Drivers and barriers: Innovation in healthcare

What will the imaging of the future be like? And what role will it play within precision medicine? Constant innovation in medical imaging implies the introduction of up-to-date technology platforms and products that can improve diagnostic accuracy for better patient care and improve health outcomes. So, what is the innovation path that a company like Bracco must pursue to always offer added value to its community?

Alberto Spinazzi, Bracco's Chief Medical & Regulatory Officer, answers some of these questions by illustrating future trends and the importance of prevention for human health.

Watch part 1 and part 2 of the interview here

– or read the full interview below the videos:

Q:

Let’s talk about innovation. Imaging has revolutionized medicine, and this is thought to continue apace, if not at a faster pace in the years to come. Research money has been and is being poured into projects and methods to advance medical imaging technology, to extend the reach of human vision. In general terms, how do you define successful innovation in medical imaging?

A:

Successful innovation has to possess two key qualities: it has to be desirable, and it has to be largely adopted by healthcare professionals. The healthcare environment has become increasingly cost-conscious, focused on bending the cost curve down, because budgets are limited, and demand is increasing. At the same time, public expectations are rising. Therefore, the healthcare environment is more and more sensitive to the value being obtained from healthcare expenditure, that is, the clinical benefit achieved for the money spent. So, desirable innovation means any innovation that provides real value to healthcare, either because it meets medical needs, it makes a transformational change for patients, it improves outcomes, and/or because it helps to reduce healthcare spending.

In medical imaging, successful innovation implies the introduction of technology platforms and products that improve the diagnostic accuracy of imaging examinations, which in turn should lead to a positive impact on patient management and improve health outcomes. At the same time, valuable innovation includes products and solutions that improve efficiency and reduce hospital spending. However, the industry developing new medical imaging products and solutions has long realized that there are many different key players, each with clear and specific perspectives of value. 

Q:

Who are those key players?

A:

First, doctors and technologists. Their perception of value gravitates around diagnostic efficacy, safety, reduced invasiveness, ease of use, and improved workflow. Administrators and payors focus on any improvement in efficiency, the ability to carry out more examinations at a lower cost, any positive impact on the cost-benefit of segments of their operations, and the overall patient experience. Patients are interested in reduced waiting times, less invasive and more tolerable examinations and a clear understanding of the risks and benefits of the procedure. A company like Bracco must have all these different needs clearly in mind and develop the evidence needed to show how our innovation provides added value over existing tools and processes for these different key players. All of them.

Q:

Who are the key stakeholders in the development and adoption of innovation in medical imaging?

A:

Well, several. First, academic health centers, which have long been traditional hotbeds of innovation, with access to patient populations, data sets, and biological materials necessary to discover, develop, validate innovation, and where countless opportunities to identify and explore unmet medical needs exist. Moreover, academia is key in supporting the validation of innovation and its’ large-scale adoption and proper use through the development of guidelines, education, and training. Second, governments should create policies that properly reward value-adding innovation, thus allowing for the return on the industry’s investment necessary to reinvest revenues back into the next generation of research.

Third, Regulatory Agencies, have developed and fostered regulatory pathways geared to address the significant challenges of developing innovative technologies (new imaging agents or devices), thus allowing the industry to bring them to patients in the most cost-effective and rapid way. 

Finally, the industry. We must focus on innovation that does provide value to health care. This can be achieved if the industry carefully identifies the (unmet) needs of the various stakeholders, aims at making a transformational change for patients, and looks to do things in a way that has the highest probability of creating better outcomes and of generating a significant impact on healthcare spending. Industry must also support the adoption and diffusion of new technology platforms by providing all the necessary evidence supporting health technology assessments, education, training, and development of guidelines.

Q:

You mentioned several times the imperative to meet medical needs: can you give us an example of unmet needs and future challenges in medical imaging?  

