Istanbul Technical University · Biomedical Imaging and AI Laboratory
BIAI Lab works on low-dose CT reconstruction, image quality assessment, self-supervised learning, and open research infrastructure. This site stays on the public side of the lab: released papers, technical posts, and live tools.
Public papers and preprints represented on this site
Research tools and lab platforms highlighted publicly
Internship participants in the current training pipeline
Publication window covered by the current paper list
What is public here
Released papers, technical posts, tooling, and high-level lab context.
How to navigate
Start with papers, then posts for method context, then tools for live systems.
Entry points
Journal articles, conference work, and public preprints from the lab.
Technical notes and public-facing articles that explain methods, systems, and context.
Live services and open repositories, including the lab visualization and IQA platforms.
The public GitHub organization where repositories, collaboration history, and code live.
Research program
The public material clusters around reconstruction, self-supervision, quality assessment, and research infrastructure.
Reconstructing diagnostically useful CT images while reducing radiation exposure remains a core lab problem.
Much of the lab's work is motivated by the reality that paired clinical targets are rare or unavailable.
The lab studies how to score and explain reconstruction quality, from classical metrics to multimodal models.
Reusable software is treated as part of the research output, not just internal scaffolding.
Papers
Peer-reviewed work and public preprints, ordered before everything else as requested.
A public preprint on prompt-conditioned multimodal quality prediction built on MedSigLIP.
A public preprint that unrolls BM3D-style denoising into a trainable network while preserving the logic of non-local collaborative filtering.
A public preprint that makes classical total-variation regularization spatially adaptive with a learned lambda map.
A public arXiv preprint studying how large multimodal models can score and explain CT image quality.
Posts
This section is reserved for explanatory writing: methods, notes, and public updates that provide context around the work.
April 12, 2026
The lab's LTV paper was presented at IEEE ISBI 2026 in London, marking a first conference-paper milestone for Yusuf Talha Basak.
Read post →January 19, 2026
The lab's paper on learnable total variation for low-dose CT denoising was accepted to IEEE ISBI 2026.
Read post →Reducing dose does not simply make CT images noisier. It changes the reconstruction problem itself, which is why the lab keeps working on data-efficient and self-supervised methods.
Read post →Tools
The public tool surface starts with the live CT visualization and IQA services, then expands into reusable repositories across reconstruction and infrastructure.
Some tool repositories are still private while documentation, packaging, and release review are being prepared. Live services are linked first; repository links appear only after public release.
The public tooling surface starts with the browser-based visualization and IQA systems. Additional repositories will be linked here when each codebase is ready for public release.
Interactive CT analysis in the browser with line profiles, ROI metrics, histogram comparison, NPS, zoom, and diff maps.
Public repository release is still being prepared.
Web-based image quality assessment workflows for reader studies, annotation, and score collection.
Public repository release is still being prepared.
Differentiable CT reconstruction operators in PyTorch for learning-based imaging pipelines.
Repository is not public yet while packaging and documentation are being finalized.
C++ reconstruction framework for digital breast tomosynthesis with classical algorithms and geometry tooling.
Repository is not public yet while packaging and documentation are being finalized.
Lab snapshot
The site focuses on public work, but the lab context behind it is visible here at a high level.
Principal investigator
Principal investigator
Leads biomedical imaging research, academic direction, and external collaborations.
