, Satoshi Hida, Hirokazu Tanaka, Kunihiko Yokoyama, Koichi Okuda, Lars Edenbrandt

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

Purpose Artificial neural networks (ANN) might help to diagnose coronary artery disease. This study aimed to determine whether the diagnostic accuracy of an ANN-based diagnostic system and conventional quantitation are comparable. Methods The ANN was trained to classify potentially abnormal areas as true or false based on the nuclear cardiology expert interpretation of 1001 gated stress/rest 99mTc-MIBI images at 12 hospitals. The diagnostic accuracy of the ANN was compared with 364 expert interpretations that served as the gold standard of abnormality for the validation study. Conventional summed stress/rest/difference scores (SSS/ SRS/SDS) were calculated and compared with receiver operating characteristics (ROC) analysis. Results The ANN generated a better area under the ROC curves (AUC) than SSS (0.92 vs. 0.82, p < 0.0001), indicating better identification of stress defects. The ANN also generated a better AUC than SDS (0.90 vs. 0.75, p < 0.0001) for stressinduced ischemia. The AUC for patients with old myocardial infarction based on rest defects was 0.97 (0.91 for SRS, p = 0.0061), and that for patients with and without a history of revascularization based on stress defects was 0.94 and 0.90 (p = 0.0055 and p < 0.0001 vs. SSS, respectively). The SSS/ SRS/SDS steeply increased when ANN values (probability of abnormality) were >0.80. Conclusion The ANN was diagnostically accurate in various clinical settings, including that of patients with previous myocardial infarction and coronary revascularization. The ANN could help to diagnose coronary artery disease.

Original languageEnglish
Pages (from-to)2280-2289
Number of pages10
JournalEuropean Journal of Nuclear Medicine and Molecular Imaging
Volume44
Issue number13
DOIs
Publication statusPublished - 01-01-2017

Fingerprint

ROC Curve
Multicenter Studies
Coronary Artery Disease
Perfusion
Validation Studies
Cardiology
Myocardial Infarction

All Science Journal Classification (ASJC) codes

  • Radiology Nuclear Medicine and imaging

Cite this

Nakajima, Kenichi ; Kudo, Takashi ; Nakata, Tomoaki ; Kiso, Keisuke ; Kasai, Tokuo ; Taniguchi, Yasuyo ; Matsuo, Shinro ; Momose, Mitsuru ; Nakagawa, Masayasu ; Sarai, Masayoshi ; Hida, Satoshi ; Tanaka, Hirokazu ; Yokoyama, Kunihiko ; Okuda, Koichi ; Edenbrandt, Lars. / Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images : A Japanese multicenter study. In: European Journal of Nuclear Medicine and Molecular Imaging. 2017 ; Vol. 44, No. 13. pp. 2280-2289.
@article{54b07f45a9ce4da79386da4d64f1033e,
title = "Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images: A Japanese multicenter study",
abstract = "Purpose Artificial neural networks (ANN) might help to diagnose coronary artery disease. This study aimed to determine whether the diagnostic accuracy of an ANN-based diagnostic system and conventional quantitation are comparable. Methods The ANN was trained to classify potentially abnormal areas as true or false based on the nuclear cardiology expert interpretation of 1001 gated stress/rest 99mTc-MIBI images at 12 hospitals. The diagnostic accuracy of the ANN was compared with 364 expert interpretations that served as the gold standard of abnormality for the validation study. Conventional summed stress/rest/difference scores (SSS/ SRS/SDS) were calculated and compared with receiver operating characteristics (ROC) analysis. Results The ANN generated a better area under the ROC curves (AUC) than SSS (0.92 vs. 0.82, p < 0.0001), indicating better identification of stress defects. The ANN also generated a better AUC than SDS (0.90 vs. 0.75, p < 0.0001) for stressinduced ischemia. The AUC for patients with old myocardial infarction based on rest defects was 0.97 (0.91 for SRS, p = 0.0061), and that for patients with and without a history of revascularization based on stress defects was 0.94 and 0.90 (p = 0.0055 and p < 0.0001 vs. SSS, respectively). The SSS/ SRS/SDS steeply increased when ANN values (probability of abnormality) were >0.80. Conclusion The ANN was diagnostically accurate in various clinical settings, including that of patients with previous myocardial infarction and coronary revascularization. The ANN could help to diagnose coronary artery disease.",
author = "Kenichi Nakajima and Takashi Kudo and Tomoaki Nakata and Keisuke Kiso and Tokuo Kasai and Yasuyo Taniguchi and Shinro Matsuo and Mitsuru Momose and Masayasu Nakagawa and Masayoshi Sarai and Satoshi Hida and Hirokazu Tanaka and Kunihiko Yokoyama and Koichi Okuda and Lars Edenbrandt",
year = "2017",
month = "1",
day = "1",
doi = "10.1007/s00259-017-3834-x",
language = "English",
volume = "44",
pages = "2280--2289",
journal = "European Journal of Nuclear Medicine and Molecular Imaging",
issn = "1619-7070",
publisher = "Springer Verlag",
number = "13",

