News Physics World  February 2015

How to create a low-cost CT scanner

Scanning architectures The two scanner designs position the patient vertically (left) and horizontally. (Ge Wang)

X-ray computed tomography (CT) has become an indispensable medical-imaging tool since the first scanners were commercialized in the 1970s. However, scanners remain bulky, lack mobility and are too expensive for many clinics in the developing world. This lack of accessible technology has motivated researchers in the US and China to develop a low-cost, mobile CT scanner suitable for clinics in rural areas in developing countries, war zones and disaster-affected regions (J. X-ray Sci. Technol. 22 673).

The research team, led by Ge Wang of the Rensselaer Polytechnic Institute, has proposed two systems, together with data-acquisition and image-reconstruction techniques. Simulations suggest that the two scanners – which position the patient vertically and horizontally – are feasible, with system costs estimated at around $50,000 and $100,000, respectively.

Unlike conventional rotational CT scanners, the new approach uses the linear translation of an X-ray source and a flat-panel detector positioned either side of the patient to acquire a truncated dataset. The system acquires images at a series of discrete positions as a fan-beam source is moved parallel, but in the opposite direction, to the moving detector. The technique cuts costs by removing the need for an accurately rotating gantry and a costly slip ring that enables the gantry to rotate at high speed during a scan.

Conventional tomographic image reconstruction is inadequate for the truncated projection dataset and would result in poor-quality images. Instead, the researchers propose using interior tomography – a method previously developed by Wang’s group to reconstruct images from sparsely sampled data – combined with compressive sensing to reconstruct a region of interest within the patient, rather than an entire axial slice.

The researchers used simulated scans of a virtual cardiac phantom to assess the system’s performance using one, two or three translations evenly spaced around the patient. Images reconstructed using two or three translations were far superior to those constructed from a single translation.

A key trade-off with the linear-scanning approach is the longer scan duration, with its greater potential for motion artefacts. The researchers estimate that an abdominal scan will take around 10 s using the horizontal-scanner design, but the artefacts may be minimized using a more intelligent reconstruction algorithm, says Wang. “Low-cost CT does not necessarily mean low performance,” he says.

The researchers now plan to build a horizontal-scanning prototype and perform a trial using large animals. Other future investigations include optimizing scan protocols and image processing, and the design of a foldable, portable prototype.

Jude Dineley