Efficacy and Accuracy of Novel Automated Mitral Valve Quantification: Three-Dimensional Transesophageal Echocardiographic Study

Autor: Tomonori Miki, B S Misako Toki, Kiyoshi Yoshida, Nobuyuki Kagiyama, Keizo Yamamoto, Masahiko Hara, B S Shuichiro Fukuda, Akihiro Hayashida, B S Shingo Aritaka, Minako Ohara, Atsushi Hirohata
Rok vydání: 2015
Předmět:
Male
medicine.medical_specialty
Echocardiography
Three-Dimensional
030204 cardiovascular system & hematology
Sensitivity and Specificity
Pattern Recognition
Automated
Machine Learning
03 medical and health sciences
0302 clinical medicine
Mitral valve
Image Interpretation
Computer-Assisted
Humans
Medicine
Mitral valve prolapse
Radiology
Nuclear Medicine and imaging
Routine clinical practice
030212 general & internal medicine
Functional mitral regurgitation
Aged
Observer Variation
Mitral regurgitation
Mitral Valve Prolapse
business.industry
Mitral Valve Insufficiency
Reproducibility of Results
Image enhancement
Image Enhancement
Software package
medicine.disease
Clinical Practice
medicine.anatomical_structure
Mitral Valve
Female
Radiology
Cardiology and Cardiovascular Medicine
business
Algorithms
Echocardiography
Transesophageal
Zdroj: Echocardiography. 33:756-763
ISSN: 0742-2822
DOI: 10.1111/echo.13135
Popis: Background Previous studies indicated that the three-dimensional features of the mitral valve (MV) have a significant impact on MV disease. However, quantification of MV with manual tracing software was too time-consuming for routine clinical practice. This study was performed to investigate the efficacy and accuracy of MV quantification with a novel highly automated commercially available software package developed for this purpose. Methods Using the manual tracing and automated package, two expert sonographers and one cardiologist individually analyzed three-dimensional datasets acquired with transesophageal echocardiography from 74 patients (15 with functional mitral regurgitation, 32 with MV prolapse, and 27 normal subjects) retrospectively. Time for analysis and inter-observer agreement were compared between the two methods, and agreement of measurements was analyzed using Cronbach's α. Results Time for analysis using the automated package was significantly shorter than manual tracing (whole cohort, 260 ± 65 vs. 381 ± 68 seconds, P < 0.001; functional mitral regurgitation, 234 ± 42 vs. 378 ± 64 seconds, P < 0.001; MV prolapse, 293 ± 69 vs. 407 ± 67 seconds, P < 0.001; normal controls, 235 ± 52 vs. 351 ± 60 seconds, P < 0.001). There was good agreement among all three observers using both methods, and measurements with the automated package agreed well with the manual tracing values. Conclusions The novel automated software package reduced time for quantification of MV with similar accuracy compared to the manual method. Automated quantification is useful and may be a key to widespread adoption of three-dimensional quantification in clinical practice.
Databáze: OpenAIRE
Externí odkaz: