James Robergé

Endocardial border detection in contrast enhanced echocardiographic cineloops using a pulse coupled neural network

Endocardial border detection is a crucial step for the quantitative analysis of overall and regional left ventricular function from echocardiographic images. Echocardiographic contrast agents with transpulmonary passage are being used increasingly in clinical practice for left ventricular cavity opacification. We investigate the feasibility of using pulse coupled neural networks to identify, the left ventricular endocardial border in contrast-enhanced echocardiographic cineloops.

Temporal analysis of regional synthetic M-mode to identify abnormal stress echocardiographic studies

The authors investigated a method to identify regional myocardial abnormalities based on the temporal pattern of LV contraction. Nineteen stress echo studies were either normal (NML=9) or abnormal (ABN=10) by standard rest and stress images. Six radial, computer-generated synthetic M-modes (SMM) front a short axis cineloop were used to measure change in time to peak contraction during stress (/spl Delta/t) for each segment. Mean /spl Delta/t for all segments was -38/spl plusmn/15% for NML and -22/spl plusmn/27% for ABN studies (p<0.001). The presence of at least one segment with /spl Delta/t/spl ges/-15% had a sensitivity of 70% and specificity of 89% for detecting ABN studies. Temporal analysis of SMM is a promising technique for detecting resting or stress-induced LV regional abnormalities.

Automated volumetric flow quantification using angle-corrected color Doppler image.

We have developed a fully automated method for measuring volumetric blood flow with angle‐corrected blood velocity from a color Doppler image. By computing the blood flow vector through a conduit, the angle of incidence between the direction of ultrasound beam and the direction of blood flow can be measured to correct the underestimated blood velocity. This correction immediately contributes to the improvement of measurement accuracy. The developed method also enhances the conduit identification procedure that is one of the most important factors affecting the accuracy of volumetric measurement. To evaluate the validity of the developed algorithm, experimental studies had been applied to 21 healthy subjects and 10 patients. Volumetric flows were measured from a color Doppler image of the left ventricular outflow track, which were compared with blood volumes that were measured by traditional pulsed‐wave (PW)‐Doppler technique. The mean stroke volume difference between two methods was −0.45 ± 11.7 (mean ± SD). The proposed algorithm is a viable method for determining blood flow volume in an automated fashion. (ECHOCARDIOGRAPHY, Volume 21, July 2004)

Fully-automated stroke volume determination from digital color flow echocardiographic images

Presents a technique for automatically computing stroke volume from 2-D color flow echocardiographic cineloops of the left ventricular outflow tract (LVOT). By utilizing the anatomical and blood velocity information in the color flow image, the measurement technique eliminated many of the technical obstacles and assumptions in existing methods. This technique determines the blood flow vector, LVOT conduit border, and ejection period without human guidance, and the stroke volume is computed automatically. The resulting automated technique improves measurement accuracy and speed and only requires one data set (a color flow cineloop).

Automated conduit detection method from a synthetic M-mode using spline curves

Previous studies have shown that correct identification of the conduit border is a crucial factor for an accurate stroke volume measurement. This paper describes a method that automatically defines conduit boundaries from a synthetically generated M-mode image of the left ventricular outflow tract (LVOT). Unlike the existing methods that define conduit boundaries based solely on single-frame analysis, this method uses LVOT boundaries of adjacent frames to measure the conduit diameter. By using spline curves that are generated from the neighboring conduit boundaries, this technique estimates the LVOT diameter even in a noisy image where conduit boundaries are not clear. The conduit diameter calculated by this method shows close agreement with manual measurement by two independent observers.

Computing stroke volume from digital 2-D color flow echocardiographic cineloops of the left ventricular outflow tract

We introduce a technique for automatically computing stroke volume from 2-D color flow echocardiographic cineloops of the left ventricular outflow tract (LVOT). This technique eliminates the need for manual specification of the direction of flow and automatically detects the LVOT conduit margins. Feedback as to the accuracy of the LVOT border assignments is provided using a synthetically-generated color M-mode image across the LVOT.

Measuring left ventricular regional wall thickening using dynamic M-mode imaging

In order to measure left ventricular (LV) regional wall thickening, one must accurately track movement of the endocardial and epicardial borders. This task can be exceedingly tedious and error-prone when performed on 2D echocardiographic cineloops. The authors have developed a computer-assisted technique, dynamic M-mode, that allows a user to specify the position of the endocardial and epicardial borders using cursors on a synthetic M-mode image that are synchronized to cursors on the 2D cineloop from which the M-mode was derived. The authors evaluated the accuracy of dynamic M-mode imaging by comparing measurements of wall thickening and rates of wall thickening performed using this technique with those produced using paired transmural ultrasonic crystals implanted on the LV epicardium and endocardium in five open-chested dogs. The results show a close correspondence between the measurements produced by these two methods.

A method for non-invasively computing left atrial pressure

Previous work has shown that left atrial pressure can be non-invasively measured from mitral regurgitation velocity at the onset of left ventricular ejection. Here, the authors introduce a computer assisted approach to computing left atrial pressure based on this technique which reduces the need for manual measurement of times and velocities from the Doppler spectra. This process includes the creation of a signal-averaged mitral regurgitation Doppler spectrum that combines data from multiple beats into a single, more coherent spectrum, with a more clearly delineated velocity envelope.