CT reconstruction

Reconstruction algorithm development

We develop state-of-the-art CT reconstruction algorithms, including iterative reconstruction (IR). Efficient algorithm design and GPU-optimized implementation enable CT image reconstruction in a short time.

In CT imaging, X-rays emitted in a fan shape from the X-ray tube are received by a flat-panel detector (FPD). In the conventional acquisition geometry (full-fan), the FPD is centered on the beam, so the imaging range (reconstruction FOV) is limited by the FPD size, which makes it difficult to image large subjects such as larger patients. Our product therefore also supports the "half-fan" geometry, in which the FPD is shifted laterally (offset) and the gantry is rotated once to greatly enlarge the reconstruction FOV. Iterative reconstruction is applied to half-fan acquisitions as well, producing high-quality CT images.

Improved spatial resolution

Recently, an X-ray device has been installed in the rotating gantry of a radiotherapy device to obtain CT images (cone beam CT: CBCT). Although the image quality is inferior to that of diagnostic CT images, it has the great advantage of being able to acquire 3D images of the treatment position/condition. However, since linacs and particle beam rotating gantry systems are larger than diagnostic CT systems, the position of the X-ray tube and detector changes during rotation, and the sensitivity of the detector is low.

Ring artifact correction

If the sensitivity correction of the detector is not successful, ring artifacts (yellow arrows) will occur in the CT image. In the case of CBCT, the position of the X-ray machine shifts with each gantry angle, so the ring artifacts are not perfectly concentric and cannot be corrected easily. By processing the image before reconstruction, it is possible to correct artifacts with higher accuracy.

CBCT artifact

CBCT has the problem that correct image reconstruction is more difficult as it approaches the end in the body axis direction. This is because the X-rays are emitted in a cone shape, so the front area on the detector side is irradiated with X-rays, but the irradiation area becomes narrower as it approaches the X-ray tube side (missing area). On the other hand, several literatures have proposed methods to correct the missing regions. This product performs artifact correction by effectively using the data remaining in the missing area.

(Left) Before image processing. There are missing areas at the top and bottom edges of the image, and the image is not reconstructed correctly. (Right) After image processing. By effectively using the data of the missing region, CT image reconstruction with reduced CBCT artifacts has become possible.

Phys. Med. Biol. 51 (2006) 3953–3965

Dual-energy CT

Dual-energy CT acquires images with X-rays at two different energies (tube voltages). Because X-ray attenuation depends on both energy and material, it enables discrimination of materials that are hard to separate on conventional CT (bone, iodine contrast, soft tissue, etc.) and estimation of electron density and effective atomic number. In radiotherapy and particle therapy it is expected to improve the accuracy of dose calculation and particle range calculation and to reduce metal artifacts. PIL is developing image reconstruction and image processing technology for dual-energy CT.

Ultra-fast computation

Efficient algorithms and GPU-optimized implementation achieve CT image reconstruction in a short time. Coordinating multiple GPUs delivers even higher speed.