Diana Spencer speaks to Dr Muhunthan Thillai, CEO at Qureight, about the company’s AI-powered imaging analytics and the technology’s potential in lung diseases and beyond. 

DS: Can you provide some background to the company?

MT: We are a core imaging laboratory using deep-learning and analytics to advance clinical trials and we are based in Cambridge UK. I co-founded the company in 2018, with a mission to advance the understanding of lung and heart diseases and transform clinical trials using innovative technology. We specialise in AI-powered data curation and advanced imaging analytics to accelerate drug development, and clinical trial design. We deploy AI and deep learning to analyse anonymised clinical data, thereby facilitating the discovery of new biomarkers and endpoints for clinical studies. We work collaboratively with pharmaceutical companies and academic partners across the globe to identify novel endpoints and enhance the efficiency of clinical trials, particularly for areas such as interstitial lung diseases (ILDs) and pulmonary hypertension (PH). 

DS: Qureight is collaborating with Remedy Cell to validate RC-0315’s mechanism of action in idiopathic pulmonary fibrosis (IPF). What will be Qureight’s contribution to the partnership?

MT: A big part of our work will involve using our deep learning imaging platform to precisely analyse localised sections of the lungs to monitor the treatment’s effects. This will specifically involve using imaging and biomarker analysis in bronchoalveolar lavage (BAL) and plasma samples. 

By integrating our deep-learning technology in the clinical phase study, we will be providing a full spectrum of imaging core lab services, including site qualification, data handling, quality control (QC) and deep learning AI image analysis.  

Part of RC-0315’s distinctiveness as a treatment is its novel application, which works by being instilled directly into a chosen lung segment by bronchoscopy. Once this is done, we will then apply innovative AI modelling techniques to sub-sections of the lungs and look for localised changes. This pioneering approach offers a more detailed, precision-focused quantification of lung segments, which will ultimately unlock new treatment options for IPF patients. 

DS: What is unique about this trial?

MT: The RC-0315 Phase Ib clinical study stands out for its integration of advanced AI imaging, localised drug delivery, and a multi-target regenerative approach, aiming to provide a deeper, more precise understanding of therapeutic effects in IPF and potentially set a new standard for early-phase fibrotic disease trials. 

What makes the trial genuinely different is its use of deep learning imaging technology for a more detailed, in-depth analysis of specific lung areas. Unlike traditional whole-lung assessments, this approach focuses on sub-sections of the lung, enabling monitoring of changes in airways, fibrosis, and blood vessels at a granular level. 

DS: Can you explain how the deep learning technology works?

MT: By moving beyond traditional methods, deep learning provides a more quantitative and objective analysis of lung scans. By processing CT scans, it generates 3D data and classifies lung structures in real time, giving researchers more detailed and reliable insights into lung changes than subjective radiologist readings or variable lung function tests.  

Earlier methods relied on 2D analysis, examining individual pixels for abnormalities. However, breakthroughs in brain imaging research introduced 3D convolutional neural networks, which we’ve successfully adapted to lung imaging, dramatically increasing accuracy. 

DS: What kind of insights can now be discovered using deep learning imaging? How can these insights advance drug discovery for IPF?

MT: Deep learning technology significantly enhances lung scan analysis by offering several key advantages. First, it enables real-time analysis, providing immediate results and scan quantification to clinicians. Second, it utilises 3D modelling, effectively replicating the combined expertise of multiple radiologists to achieve a highly detailed analysis. 

In fact, a recent study published in the American Journal of Respiratory and Critical Care Medicine (AJRCCM) demonstrated that 3D models achieved a DICE score exceeding 85%, indicating they are over 85% as accurate as a panel of highly trained radiologists. This combination of speed and accuracy is a crucial advantage in advancing lung scan analysis. 

DS: You are also collaborating with Avalyn Pharmaceuticals on PPF studies. What will be Qureight’s role in these trials?

MT: Our work with Avalyn Pharma forms part of the ongoing MIST study, a Phase IIb clinical trial evaluating the safety and efficacy of AP01 (inhaled pirfenidone) in patients with progressive pulmonary fibrosis (PPF). This collaboration marks the first time deep-learning imaging technologies have been used in a PPF clinical study. 

Deep learning techniques will be used to assess a MIST study clinical outcome measure linking imaging more closely to clinical endpoints. This is especially important for PPF, where diagnosis, monitoring and clinical development pathways can be complex and the data fragmented.  

This will allow us to measure changes in fibrosis volume more sensitively while also enabling us to quantify changes across multiple compartments of the lung that we believe are also linked with disease progression. We expect to release additional promising data that analyses using Qureight technology at future medical conferences, so watch this space. 

DS: Can the technology be applied to other diseases and other organs beyond the lungs? 

MT: Deep imaging technology is adaptable for a wide range of clinical trial data and imaging modalities. The use of AI-powered data curation and analytics is certainly not limited to specific issues like lung tissue. Integrating imaging, physiological, and biomarker data, is applicable to other organ systems, and this type of techbio platform approach can pave the way for early detection and improved intervention strategies. We have ambitions to extend the use of the technology to new diseases such as lung cancer, and potentially beyond heart and lung diseases. 

DS: How do you foresee AI/deep learning imaging improving drug discovery in the future?

MT: Beyond reducing time and costs, AI can create synthetic, or ‘digital twins,’ of clinical studies, enabling significantly more detailed comparisons between patient groups and their outcomes. For instance, AI can generate synthetic matched placebo groups, allowing us to compare treatment results against a simulated control. While this is still an emerging area, we have already successfully implemented this approach with some of our partners.  

By carefully matching patients using baseline CT scans and fibrosis markers, we’ve created synthetic matched placebo arms, demonstrating the likely outcomes if an imaging-matched control had been used. These advancements represent a significant leap forward in our fight against life-threatening lung conditions like progressive pulmonary fibrosis, making this a truly exciting time for the sector. 

Biography 

Muhunthan is a board certified pulmonologist. He was awarded a PhD in molecular immunology from Imperial College London before being appointed as a consultant physician. He was previously Director of Interstitial Lung Diseases at Royal Papworth and Addenbrooke’s Hospitals Cambridge UK and is currently an Honorary Associate Professor at the University of East Anglia.