A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations

A. Sreekumari, D. Shanbhag, D. Yeo, T. Foo, J. Pilitsis, J. Polzin, U. Patil, A. Coblentz, A. Kapadia, J. Khinda, A. Boutet, John D Port, I. Hancu

Research output: Contribution to journalArticle

Abstract

BACKGROUND AND PURPOSE: MR imaging rescans and recalls can create large hospital revenue loss. The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. MATERIALS AND METHODS: A deep learning- based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. These series were assumed to be scanned for 2 scan indications: screening for multiple sclerosis and stroke. RESULTS: The image-quality rating was found to be scan indication- And reading radiologist- dependent. Of the 49 test datasets, technologists created a mean ratio of rescans/recalls of (4.7±5.1)/(9.5±6.8) for MS and (8.6±7.7)/(1.6±1.9) for stroke. With thresholds adapted for scan indication and reading radiologist, deep learning created a rescan/recall ratio of (7.3±2.2)/(3.2±2.5) for MS, and (3.6± 1.5)/(2.8±1.6) for stroke. Due to the large variability in the technologists' assessments, it was only the decrease in the recall rate for MS, for which the deep learning algorithm was trained, that was statistically significant (P = .03). CONCLUSIONS: Fast, automated deep learning- based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

Original languageEnglish (US)
Pages (from-to)217-223
Number of pages7
JournalAmerican Journal of Neuroradiology
Volume40
Issue number2
DOIs
StatePublished - Feb 1 2019

Fingerprint

Learning
Reading
Stroke
Artifacts
Multiple Sclerosis
Radiologists
Brain

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Clinical Neurology

Cite this

Sreekumari, A., Shanbhag, D., Yeo, D., Foo, T., Pilitsis, J., Polzin, J., ... Hancu, I. (2019). A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations. American Journal of Neuroradiology, 40(2), 217-223. https://doi.org/10.3174/ajnr.A5926

A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations. / Sreekumari, A.; Shanbhag, D.; Yeo, D.; Foo, T.; Pilitsis, J.; Polzin, J.; Patil, U.; Coblentz, A.; Kapadia, A.; Khinda, J.; Boutet, A.; Port, John D; Hancu, I.

In: American Journal of Neuroradiology, Vol. 40, No. 2, 01.02.2019, p. 217-223.

Research output: Contribution to journalArticle

Sreekumari, A, Shanbhag, D, Yeo, D, Foo, T, Pilitsis, J, Polzin, J, Patil, U, Coblentz, A, Kapadia, A, Khinda, J, Boutet, A, Port, JD & Hancu, I 2019, 'A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations', American Journal of Neuroradiology, vol. 40, no. 2, pp. 217-223. https://doi.org/10.3174/ajnr.A5926
Sreekumari, A. ; Shanbhag, D. ; Yeo, D. ; Foo, T. ; Pilitsis, J. ; Polzin, J. ; Patil, U. ; Coblentz, A. ; Kapadia, A. ; Khinda, J. ; Boutet, A. ; Port, John D ; Hancu, I. / A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations. In: American Journal of Neuroradiology. 2019 ; Vol. 40, No. 2. pp. 217-223.
@article{2289df40047d4c828e35284fce33c394,
title = "A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations",
abstract = "BACKGROUND AND PURPOSE: MR imaging rescans and recalls can create large hospital revenue loss. The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. MATERIALS AND METHODS: A deep learning- based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. These series were assumed to be scanned for 2 scan indications: screening for multiple sclerosis and stroke. RESULTS: The image-quality rating was found to be scan indication- And reading radiologist- dependent. Of the 49 test datasets, technologists created a mean ratio of rescans/recalls of (4.7±5.1)/(9.5±6.8) for MS and (8.6±7.7)/(1.6±1.9) for stroke. With thresholds adapted for scan indication and reading radiologist, deep learning created a rescan/recall ratio of (7.3±2.2)/(3.2±2.5) for MS, and (3.6± 1.5)/(2.8±1.6) for stroke. Due to the large variability in the technologists' assessments, it was only the decrease in the recall rate for MS, for which the deep learning algorithm was trained, that was statistically significant (P = .03). CONCLUSIONS: Fast, automated deep learning- based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.",
author = "A. Sreekumari and D. Shanbhag and D. Yeo and T. Foo and J. Pilitsis and J. Polzin and U. Patil and A. Coblentz and A. Kapadia and J. Khinda and A. Boutet and Port, {John D} and I. Hancu",
year = "2019",
month = "2",
day = "1",
doi = "10.3174/ajnr.A5926",
language = "English (US)",
volume = "40",
pages = "217--223",
journal = "American Journal of Neuroradiology",
issn = "0195-6108",
publisher = "American Society of Neuroradiology",
number = "2",

}

TY - JOUR

T1 - A deep learning-based approach to reduce rescan and recall rates in clinical MRI examinations

AU - Sreekumari, A.

AU - Shanbhag, D.

AU - Yeo, D.

AU - Foo, T.

AU - Pilitsis, J.

AU - Polzin, J.

AU - Patil, U.

AU - Coblentz, A.

AU - Kapadia, A.

AU - Khinda, J.

AU - Boutet, A.

AU - Port, John D

AU - Hancu, I.

PY - 2019/2/1

Y1 - 2019/2/1

N2 - BACKGROUND AND PURPOSE: MR imaging rescans and recalls can create large hospital revenue loss. The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. MATERIALS AND METHODS: A deep learning- based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. These series were assumed to be scanned for 2 scan indications: screening for multiple sclerosis and stroke. RESULTS: The image-quality rating was found to be scan indication- And reading radiologist- dependent. Of the 49 test datasets, technologists created a mean ratio of rescans/recalls of (4.7±5.1)/(9.5±6.8) for MS and (8.6±7.7)/(1.6±1.9) for stroke. With thresholds adapted for scan indication and reading radiologist, deep learning created a rescan/recall ratio of (7.3±2.2)/(3.2±2.5) for MS, and (3.6± 1.5)/(2.8±1.6) for stroke. Due to the large variability in the technologists' assessments, it was only the decrease in the recall rate for MS, for which the deep learning algorithm was trained, that was statistically significant (P = .03). CONCLUSIONS: Fast, automated deep learning- based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

AB - BACKGROUND AND PURPOSE: MR imaging rescans and recalls can create large hospital revenue loss. The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. MATERIALS AND METHODS: A deep learning- based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. These series were assumed to be scanned for 2 scan indications: screening for multiple sclerosis and stroke. RESULTS: The image-quality rating was found to be scan indication- And reading radiologist- dependent. Of the 49 test datasets, technologists created a mean ratio of rescans/recalls of (4.7±5.1)/(9.5±6.8) for MS and (8.6±7.7)/(1.6±1.9) for stroke. With thresholds adapted for scan indication and reading radiologist, deep learning created a rescan/recall ratio of (7.3±2.2)/(3.2±2.5) for MS, and (3.6± 1.5)/(2.8±1.6) for stroke. Due to the large variability in the technologists' assessments, it was only the decrease in the recall rate for MS, for which the deep learning algorithm was trained, that was statistically significant (P = .03). CONCLUSIONS: Fast, automated deep learning- based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

UR - http://www.scopus.com/inward/record.url?scp=85061485701&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85061485701&partnerID=8YFLogxK

U2 - 10.3174/ajnr.A5926

DO - 10.3174/ajnr.A5926

M3 - Article

VL - 40

SP - 217

EP - 223

JO - American Journal of Neuroradiology

JF - American Journal of Neuroradiology

SN - 0195-6108

IS - 2

ER -