A:

Unmet medical needs persist across many diseases and conditions. Let’s take cancer, for instance: the most critical need and challenge have always been, and still is, its early detection and, possibly, complete removal. Not only would the prognosis of patients with malignant neoplasms be dramatically improved, but also healthcare spending to treat the disease would be remarkably reduced. Early detection means screening healthy and high-risk populations. Imaging plays a key role in the screening and early detection and localization of cancer lesions. 

Screening programs for breast, colorectal, and cervical cancers have clearly demonstrated the power of early cancer detection, but they are still not universally adopted. Bracco has invested and is investing in the development of methods to improve the acceptability and effectiveness of screening, such as virtual colonoscopy, contrast-enhanced mammography, and abbreviated breast MR protocols. Sadly, however, easy screening methods are currently unavailable for some of the world’s deadliest cancers.

Q:

What happens when cancer is not detected early?

A:

Unfortunately, approximately 50% of cancers are at an advanced stage when diagnosed. The chance to eliminate the cancer is much reduced, and the goal becomes to stop or slow the progression of the disease, possibly minimizing short- and long-term untoward side effects of treatment. Finding a way to possibly cure or at least improve survival and quality of care is key. Developing systems that support precision medicine, that is tailoring management to specific subsets of cancer patient populations, or possibly to individuals with cancer is necessary. 

Imaging plays and will continue to play a vital role in precision medicine. Our research programs in several imaging modalities are aimed at integrating the description of anatomy and morphologic parameters with molecular, metabolic, and functional biomarkers and image feature analysis systems to detect imaging phenotypes. The integration of imaging with pathology and clinical data will support adaptive treatment tailoring. Preliminary results are very encouraging, showing that tailoring and monitoring of antiangiogenic treatment and immunotherapy with functional and molecular imaging can improve survival.

Q:

One final question: what is the role that artificial intelligence systems will play in all this?

A:

Well, how much time do we have? There is no doubt that artificial intelligence, or AI, is one of the hottest topics today in medical imaging, and healthcare in general. In a nutshell, my feeling is that the limitless power of computers makes AI an ideal candidate to provide the standardization, consistency, and dependability needed to support medical imaging professionals in their mission to provide high-quality patient care. 

The solutions available today rely on a machine- or deep-learning methodologies that perform various image analysis tasks, such as enhancement of image quality, segmentation, and detection of abnormalities, as well as estimation of the likelihood of malignancy. 

The use of AI in current medical imaging technology is expected to grow dramatically in the next decade. If you think that it took 50 years, from 1950 until 2000, for the amount of data in the medical field to double, and that, by 2010, that timeframe had shrunk to 3.5 years, with data all in digital format. In 2020, the amount of time it took for healthcare data to double was just 73 days. Imaging data availability is also growing exponentially, complemented by massive amounts of associated data-rich medical records, and growing databases of genetic and phenotypic information. As such, the research community is aiming at harnessing the full potential of the wealth of data that are now available at the individual patient level underpinning precision medicine. 

There are still challenges, however. The first challenge posed by AI is the substantial number of high-quality data required to train the algorithms and develop robust AI. AI systems are heavily dependent on a variety of co-factors such as disease subtypes or stages, patient populations, pathology data, genotypes, and phenotypes, among many others. Given a large number of combinations, the data from one single institution are likely to be vastly insufficient for AI algorithms to achieve their full potential. Therefore, sharing data between AI developers is vital for training machine learning algorithms.  Unfortunately, multi-institutional and multi-company collaborations have not often been pursued so far. 

The generation of such high-quality large datasets also requires the validation and standardization of information and data entry, and that also requires multi-institutional collaboration. The use of a pool of relevant retrospective data is further complicated by the ever-changing nature of the clinical practice. Notably, given that the amount of data utilized to train the machine-learning system dictates the confidence in the power of the prediction or judgment of the output, it is necessary to have diversified data, without which there may be inherent inaccuracies for the application of such algorithms among groups who are under-represented in the population under test. Finally, AI may introduce highly consequential systemic errors and it’s not yet clear how to best integrate AI models into clinical workflows. 

All in all, identifying and overcoming the foregoing challenges will be crucial in the development of AI models which are measurably and reproducibly beneficial in clinical settings.