Rapid signals, sharp images, clear diagnoses

High-resolution three-dimensional images, even in dense breast tissue – this is the promise of ultrasonic computer tomography. The imaging procedure which could revolutionise breast cancer diagnostics is currently being developed at a research centre at the Karlsruhe Institute of Technology (KIT).
In the Western world, the incidence of breast cancer stands at some 10 percent. The earlier the suspected nodules in the breast are detected, the better are patients’ prospects of recovery. Currently, the average size of tumours at the time of diagnosis is one centimetre. When the tumour is this size, there is a 30 percent likelihood that metastasis, associated with poor prognosis, has already occurred.

To equip medicine with a better set of diagnostic instruments, scientists at the Institute for Data Processing and Electronics (IPE) are attempting to build the world’s first 3D Ultrasound Computer Tomography Scanner (3D-USCT) to generate high-resolution three-dimensional images of the female breast.

The essential elements of the measurement technology of this new system, the ultrasound converters, were specially developed for this project at the IPE’s production centre. Although this type of converter can work effectively as an ultrasound transmitter and receiver at the same time, the functions are kept separate in order to optimise signal quality. The current demonstrator works with 628 transmission and 1,413 reception elements.

To obtain the images, the ‘transmission converters’ transmit the ultrasound into a water-filled semi-ellipsoid which acts as the examination receptacle for the breast. Breast tissue modifies the ultrasonic signals as a function of the specific properties of the particular tissue types. The signal waves are broken up, reflected and damped on the outer edges of the tissue. Initial experiments of scientists as far back as the 1970s showed that glandular tissue, fat tissue and certain carcinomas could be clearly distinguished from each other because of the differing speeds of sound passing through them. The ultrasound receivers can receive signals in any combination. “Small movements in the entire sensor system make it possible to raise the number of perspectives,” reveals project head Dr. Nicole Ruiter. To able to deal with breast tissue, the system works with a mean frequency of 2.5 MHz.

The way ahead from here sounds quite simple in principle. The hardware, specially designed for this task, digitalises and processes the captured raw data and forwards it to a data acquisition computer. The computer initialises all the measurements and stores the data obtained from this. Using suitable calculation algorithms, software then creates the volume images to enable 3D reconstruction – something which is not quite as simple as it sounds.

The reason is that this process involves such huge volumes of raw data that, given the current state of technical progress, the procedure still needs several weeks of computer time to generate a maximum resolution three-dimensional image. About one billion image points need to be calculated from the 20 GBytes of raw data.

A group of IPE specialists spearheaded by Matthias Balzer is working on designing hardware capable of handling very large data volumes and high computer performance levels. The developers have ingeniously decided to use Field Programmable Gate Arrays (FPGAs), which are already available for data logging, as computers at the same time. “The 80 parallel FPGAs provide us with an enormous speed advantage. And using a dynamic configuration – i.e. by re-using available chip space for several calculations – we get, virtually speaking, several processors on the chip”, explains Dr. Michael Hübner from the Institute for Information Processing Technologies (ITIV). And this elegant solution brings a further benefit – it cuts energy consumption.

The scientists’ remaining tasks include the gradual enhancement of image reconstruction effectiveness, improving data compression and signal processing and optimising hardware architectures. The aim is to find the most efficient way possible to bundle different computer architectures. It is not just a matter of achieving a high level of computing power – it also involves the flexible handling of algorithms, which are being constantly developed.

Components of the image quality

High image quality depends on numerous components. Important measurement factors to be addressed are: The arrangement and number of ultrasound converters in the room, their bandwidth and aperture angle. Also important are the damping of ultrasonic signals by the breast and water and the exact position of the converters and the accuracy of the known sound speed of the object being measured and of water. It is also necessary to have precise knowledge of the ultrasonic properties of the water. For example, the sound speed varies with the temperature of the water. Air bubbles and impurities in the water can also increase the damping effect.

Each USCT record simultaneously produces three different images of the different physical properties of the breast, i.e. reflectivity, sound speed and absorption. A crucial factor to be addressed when attempting to generate high quality images is that each of these properties is altered by tumour tissue. So, the greater the amount of data that can be obtained on each image point, the more definitely can a distinction be made between ‘normal’ and ‘abnormal’. This can be of vital importance, especially in diagnostically difficult cases.

