Results
Twenty-two patients were prospectively included in this study with three
novice operators and two expert operators performing TEE exams. Patient
demographics are shown in Table 1. A total of 44 TEE image acquisition
protocols were acquired (22 manual and 22 automated). The average TEE
image quality score for manual and automated protocols was similar and
not statistically different. (1.9 vs. 1.9 for manual and automated
protocols, p>0.05). Distribution of automated protocols
used can be seen in Table 1 .
The average duration of the automated protocol (10:06 ± 2 min) was
faster when compared with the manual protocol (11:29 ± 2 min,
p<0.05). This difference in study duration between automated
and manual protocols persisted when compared between novice and expert
operators (Figure 2-A) . The TEE protocol performed second after
randomization (i.e. the second protocol performed in each patient) was
shorter regardless of protocol method (manual or automated), however,
the percentage decrease in duration of the second exam in each patient
was significantly greater when the automated protocol was performed
second compared to the manual protocol performed second (27% vs. 6%)(Figure 2-B) .
The automated protocol required fewer sonographer keystrokes than the
manual protocol, regardless of operator experience level (novice 28%,
expert 27% reduction); Figure 2-C ) or the protocol order
(manual first 28%, automated first 33% reduction; Figure
2-D ). Automated protocols required a significantly fewer number of
angle corrections, gain adjustments, 2D optimizations, imaging plane
navigations, and miscellaneous keystrokes (Figure 3) . A total
number of images acquired was similar between automated (43 ± 4) and
manual (44 ± 4 images) protocols (p>0.05). The greatest
difference in specific keystrokes was found in 2D optimization (manual =
27 ± 5 vs automated 7 ± 4, p<0.05) and XYZ navigation (manual
= 36 ± 10 vs automated 8 ± 4, p<0.05; Figure 3 ).
When using the more complex mitral valve protocol, the automated
protocol was on average 1:01 ± 3 minutes faster than the manual protocol
(p>0.05). When using the AS or AF protocol, the automated
protocol was on average 1.43 ± 2 minutes faster (p<0.05).
There was no significant difference in percent decrease in study
duration or study keystrokes when comparing the more “complex” MR
protocol versus the “simpler” AS and AF protocols (Figure 4-A,
B).