iTRIM - A Novel CT Image Reconstruction Algorithm to Enhance Temporal Resolution

Not for distribution in the US. iTRIM – A Novel CT Image Reconstruction Algorithm to Enhance Temporal Resolution Sebastian Vogt, PhD, Harald Schöndube, PhD, Thomas Allmendinger, PhD, Johan Sunnegårdh, PhD, Karl Stierstorfer, PhD, Herbert Bruder, PhD, and Thomas Flohr, PhD SIEMENS iTRIM – A Novel CT Image Reconstruction Algorithm to Enhance Temporal Resolution Sebastian Vogt, PhD, Harald Schöndube, PhD, Thomas Allmendinger, PhD, Johan Sunnegårdh, PhD, Karl Stierstorfer, PhD, Herbert Bruder, PhD, and Thomas Flohr, PhD One of the most relevant attributes of a CT scan Alternatively to the more complex hardware designs, mode for cardiac imaging is its temporal resolution. we developed an advanced image reconstruction Utilizing conventional image reconstruction algorithm, dubbed iTRIM (Iterative Temporal algorithms, the highest achievable temporal Resolution Improvement Method), which is designed resolution in the isocenter of a CT image is to reduce temporal resolution of cardiac CT images determined by the time the scanner requires to to values below half the gantry rotation time of acquire 180º of CT projections [1]. This amounts single source scanners with moderate rotation times to a temporal resolution of half the gantry rotation [4]. Based on a novel iterative image reconstruction time for single source scanners and approximately algorithm, iTRIM improves the temporal resolution of a quarter of the gantry rotation time for our to a value of about 80% of that of conventionally Dual Source systems. Hence, high-end CT scanners reconstructed images. As such, the algorithm is owe their cardiac imaging performance and capable of effectively reducing motion artifacts in fast acquisition speeds to a markedly increased CT images at moderate gantry rotation times while complexity of the scanner hardware. still maintaining a good overall image quality and low image noise. An evaluation of clinical images will follow the description of the underlying iTRIM algorithm below. 1 More than quickscan data Cardio CT data Subset of cardio CT data Iteration Start Low temporal resolution Motion magnitude Number image of iterations Local adaptation of histogram Histogram High temporal High temporal resolution resolution image image mix Final result Fig. 1: Flow chart of the iTRIM algorithm, which is based on a novel iterative reconstruction scheme with a local histogram constraint and which is designed to enhance temporal resolution of cardiac CT images below the time needed to acquire 180º of CT projections. 2 The iTRIM algorithm is based on the observation according to the respective histogram of the low that the presence of motion artifacts does not temporal resolution image: Pixels with an HU value significantly change the histogram of a CT image. close to a maximum of the histogram are left Following the idea of an earlier published iterative unchanged; pixels with an HU value far from any algorithm for extended field reconstruction [2] this maximum are adjusted slightly towards the closest information is used to reconstruct an image from maximum. less than half a helix turn of data. Figure 1 depicts a flow chart of the iTRIM algorithm. First, a cardiac To maximize dose usage, a motion detection short-scan weighted filtered back projection (WFBP) technique is used in a last step after the iteration has reconstruction is performed, resulting in a temporal finished. The iTRIM image from the final iteration of resolution equivalent to 180º of rebinned CT the iterative reconstruction is combined with the data. For each pixel the system then computes a 180º WFBP image in the following way: In image histogram within a quadratic region centered on regions that exhibit motion the iTRIM image is used the pixel, as shown in Figure 2. Then, an iterative for the final image, whereas in static regions the reconstruction algorithm using only a subset of WFBP image is used. Motion detection is realized by the full 180° cardiac data set is started. The length reconstructing a second image at a slightly different of this subset (e.g., 140°) is adapted to the cardiac phase and then computing the difference target temporal resolution. In order to expedite between the low resolution WFBP image and this convergence, the low temporal resolution WFBP motion reference image. Using a non-isotropic image is used as a start image for this iterative filter the resulting difference image is furthermore algorithm. The iterative loop then consists of smoothed. The final CT image is created by using two steps: First, the image is updated with the the image regions from the WFBP image where the projection data subset defined above using the SART filtered motion difference image stays below a first (Simultaneous Algebraic Reconstruction Technique) threshold and by using the image regions from the iterative reconstruction framework [3]. After every iTRIM image where the filtered motion difference SART iteration, an additional step is performed image exceeds a second, higher threshold. To ensure where the HU values of each pixel are adjusted a smooth image appearance, a linear combination 2 N 32 32 -1000 0 HU 512 512 Fig. 2: For each image pixel a local histogram is generated, which is virtually unaffected by the presence of motion artifacts and which functions as a constraint during the iterations in iTRIM. (Raw dataset courtesy of Shanghai Jiangong