On the cover of a prestigious magazine: FIT’s supercomputing technologies have been recognised by the professional community

IT research applied in critical areas is the icing on the cake of the academic activities at FIT, Brno University of Technology. This is doubly true of technologies related to healthcare. Approximately one in twenty women worldwide will develop breast cancer during their lifetime; there are currently 2.3 million cases a year. In the vast majority of countries, it is the most common type of cancer affecting women. The Supercomputing Technologies research group (SC@FIT), led by Prof. Jiří Jaroš, has been involved for many years in the major European project PAMMOTH. This project focused on the development of a new non-invasive breast imaging method with the aim of improving the accuracy of identification and facilitating the assessment of suspicious findings. The project’s results are now summarised in an article, to which our experts Jiří Jaroš, Gabriel Bordovský and Jakub Budiský contributed, and which has been published in the prestigious (Q1) journal Photoacoustics by Elsevier. The editors also selected it specifically for the cover of the current issue. What made the research so compelling to the scientific community?

More gentle and accurate thanks to a combination of methods

The main output of the project is the new PAM3 imaging system. It is capable of detecting common cysts, calcifications and other pathological phenomena that may develop into cancerous growths and which current technologies may overlook or misinterpret. PAM3 combines photoacoustic and ultrasound tomography into a single three-dimensional breast examination model. Thanks to this combination, it is able to obtain not only a detailed image of the tissue structure, but also information about its functional properties. Photoacoustic tomography uses very short laser pulses that safely illuminate the tissue being examined. Different types of tissue (e.g. fat, muscle or skin) absorb light in different ways. This absorption generates weak acoustic waves, which are detected by the device and processed by the software. Based on these, the system is then able to distinguish between different types of tissue and obtain information on, for example, blood oxygenation. However, to create an accurate photoacoustic image, it is necessary to know the speed of sound in individual parts of the tissue. This is first determined by the ultrasound component of the system. The resulting sound velocity map is then used to refine and sharpen the photoacoustic image so that even very fine details are visible. It is precisely the combination of both imaging methods with advanced reconstruction algorithms that represents the main benefit of the PAM3 system, as Professor Jaroš confirms: “A key advantage of photoacoustic tomography is the use of a combination of light and ultrasound instead of the ionising radiation used in conventional mammography. The examination therefore does not expose the patient to radiation and can be repeated more safely, for example during regular health monitoring. Furthermore, unlike computed tomography, the method does not require the use of contrast agents. Compared with a standard ultrasound scan, it can provide more detailed information about the structure and properties of tissues, which helps to better distinguish healthy tissue from potentially cancerous changes.”

During the examination, the breast is placed in a transparent, cup-shaped casing filled with fluid. An array of 512 ultrasound sensors and 40 optical outputs rotates around it, illuminating it from many directions. The benefit to the patients themselves, as already mentioned, is crucial: photoacoustic imaging operates without X-rays and without the administration of a contrast agent. The research has therefore laid the foundations for gentler and more accurate imaging of breast tissue; its clinical application will now be tested in a study involving patients.

The results of the PAMMOTH project include not only the examination device itself, but also the development of an entire technological platform. This comprises, for example, a tunable laser system, highly sensitive ultrasound transducers, special test objects and breast models, as well as methods for three-dimensional image reconstruction and algorithms for correcting photoacoustic imaging. Alongside our researchers, experts from eight other European research institutions also took part in this long-term project. “As always, our team was involved in implementing the software for this device. This included a network interface through which data is received from the scanner to the computer, as well as the complete software that generates the final image from the measured data. We devised ways to accelerate the reconstruction technique so that the entire process could be handled on a single computer,” says Jiří Jaroš, commenting on the role of the experts from the SC@FIT research group.

A key challenge for the future is the speed of calculations. The authors plan to significantly speed up the algorithms, using, amongst other things, multi-level methods and machine learning. “One of the main technological challenges of the project is speeding up the processing of large volumes of measured data. The photoacoustic part of the system is already at an advanced stage of development, and thanks to the use of the computing power of graphics cards, it can generate the final image in less than 30 minutes. The ultrasound component is currently at the prototype stage, and its computational demands are significantly higher. Whilst a single photoacoustic examination generates approximately 50 GB of data, the ultrasound component can generate up to 500 GB. “Further development is therefore focused primarily on optimising the algorithms and speeding up the entire process so that the system can be used effectively in clinical practice,” says Jiří Jaroš, summarising the immediate goals. The next important step will be extensive clinical trials, during which doctors will verify the benefits of the acquired image data and propose procedures for their correct interpretation. The aim is to enable the most accurate possible detection of tumour tissue whilst keeping the number of false-positive findings to a minimum.

We congratulate the researchers from the Supercomputing Technologies group on their publication in the journal – and, above all, on the significant, highly applicable outcomes of their long-term research.