TIA Centre: News

29^th UK Conference on Medical Image Understanding and Analysis, Leeds (15^th - 17^th July 2025)

This July, I was fortunate to attend my first in-person academic conference: the 29th UK Conference on Medical Image Understanding and Analysis (MIUA 2025) in Leeds, which brings together an international community focused on advancing the use of AI in medical imaging. As a PhD student in this area, I’d previously experienced conferences only through my screen. Participating virtually in ICLR 2022 felt like a very different experience due to the pandemic. The unfamiliarity of Leeds added a sense of adventure; some last-minute research into the city’s history and sights piqued my excitement for both the place and the gathering.

The journey began bleary-eyed with an early morning departure something I, like many academics, would gladly avoid if possible. Luckily, I was able to catch a ride, which not only spared me the long drive but was a good bonding opportunity with colleagues and gave me time to mentally prepare. In hindsight, a pre-conference night in Leeds would have made the day smoother. For future newcomers: plan to arrive early to save energy for the busy days ahead.

The conference wasted no time plunging us into a full schedule. The first day opened with a series of compelling talks spanning diverse data modalities. It was great to observe common themes across different fields while also gaining insight into the unique challenges each specialty encounters. That morning, I somehow ended up not only front and centre in the group photo, but, more importantly, first in the queue for lunch!

Heading into the computational pathology session, my focus was squarely on my own presentation, I felt that blend of nerves and anticipation familiar to any first timer. Most of my slides were repurposed from earlier lab meetings, and my rehearsal time had been limited but I was confident I knew the content having looked at it nearly every day for 2.5 years! My talk, which cantered on segmenting endometrial tissue, was scheduled just after the tea break, leaving me too distracted to concentrate on much else beforehand. Taking the stage, I was acutely aware of the size of the audience; it was by far the largest I’d addressed. Those few minutes on stage passed quickly, but being asked two insightful questions and some engaging discussion afterward was a validating sign that people were interested in the work, and I had done it justice presenting it at a level people could understand.

There was a sense of relief once my talk was done. I’ve received recognition at smaller events like the Warwick Postgraduate Colloquium on Computer Science (WPCCS), presenting at MIUA was on another level. It was heartening to see a strong turnout from TIA, with six presentations in our session, and the whole event was expertly chaired by our own Adam, Mostafa, and Neda, adding a comforting sense of familiarity with our own lab meetings.

After the session, prizes were awarded for best paper, with winners determined by both reviewer feedback and an audience vote. To my surprise, I won the best paper runner up! A moment that felt surreal after simply hoping to get through my talk without any mishaps. Being recognized by both the reviewers and the audience was humbling and is something I can take great confidence from. Congratulations are also in order for our own Jiaqi, who took the top spot with a truly impressive contribution. As the celebrations wrapped up, we were treated to a sudden downpour, a classic English summer welcome, but not even the rain could dampen my spirits.

Day two kept up the momentum: a packed program of talks and posters, covering a wide range of topics ranging from novel methods in dentistry to advances in X-ray imaging, drawing contributors from across the globe. Bashayer from the TIA showcased an excellent poster, and we all enjoyed a thought-provoking keynote. The day ended with a lively panel discussion involving clinicians and academics on the real-world deployment of AI in healthcare. Hearing a clinical perspective is always enjoyable, and it emphasises how AI research can (and must) impact real-world healthcare. This reinforces why the work we do matters beyond the academic bubble.

That evening, the conference dinner took place at Leeds Armory, one of the world’s oldest museums, dating back to 1323. The venue was superb, with the drinks reception actually held among the exhibits. This gave us a unique chance to browse history’s artifacts and sparked some interesting conversations. There was also some shared amusement at the irony of spending the day focused on improving health, only to spend the evening surrounded by instruments for the opposite.

The final day brought more talks, the closing prize ceremony, and an excellent take away lunch. After three non-stop days, I left feeling energised about my research and better connected to the field. Walking back to the car, we noticed a poignant quote from Rob Burrow that seemed to sum up not just MIUA 2025 but the entire journey of a PhD student in medical research:

“In a world full of adversity, we must dare to dream.”

Contributions from the TIA Centre:

Ethar Alzaid [Oral Presentation]: Transductive Survival Ranking for Pan-cancer Automatic Risk Stratification using Whole Slide Images.

Fayyaz Minhas [Oral Presentation]: Learnable Moran’s Index for Modeling Spatial Autocorrelation in Whole Slide Images to Predict Breast Cancer Outcomes.

George Wright [Oral Presentation]: ASTER: Automated Segmentation of Endometrial Histology Images for Reproductive Health Assessment.

Esha Nasir [Oral Presentation]: SWIFT-Reg: Slide-Wide Intelligent Feature-based Tissue Registration.

Esha Nasir [Oral Presentation]: MitoNet: Efficient Ki-67 Detection in H&E-Stained Images.

Jiaqi Lv [Oral Presentation]: Leveraging Pathology Foundation Models for Panoptic Segmentation of Melanoma in H&E Images.

Bashayer Abdallah [Poster Presentation]: DepthClassNet: A Multitask Framework for Monocular Depth Estimation and Texture Classification in Endoscopic Imaging.

By George Wright