The value of semi-automatic threshold-based segmentation software in cardiac MR for left ventricular non-compaction cardiomyopathy.

Houten, L. van (Luutsen) (2014) The value of semi-automatic threshold-based segmentation software in cardiac MR for left ventricular non-compaction cardiomyopathy. thesis, Medicine.

Text HoutenvanL.pdf Restricted to Registered users only Download (1MB)

Abstract

Aims: Left ventricular non-compaction (LVNC) is a rare myocardial disorder, characterized by a prominent trabecular meshwork and deep recesses. Diagnosing LVNC is difficult, since there is poor agreement between different diagnostic criteria. Genetic defects are identified in approximately 40% of these patients. Our aim was to develop a new cardiac MR method for diagnosing genetically confirmed LVNC and to compare this method to the currently available criteria. Methods: We retrospectively analyzed a group of 47 patients screened for LVNC, because of symptoms, familiar appearance or incidental findings associated with LVNC. Imaging data were analyzed on the Jenni (echo) and Petersen (MR) criteria. Because limitations of these criteria had been described before, Massk CMR method, based on a continuous stack of short axis planes combined with semi-automatic threshold-based segmentation software, was introduced, measuring the left ventricular (LV) trabecular mass as a percentage of total LV mass. For good comparison, receiver operating characteristics (ROC) analysis was performed for all criteria, using genetically confirmed LVNC as golden standard. Results: The Jenni and Petersen criteria disagreed for diagnosis of LVNC in 27 patients (58%). ROC analysis identified a value of LV trabecular mass/ total mass of > 40% as the most valuable cut-off point for distinguishing between patients with and without genetic confirmed LVNC, having an area under the curve of 0.57 [95% confidence interval (CI), 0.39-0.76]. The Jenni and Petersen criteria had an area under the curve of 0.45 (CI, 0.27-0.64) and 0.44 (CI, 0.25-0.63) respectively. Conclusion: Currently available criteria for LVNC show poor agreement and therefore, there is need of new criteria for LVNC, generated prospectively in large multicenter patient cohort studies. The Massk CMR method described is a better tool for diagnosing genetically confirmed LVNC, compared to the currently available echo and MR criteria.

Item Type:	Thesis (Thesis)
Supervisor name:	Willems, Dr. T.P.
Faculty:	Medical Sciences
Date Deposited:	25 Jun 2020 11:01
Last Modified:	25 Jun 2020 11:01
URI:	https://umcg.studenttheses.ub.rug.nl/id/eprint/2137

Actions (login required)

View Item