Mathematical Modeling Improves Computed Tomography Diagnosis of Traumatic Aortic Injury

David T. Fetzer, Charles Green, O. Clark West

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Rationale and Objectives: Acute traumatic aorta injuries (ATAIs) following blunt thoracic trauma require rapid and accurate diagnosis for institution of lifesaving treatment. The use of computed tomography (CT) in the diagnosis of such injuries continues to improve and has the potential to become the diagnostic modality of choice in many trauma centers. A standardized diagnostic model may contribute to improvements in radiologist interpretation of CT for ATAIs. Materials and Methods: The following diagnostic criteria were used to develop a diagnostic model for ATAIs: 11 areas of potential hematoma formation were identified in the mediastinum. Maximum short- and long-axis cross-sectional diameters of the aorta were measured. Qualitative morphologic information (contour change, intimal flap) was recorded. Smoothness of the aorta wall was assessed. These characteristics were quantified and analyzed for statistical significance, allowing for the development of an injury assessment model. Results: The diagnostic model was used to score 69 blunt thoracic trauma patient cases. Average weighted kappa was 0.74, showing strong agreement among two observers and reproducibility of the model. The model improved injury assessment by classifying equivocal cases as either positive or negative. The ROC curve calculated from the original radiologist interpretation contained 86.1% area under the curve, while the curve for the new model contained 97.5%. The likelihood ratio increased from 30.06 to 48.67. The degree to which the new measure improved prediction over the original radiologist reading was tested using a nested model and yielded a reliable increment in model fit (χ2 analysis: Δχ2(3) = 20.929, P ≤ .0001). Finally, beta weights calculated from each variable were used to create a quantitative best-fit diagnostic model for future use. Conclusion: We have developed a diagnostic tool that may help radiologists better evaluate CT for ATAIs.

Original languageEnglish (US)
Pages (from-to)1244-1253
Number of pages10
JournalAcademic Radiology
Volume13
Issue number10
DOIs
StatePublished - Oct 2006

Fingerprint

Tomography
Aorta
Wounds and Injuries
Thorax
Tunica Intima
Nonpenetrating Wounds
Trauma Centers
Mediastinum
ROC Curve
Hematoma
Area Under Curve
Reading
Weights and Measures
Radiologists

Keywords

  • Aorta
  • aortic rupture
  • computed tomography
  • computer-assisted
  • helical
  • image interpretation
  • models
  • nonpenetrating
  • regression analysis
  • statistical
  • thoracic
  • thoracic injuries
  • wounds

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Mathematical Modeling Improves Computed Tomography Diagnosis of Traumatic Aortic Injury. / Fetzer, David T.; Green, Charles; West, O. Clark.

In: Academic Radiology, Vol. 13, No. 10, 10.2006, p. 1244-1253.

Research output: Contribution to journalArticle

Fetzer, David T. ; Green, Charles ; West, O. Clark. / Mathematical Modeling Improves Computed Tomography Diagnosis of Traumatic Aortic Injury. In: Academic Radiology. 2006 ; Vol. 13, No. 10. pp. 1244-1253.
@article{df93c27287184357a6747ada9b9549b5,
title = "Mathematical Modeling Improves Computed Tomography Diagnosis of Traumatic Aortic Injury",
abstract = "Rationale and Objectives: Acute traumatic aorta injuries (ATAIs) following blunt thoracic trauma require rapid and accurate diagnosis for institution of lifesaving treatment. The use of computed tomography (CT) in the diagnosis of such injuries continues to improve and has the potential to become the diagnostic modality of choice in many trauma centers. A standardized diagnostic model may contribute to improvements in radiologist interpretation of CT for ATAIs. Materials and Methods: The following diagnostic criteria were used to develop a diagnostic model for ATAIs: 11 areas of potential hematoma formation were identified in the mediastinum. Maximum short- and long-axis cross-sectional diameters of the aorta were measured. Qualitative morphologic information (contour change, intimal flap) was recorded. Smoothness of the aorta wall was assessed. These characteristics were quantified and analyzed for statistical significance, allowing for the development of an injury assessment model. Results: The diagnostic model was used to score 69 blunt thoracic trauma patient cases. Average weighted kappa was 0.74, showing strong agreement among two observers and reproducibility of the model. The model improved injury assessment by classifying equivocal cases as either positive or negative. The ROC curve calculated from the original radiologist interpretation contained 86.1{\%} area under the curve, while the curve for the new model contained 97.5{\%}. The likelihood ratio increased from 30.06 to 48.67. The degree to which the new measure improved prediction over the original radiologist reading was tested using a nested model and yielded a reliable increment in model fit (χ2 analysis: Δχ2(3) = 20.929, P ≤ .0001). Finally, beta weights calculated from each variable were used to create a quantitative best-fit diagnostic model for future use. Conclusion: We have developed a diagnostic tool that may help radiologists better evaluate CT for ATAIs.",
keywords = "Aorta, aortic rupture, computed tomography, computer-assisted, helical, image interpretation, models, nonpenetrating, regression analysis, statistical, thoracic, thoracic injuries, wounds",
author = "Fetzer, {David T.} and Charles Green and West, {O. Clark}",
year = "2006",
month = "10",
doi = "10.1016/j.acra.2006.07.002",
language = "English (US)",
volume = "13",
pages = "1244--1253",
journal = "Academic Radiology",
issn = "1076-6332",
publisher = "Elsevier USA",
number = "10",

}

TY - JOUR

T1 - Mathematical Modeling Improves Computed Tomography Diagnosis of Traumatic Aortic Injury

AU - Fetzer, David T.

AU - Green, Charles

AU - West, O. Clark

PY - 2006/10

Y1 - 2006/10

N2 - Rationale and Objectives: Acute traumatic aorta injuries (ATAIs) following blunt thoracic trauma require rapid and accurate diagnosis for institution of lifesaving treatment. The use of computed tomography (CT) in the diagnosis of such injuries continues to improve and has the potential to become the diagnostic modality of choice in many trauma centers. A standardized diagnostic model may contribute to improvements in radiologist interpretation of CT for ATAIs. Materials and Methods: The following diagnostic criteria were used to develop a diagnostic model for ATAIs: 11 areas of potential hematoma formation were identified in the mediastinum. Maximum short- and long-axis cross-sectional diameters of the aorta were measured. Qualitative morphologic information (contour change, intimal flap) was recorded. Smoothness of the aorta wall was assessed. These characteristics were quantified and analyzed for statistical significance, allowing for the development of an injury assessment model. Results: The diagnostic model was used to score 69 blunt thoracic trauma patient cases. Average weighted kappa was 0.74, showing strong agreement among two observers and reproducibility of the model. The model improved injury assessment by classifying equivocal cases as either positive or negative. The ROC curve calculated from the original radiologist interpretation contained 86.1% area under the curve, while the curve for the new model contained 97.5%. The likelihood ratio increased from 30.06 to 48.67. The degree to which the new measure improved prediction over the original radiologist reading was tested using a nested model and yielded a reliable increment in model fit (χ2 analysis: Δχ2(3) = 20.929, P ≤ .0001). Finally, beta weights calculated from each variable were used to create a quantitative best-fit diagnostic model for future use. Conclusion: We have developed a diagnostic tool that may help radiologists better evaluate CT for ATAIs.

AB - Rationale and Objectives: Acute traumatic aorta injuries (ATAIs) following blunt thoracic trauma require rapid and accurate diagnosis for institution of lifesaving treatment. The use of computed tomography (CT) in the diagnosis of such injuries continues to improve and has the potential to become the diagnostic modality of choice in many trauma centers. A standardized diagnostic model may contribute to improvements in radiologist interpretation of CT for ATAIs. Materials and Methods: The following diagnostic criteria were used to develop a diagnostic model for ATAIs: 11 areas of potential hematoma formation were identified in the mediastinum. Maximum short- and long-axis cross-sectional diameters of the aorta were measured. Qualitative morphologic information (contour change, intimal flap) was recorded. Smoothness of the aorta wall was assessed. These characteristics were quantified and analyzed for statistical significance, allowing for the development of an injury assessment model. Results: The diagnostic model was used to score 69 blunt thoracic trauma patient cases. Average weighted kappa was 0.74, showing strong agreement among two observers and reproducibility of the model. The model improved injury assessment by classifying equivocal cases as either positive or negative. The ROC curve calculated from the original radiologist interpretation contained 86.1% area under the curve, while the curve for the new model contained 97.5%. The likelihood ratio increased from 30.06 to 48.67. The degree to which the new measure improved prediction over the original radiologist reading was tested using a nested model and yielded a reliable increment in model fit (χ2 analysis: Δχ2(3) = 20.929, P ≤ .0001). Finally, beta weights calculated from each variable were used to create a quantitative best-fit diagnostic model for future use. Conclusion: We have developed a diagnostic tool that may help radiologists better evaluate CT for ATAIs.

KW - Aorta

KW - aortic rupture

KW - computed tomography

KW - computer-assisted

KW - helical

KW - image interpretation

KW - models

KW - nonpenetrating

KW - regression analysis

KW - statistical

KW - thoracic

KW - thoracic injuries

KW - wounds

UR - http://www.scopus.com/inward/record.url?scp=33748474190&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33748474190&partnerID=8YFLogxK

U2 - 10.1016/j.acra.2006.07.002

DO - 10.1016/j.acra.2006.07.002

M3 - Article

VL - 13

SP - 1244

EP - 1253

JO - Academic Radiology

JF - Academic Radiology

SN - 1076-6332

IS - 10

ER -