Automatic determination of minimal cardiac motion phases for computed tomography imaging: initial experience

Abstract

Low motion phases for cardiac computed tomography reconstructions are currently detected manually in a user-dependent selection process which is often time consuming and suboptimal. The concept of motion maps was recently introduced to achieve automatic phase selection. This pilot study compared the accuracy of motion-map phase selection to that with manual iterative selection. The study included 20 patients, consisting of one group with low and one with high heart rate. The technique automatically derives a motion strength function between multiple low-resolution reconstructions through the cardiac cycle, with periods of lowest difference between neighboring phases indicating minimal cardiac motion. A high level of agreement was found for phase selection achieved with the motion map approach compared with the manual iterative selection process. The motion maps allowed automated quiescent phase detection of the cardiac cycle in 85% of cases, with best results at low heart rates and for the left coronary artery. They can also provide additional information such as the presence of breathing artifacts. Motion maps show promise as a rapid off-line tool to automatically detect quiescent cardiac phases in a variety of patients.

Detection of Relevant Coronary Artery Disease Using Dual-Source Computed Tomography in a High Probability Patient Series Comparison With Invasive Angiography

“…6 The first image reconstruction was performed at 70% of the RR-cycle, followed by an automatically generated reconstruction using dedicated reconstruction software, which automatically generates the most quiescent phase during the RR-cycle by calculating a motion strength function between several reconstructions at low resolution over the cardiac cycle and identifying periods of low difference between neighboring phases (BestDiast/ BestSyst  , Siemens Medical Solutions). 18 If motion artefacts were present, additional reconstructions in 5% increments and decrements were rendered, as described. 5,[19][20][21][22] For image reconstruction, a slice thickness of 0.6 mm and an increment of 0.3 mm were chosen.…”

Automaticvssemi-automatic global cardiac function assessment using 64-row CT

“…In addition to this we also used the scanner's motion map-based software to automatically determine and reconstruct an image series at the time of least cardiac motion during systole (auto-systole) [20].…”

“…As a third region-growing approach, instead of using all 10 phases from the cardiac circle, we used only the phase centred on 0% as the diastolic phase, and the autosystole phase, determined by automatic motion mapping [20], as the systolic phase, applying the same method for manual adjustment as described above.…”

“…The time taken was reduced by 85% (to less than 2 min) for the automatic approach and by 75% (to less than 3 min) for the manually adjusted approach, a significant reduction in each case. The time needed was even further reduced using only two phases, obtained by using motion mapping [20], resulting in a reduction of 95% to less than 1 min. However, using only two phases resulted in only moderate correlation with the short-axis approach for EF, and it also showed a significant overestimation of myocardial mass and significantly larger limits of agreement for EF and ESV with respect to the region-growing approach using 10 phases, thus proving to be inferior to the manually adjusted region-growing approach.…”

“…The aim of this study was therefore to evaluate an automatic four-dimensional (4D) analysis of a regiongrowing segmentation algorithm and motion mapping to automatically identify end-systole [20] for the assessment of LV function, taking into account the time needed and the degree of interobserver variability, as well as accuracy when compared with an established semi-automatic short-axis approach.…”