Paolo Pellegretti, Head of R&D of Ultrasound systems Esaote

The intuitive image: immediate diagnostic reading, made possible by the complexity of research

Paolo Pellegretti, Head of R&D of Ultrasound systems Esaote

Paolo Pellegretti, Genoa, Italy, 1963. 
Holder of a PhD in electronic engineering, he worked in various technical roles at Esaote until 1995, when he took up his current position as Head of Research and Development of ultrasound systems. Photography and travel are two of his great passions.

Every image triggers a process of interpretation and resolution. We also look to gaming technologies and computer graphics to help with patients' medical histories.

An ultrasound image tells the doctor a story and, like any good story, it offers all the necessary elements to allow those who explore it to better orient themselves in their decisions, to choose what is important and what is not, to investigate, act and resolve.

The people at Esaote’s R&D department know that their role is to create the right bridge between the image and its interpretation, known in medicine as diagnosis. As in the case of photography, a good ultrasound image maintains the complexity of the story of the data but does not confuse, lead to dead ends, or risk becoming muddled. It leaves room for the experience of the interpreter, while keeping the observer focused on the experience of the object they are observing, albeit expanding the field of observation from the experiences of those who have already gone through the same set of questions. The function of an ultrasound image is to convey an important message to the physician or operator through an effective balance of information. In order to enable the correct diagnosis, this message must be carried as neutrally as possible, not masked or altered by elaboration or interpretation. The more we manage to make the reading of an ultrasound image effective and immediate, the more we can say we have done a good job – similarly to how a photographer communicates with their audience.

A story conveyed by a photographer in a single image can be seen as the equivalent of a diagnosis based on a single image, made by a researcher in our research and development laboratories. The immediacy and ease of reading of a photograph is equivalent to the simplicity of interpreting an ultrasound image, which is a reflection of the technological and methodological complexity required to do so.

The story in a single image can also be interpreted in the medical diagnostic industry as the need to pursue productivity and profitability in the process of performing an examination. Here again, we come across the challenge of bringing together, ideally in a single meaningful image, all the information needed to make a fast and reliable diagnosis – in other words, a summary of complexity – and to pursue simplicity in the diagnostic process. This is another important aspect of our concept of complexity, which is aimed at simplification. Due to its complexity, the story may not be told by a single image for various reasons, ranging from the heterogeneity of the factors and concepts to be summarized, to not having seized the moment that summarizes the intended picture.

This is another level of complexity currently being addressed; Esaote is a pioneer in this field with the real-time integration of the ultrasound technique with volumes from other acquisition modalities, such as MRI and computerized axial tomography. The synergistic fusion of multiple diagnostic channels is an effective response to the need to provide comprehensive information in order to make fast and reliable diagnoses.

Image fusion, however, also presents its own particular complexity in the process of volumetric registration of different information channels. The procedures available today require specific training on the part of operators and are long and complex to implement. Esaote is committed to simplifying these procedures through the use of Artificial Intelligence (AI) techniques, with the ultimate goal of providing a solution for their complete automation.

Despite all its positive features, ultrasound diagnostics is however also characterized by the fact that the quality and accuracy of the diagnosis is strongly operator-dependent.

This is another complexity that we are tackling, by developing solutions aimed at simplifying the use of the devices, making them more intuitive and making the information they produce increasingly quantitative, easy and unambiguous to read. This is a response to the need for practicality that, in emergency situations and when the system is under stress, has proved to be a priority for the proper functioning of the healthcare system.

We are also working to refine the diagnostic possibilities by developing new processing tools, delivered via other advanced areas such as radar and sonar imaging for military and commercial applications. All this has been made possible by the ever-increasing processing power offered by technological developments in leading sectors such as computer graphics, gaming, and home entertainment.

These new and powerful investigation tools, if not properly adapted to our specific field, could however end up making the use of new ultrasound platforms more complex. Hence the need for simplification, which can be effectively addressed using AI techniques. By exploiting prior knowledge, which is suitably structured and encoded by learning engines, such as neural networks, these techniques can be used to assist the operator as they use the equipment, as well as to support interpretation and reporting.

However, AI brings with it two complexities, namely that performance is highly dependent on the level of knowledge available and, more generally, on a high computational burden. The simplification model under consideration is the extension of the “Internet of Things” concept to ultrasound equipment, i.e. the adoption of distributed processing and knowledge models in which the ultrasound machine is no longer an independent diagnostic tool, but a terminal in a cooperative network of units that work together and share knowledge and computational resources.

A final area of research and development in which we are actively engaged is looking for new ways to exploit the information that can be obtained from the ultrasound-tissue interaction. In fact, it has been shown that tissue texture analysis can be conducted using ultrasound as well as investigations of attenuation and diffusion rate variation. Research is underway into the correlation of these new physical parameters with pathological effects and, therefore, to construct their own semiotics of reading. This plethora of alternative ultrasound investigations can be seen as a single line of applied research, summarized by Esaote’s concept of “never stop seeing the unseen”. In other words, “search for what is not immediately visible”.

The challenges in this area are obviously very complex but, at the same time, they are exciting and rewarding. Here again, our role is to develop the technology, maximize the sensitivity of our equipment to the monitoring and measurement of various new biophysical parameters, and make it available for use by means of appropriate simplifications for users. Success in this line of advanced research comes from cooperation between industry and research institutions that, due to their formation and purpose, are naturally the most productive and innovation oriented. The task of companies like ours is to establish cooperation with these bodies, supporting their initiatives both technologically and with human and financial resources.

Research and Development activity is the beating heart of Esaote and our strong financial commitment in this sector (more than 10% of investment) is proof of how innovation is at the center of the company’s growth strategies within a context of strong and continuous transformation.

Stay updated on Esaote's world

Don't miss the opportunity to stay updated on upcomings events, educational resources, and all latest from Esaote.

Login required to access this resource.
Go to login page