AI: the intelligent way to process clinical imaging data

By Annabel Kartal-Allen | Published: 20-Feb-2024

AI can aid the mass extraction of quantitative features from image data and reduce handling costs

The use of Artificial Intelligence (AI) in the pharmaceutical industry has gained significant popularity during the last decade, with many companies involving the software in their daily functioning.

It can be utilised in all stages of pharmaceutical development, from drug discovery to commercialisation and can help to handle medical images, driving novel innovations and medicines. Annabel Kartal-Allen spoke to Fuensanta Bellvís, VP of Clinical Studies at Quibim to find out more.


Simplifying data handling

The pharmaceutical industry is vastly data-driven, and handling large amounts manually can be laborious and inefficient. AI can assist scientists in accurately and rapidly conducting mass data handling: “To handle the huge amount of data involved in these processes, radiomics and imaging biomarkers need to be coupled with AI to maximise efficiency,” explains Ms Bellvís.

AI softwares, with time and data exposure, can learn automatically: “By leveraging computational algorithms provided by AI, we can extract and learn from hidden data patterns. This acquired knowledge then empowers the model to make accurate predictions on previously unseen datasets, leading to cancers being spotted sooner and with more accuracy.”


Like this story? Subscribe to Manufacturing Chemist magazine for the latest news, updates and expert-written articles from the global pharmaceutical and biopharma sectors. For more information click here.


Who Quibim are and what they do

Quibim is a global company striving to bring AI to the forefront of the healthcare industry, with the aim of improving treatment options and, in turn, patient livelihood.

They have been developing their technologies further to enhance medical image analysis, focusing on MRI, CT and PET to detect and treat cancers earlier, as well as determine patient prognosis and predict various primary endpoints.


Reducing costs with AI 

There are significant expenses involved in the large-scale analysis of clinical data, both in time and labour.

However, AI has the ability to positively influence this: “There is definitely a cost-reduction benefit to using AI, although that always comes second to accuracy and the usability of solutions being rolled out,” explains Ms Bellvís.

“When modelled by experts, AI-powered solutions increase overall patient access to precision medicine and novel therapeutics, whilst also improving the efficiency of clinical trials,” she continues. 

Utilising AI to accelerate the approval of therapeutics doesn’t stop at cancer therapy — this initiative could be extrapolated to almost all therapeutic areas.

This can improve patient accessibility to novel treatments,  whilst also aiding pharma companies in identifying compounds for early-stage drug development:

“Imaging AI-based predictive models can empower life sciences companies to delve deeper into medical images, patterns and trends, harnessing this data in a non-invasive, cost-effective and widely accessible manner. This enables the analysis of a variety of data types, ensuring consistent monitoring over time and offering a clear trajectory of disease progression or response to treatment.”


Despite 10M daily imaging exams worldwide, less than 0.1% are used for clinical research or creating AI models


Expanding AI’s footprint in pharma

Quibim believe in the power of AI to enhance pharmaceutical development, so they want to encourage an increase in its influence.

The company plans to do this by conducting real-world studies: “Although there is an exponential growth of the publications using AI to develop diagnostic, prognostic and predictive models, most of them are based on small cohorts from single centres,” states Fuensanta Bellvís.

“We want to champion an approach to research which ensures more extensive and representative datasets, promotes data harmonisation and emphasises the importance of the interpretability and explainability of results, addressing concerns related to the "black box" nature of AI algorithms.


Fuensanta Bellvís, VP of Clinical Studies at Quibim

Fuensanta Bellvís, VP of Clinical Studies at Quibim


Building the foundations of reputability 

With many still tentative about employing AI into their business, Fuensanta acknowledges the importance of building trust in AI within the industry: “We want to develop more rigorous model validation, preferably on external and independent groups, with a focus on prospectively collected data within RWE studies.

This not only ensures the robustness and generalisability of our AI models but also contributes to building trust within the pharmaceutical and medical communities, which is absolutely key to this emerging technology being used at scale.”

AI, though a relatively novel concept in pharma, only continues to grow in prevalence. To make the software the best it can be in terms of reliability and validity, more data will have to be made available to enhance its knowledge. “Despite 10M daily imaging exams worldwide, less than 0.1% are used for clinical research or creating AI models.” Stresses Ms Bellvís. With an increase of data available for training AI, there is a tangible chance to improve the capability and benefits of AI technology for all in the industry.

You may also like