This graphically illustrates the advances that sophisticated USCT offers. It holds out the prospect of detailed images in which even significantly smaller tumours could be detected.

Unlike with X-ray mammography, with this system the various breast tissues are not merely presented in a composite projection. Instead, they are shown as three-dimensional images accessible from all sides. Because the ultrasonic data is received in the USCT method from all directions, reflection tomography regularly delivers illuminated volumes at resolutions of up to 200 micrometres.

This enables a differentiated view of a tumour even when it is located in dense glandular tissue. Modern X-ray mammography cannot detect tumorous growths until they are one centimetre in size on average. The target set for the USCT system is accurate diagnosis of tumours with an average size of only 5 millimetres. “The likelihood that metastasis has already occurred at this stage,” stresses Nicole Ruiter “would only be 5 percent.”

With X-ray mammography, breast tissue, given its elasticity, is deformed when compressed between two plates and so each examination results in different deformations. This makes it difficult to compare mammography images with each other. Similar problems also occur with hand-controlled conventional ultrasound equipment.

The difference with USCT breast examinations is that they are conducted with the patient in a reproducible position and the breast is not compressed, so 3D images obtained at different times can easily be compared with each other. Also, unlike with X-ray based procedures, there is no harmful ionising radiation to consider, so images can be taken as often as necessary without the risk of injury from that source.

Moreover, USCT screening comes with a significant level of comfort. During the examination, the patient lies on her stomach on a couch which has recesses for the breasts. Since the breast is not compressed, women find that the procedure is considerably more pleasant. The depression is filled with water at body temperature, so the breast is freely suspended.

Unlike the conventional ultrasonic method, with USCT images of the breast are taken from all directions. This reduces the number of image distortions, which considerably improves image quality. It is true that differentiated 3D images of breast tissue can also be obtained with magnetic resonance tomography (MRT), but the work involved in this procedure is much greater and it is considerably more expensive. Future application options for USCT systems will be check-up and preventive examinations of all kinds. Young women in particular could benefit from this method, and also women with dense glandular tissue because small growths have so far been difficult to detect in this sort of tissue using conventional mammography. The procedure will also be appropriate by way of additional clarification if the results of mammography, MRT and conventional sonography prove difficult to interpret.

But that all still lies in the future. The immediate goal of KIT researchers is acceleration of the procedure by a factor of 100. The long-term goal is data processing in real time. However, according to estimates, another three to five years are needed before the 3D demonstrator will be ready to go on the market as a diagnostic tool. Before this can happen, the ultrasound computed tomography scanner will first have to prove its fitness for use in clinical studies.

The Deutsche Forschungsgemeinschaft (DFG) is providing two years’ funding for the collaborative project of the Institute for Data Processing and Electronics (IPE) under the direction of Prof. Marc Weber and the Institute for Information Processing Technologies (ITIV) under the direction of Prof. Jürgen Becker.

Dr. Helmut Bruckner

German Summary

Brustkrebs wird häufig erst entdeckt, wenn sich bereits Metastasen gebildet haben. Mit ihren hochauflösenden und reproduzierbaren Bildern verspricht die 3D-Ultraschall-Computertomographie (USCT) nach den ersten Testläufen ein Verfahren zur frühzeitigen Erkennung von Brustkrebs zu werden. Doch bis daraus ein Standard-Verfahren wird, gibt es noch einige Probleme zu lösen: Um die enormen Rohdatenmengen in Echtzeit verarbeiten zu können, müssen die Rechner-Architekturen und Algorithmen des Systems noch erheblich verbessert werden. Doch sobald die Technologie optimal läuft, verspricht sie außerordentliche Vorteile: detailgenaue Bilder sind in 3-D-Format erhältlich und beliebig reproduzierbar, die Untersuchungen sind für die Patientinnen völlig schmerzfrei und ungefährlich, da sie mit keiner Strahlenbelastung verbunden sind. Langfristig könnte USCT die bisherige Röntgentechnik-Mammographie ersetzen, auf der die bisherigen Brustkrebs-Screenings basieren. Der deutschsprachige Beitrag ist nachzulesen auf www.meditec-international.com/medi0511usct