}

Nakajima, K, Kudo, T, Nakata, T, Kiso, K, Kasai, T, Taniguchi, Y, Matsuo, S, Momose, M, Nakagawa, M, Sarai, M, Hida, S, Tanaka, H, Yokoyama, K, Okuda, K & Edenbrandt, L 2017, 'Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images: A Japanese multicenter study', European Journal of Nuclear Medicine and Molecular Imaging, vol. 44, no. 13, pp. 2280-2289. https://doi.org/10.1007/s00259-017-3834-x

Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images : A Japanese multicenter study. / Nakajima, Kenichi; Kudo, Takashi; Nakata, Tomoaki; Kiso, Keisuke; Kasai, Tokuo; Taniguchi, Yasuyo; Matsuo, Shinro; Momose, Mitsuru; Nakagawa, Masayasu; Sarai, Masayoshi; Hida, Satoshi; Tanaka, Hirokazu; Yokoyama, Kunihiko; Okuda, Koichi; Edenbrandt, Lars.

In: European Journal of Nuclear Medicine and Molecular Imaging, Vol. 44, No. 13, 01.01.2017, p. 2280-2289.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images

T2 - A Japanese multicenter study

AU - Nakajima, Kenichi

AU - Kudo, Takashi

AU - Nakata, Tomoaki

AU - Kiso, Keisuke

AU - Kasai, Tokuo

AU - Taniguchi, Yasuyo

AU - Matsuo, Shinro

AU - Momose, Mitsuru

AU - Nakagawa, Masayasu

AU - Sarai, Masayoshi

AU - Hida, Satoshi

AU - Tanaka, Hirokazu

AU - Yokoyama, Kunihiko

AU - Okuda, Koichi

AU - Edenbrandt, Lars

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Purpose Artificial neural networks (ANN) might help to diagnose coronary artery disease. This study aimed to determine whether the diagnostic accuracy of an ANN-based diagnostic system and conventional quantitation are comparable. Methods The ANN was trained to classify potentially abnormal areas as true or false based on the nuclear cardiology expert interpretation of 1001 gated stress/rest 99mTc-MIBI images at 12 hospitals. The diagnostic accuracy of the ANN was compared with 364 expert interpretations that served as the gold standard of abnormality for the validation study. Conventional summed stress/rest/difference scores (SSS/ SRS/SDS) were calculated and compared with receiver operating characteristics (ROC) analysis. Results The ANN generated a better area under the ROC curves (AUC) than SSS (0.92 vs. 0.82, p < 0.0001), indicating better identification of stress defects. The ANN also generated a better AUC than SDS (0.90 vs. 0.75, p < 0.0001) for stressinduced ischemia. The AUC for patients with old myocardial infarction based on rest defects was 0.97 (0.91 for SRS, p = 0.0061), and that for patients with and without a history of revascularization based on stress defects was 0.94 and 0.90 (p = 0.0055 and p < 0.0001 vs. SSS, respectively). The SSS/ SRS/SDS steeply increased when ANN values (probability of abnormality) were >0.80. Conclusion The ANN was diagnostically accurate in various clinical settings, including that of patients with previous myocardial infarction and coronary revascularization. The ANN could help to diagnose coronary artery disease.

AB - Purpose Artificial neural networks (ANN) might help to diagnose coronary artery disease. This study aimed to determine whether the diagnostic accuracy of an ANN-based diagnostic system and conventional quantitation are comparable. Methods The ANN was trained to classify potentially abnormal areas as true or false based on the nuclear cardiology expert interpretation of 1001 gated stress/rest 99mTc-MIBI images at 12 hospitals. The diagnostic accuracy of the ANN was compared with 364 expert interpretations that served as the gold standard of abnormality for the validation study. Conventional summed stress/rest/difference scores (SSS/ SRS/SDS) were calculated and compared with receiver operating characteristics (ROC) analysis. Results The ANN generated a better area under the ROC curves (AUC) than SSS (0.92 vs. 0.82, p < 0.0001), indicating better identification of stress defects. The ANN also generated a better AUC than SDS (0.90 vs. 0.75, p < 0.0001) for stressinduced ischemia. The AUC for patients with old myocardial infarction based on rest defects was 0.97 (0.91 for SRS, p = 0.0061), and that for patients with and without a history of revascularization based on stress defects was 0.94 and 0.90 (p = 0.0055 and p < 0.0001 vs. SSS, respectively). The SSS/ SRS/SDS steeply increased when ANN values (probability of abnormality) were >0.80. Conclusion The ANN was diagnostically accurate in various clinical settings, including that of patients with previous myocardial infarction and coronary revascularization. The ANN could help to diagnose coronary artery disease.

UR - http://www.scopus.com/inward/record.url?scp=85029787148&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029787148&partnerID=8YFLogxK

U2 - 10.1007/s00259-017-3834-x

DO - 10.1007/s00259-017-3834-x

M3 - Article

AN - SCOPUS:85029787148

VL - 44

SP - 2280

EP - 2289

JO - European Journal of Nuclear Medicine and Molecular Imaging

JF - European Journal of Nuclear Medicine and Molecular Imaging

SN - 1619-7070

IS - 13

ER -