Hospital, Shanghai, People’s Republic of China) 3 3 WFBP iTRIM Fig. 3: Reconstructions using a cardiac WFBP (top row) and the iTRIM algorithm (bottom row). The examination was performed on a scanner with a rotation time of 500 ms and at a patient heart rate of 60 bpm. (Raw dataset courtesy of Shanghai Jiangong Hospital, Shanghai, People’s Republic of China) of both the WFBP image and the iTRIM image is with a gantry rotation time of 500 ms. The reduction employed in those image regions where the value of in motion artifacts using the iTRIM algorithm is the motion difference image lies between the two clearly visible, while maintaining the same noise thresholds. level and overall image quality. As scientifically validated in [4], the temporal resolution of the iTRIM Figure 3 shows image examples of a cardiac dataset, reconstruction technique is below 200 ms and thus reconstructed with the typical cardiac WFBP, in direct far superior to the temporal resolution of 250 ms comparison to a iTRIM reconstruction of the same of the corresponding cardiac WFBP reconstruction. dataset. The sinogram (raw) data was acquired 4 4 180° & WFBP 140° & WFBP 140° & iTRIM Fig. 4: CT reconstructions of another raw dataset, from left to right: (a) with a cardiac WFBP utilizing the usual 180º of CT data, the image exhibits motion artifacts; (b) with a cardiac WFBP utilizing only 140º of CT data, resulting in severe limited view angle artifacts; (c) with the iTRIM algorithm utilizing 140º of CT data in the iterative loop, resulting in an image without motion or limited view angle artifacts. (Raw dataset courtesy of Medical University of South Carolina, Department of Radiology and Radiological Science) The three images in Figure 4 demonstrate the On the new Siemens SOMATOM Perspective with a effectiveness of the iTRIM algorithm with a second rotation time of 480 ms iTRIM delivers an equivalent dataset, which was also acquired on a scanner with rotation speed of 390 ms and an effective temporal a gantry rotation time of 0.5 s and at a patient heart resolution of 195 ms, far superior to the temporal rate of about 64 bpm. The typical cardiac WFBP resolution of 240 ms of the corresponding standard reconstruction with a temporal resolution of 250 ms cardiac WFBP reconstruction. exhibits visible motion artifacts around the coronary artery. When using less than 180° of CT data for the In summary, our novel iterative image reconstruction WFBP reconstruction the motion artifacts can be algorithm utilizing a histogram constraint provides a reduced, as exemplified by a second cardiac WFBP superior temporal resolution, reduced to a value of reconstruction with only 140º of CT data. However, about 80% of that of conventionally reconstructed the resulting image is distorted by severe limited CT images, while maintaining the same overall view angle artifacts rendering the image non- image impression. diagnostic. In comparison, the iTRIM algorithm, which also utilizes only 140º of CT data within its iterative reconstruction loop, yields a diagnostic image without limited view angle artifacts. In addition, the iTRIM image is free of the motion artifacts that first WFBP image exhibits while image noise and overall image impression are maintained. [1] Ohnesorge, B. M., Flohr, T. G., Becker, C. R., Knez, A., [3] Kak, A. C. and Slaney, M., Principles of Computerized and Reiser, M. F., Multi-slice and Dual-source CT in Cardiac Tomographic Imaging, IEEE Press (1998), available online Imaging, Springer Verlag, Berlin, second ed. (2007). at http://www.slaney.org/pct/pct-toc.html. [2] Kunze, H., Härer, W., and Stierstorfer, K., Iterative extended [4] Schöndube, H., Allmendinger, T., Stierstorfer, K., field of view reconstruction, in Medical Imaging: Physics Bruder, H., and Flohr, T., Evaluation of a novel CT image of Medical Imaging, Hsieh, J. and Flynn, M. J., eds., Proc. reconstruction algorithm with enhanced temporal resolution, SPIE 6510(1), 65105X (2007). Proc. SPIE 7961, 79611N (2011); doi:10.1117/12.877734 5 In the event that upgrades require FDA approval, Siemens cannot predict whether or when the FDA will issue its approval. Therefore, if regulatory clearance is obtained and is applicable to this package, it will be made available according to the terms of this offer. On account of certain regional limitations of sales rights and service availability, we cannot guarantee that all products included in this brochure are available through the Siemens sales organization worldwide. Availability and packaging may vary by country and are subject to change without prior notice. Some/All of the features and products described herein may not be available in the United States. The information in this document contains general technical descriptions of specifications and options as well as standard and optional features which do not always have to be present in individual cases. Siemens reserves the right to modify the design, packaging, specifications, and options described herein without prior notice. Please contact your local Siemens sales representative for the most current information. Note: Any technical data contained in this document may vary within defined tolerances. Original images always lose a certain amount of detail when reproduced. Siemens Healthcare Headquarters Siemens Healthcare GmbH Henkestr. 127 91052 Erlangen Germany Phone: +49 9131 84 0 siemens.com/healthcare siemens.com/healthcare CC CT online © Siemens Healthcare GmbH, 2016