<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Azar Kazemi |</title><link>https://albarqouni.github.io/authors/azar-kazemi/</link><atom:link href="https://albarqouni.github.io/authors/azar-kazemi/index.xml" rel="self" type="application/rss+xml"/><description>Azar Kazemi</description><generator>HugoBlox Kit (https://hugoblox.com)</generator><language>en-us</language><image><url>https://albarqouni.github.io/media/authors/azar-kazemi_hu_c589bd99f8950c25.jpg</url><title>Azar Kazemi</title><link>https://albarqouni.github.io/authors/azar-kazemi/</link></image><item><title>MISTAI Computational Pathology Workshop</title><link>https://albarqouni.github.io/community/mistai-workshop2026/</link><pubDate>Thu, 03 Dec 2026 00:00:00 +0000</pubDate><guid>https://albarqouni.github.io/community/mistai-workshop2026/</guid><description>&lt;p&gt;On &lt;strong&gt;3 December 2026&lt;/strong&gt; we host a full-day &lt;strong&gt;Computational Pathology Workshop&lt;/strong&gt; in Bonn as part of
, our cooperation with &lt;strong&gt;Dnipro State Medical University (DSMU)&lt;/strong&gt; in Ukraine, funded by the &lt;strong&gt;Ministry of Culture and Science of North Rhine-Westphalia (MKW NRW)&lt;/strong&gt;. The workshop takes place during the one-week exchange visit of the DSMU team, which also includes a round table on collaboration projects and EU/DFG grant strategy (30 November), the
in Essen (1 December), the Ukraine Science Days at Ruhr University Bochum (2 December) and lab tours in Bonn (4 December).&lt;/p&gt;
&lt;p&gt;It brings together pathologists, clinicians and AI researchers around one question: how can AI and digital pathology keep diagnostics, research and teaching going when physical infrastructure is disrupted? Two keynotes open the day and a third, by
(Indiana University School of Medicine), closes it, on moving from foundation models to multimodal clinical intelligence beyond the slide. The opening keynotes are by
(Helmholtz AI / TUM) on multimodal foundation models for computational pathology, and by
(Roche Diagnostics) on computational co-intelligence in diagnostic pathology. They are followed by talks on the shared Bonn–Dnipro data platform for &lt;strong&gt;muscle-invasive bladder cancer&lt;/strong&gt;, AI-assisted grading of conjunctival melanocytic lesions in QuPath, and innovations for molecular diagnostics. In the afternoon, young researchers from Bonn and Dnipro present their work in &lt;strong&gt;spotlight talks&lt;/strong&gt; — including results from fine-tuning &lt;strong&gt;pathology foundation models&lt;/strong&gt; and &lt;strong&gt;live demos in QuPath&lt;/strong&gt; (MISTAI and INTERACT-CMIL) — followed by the new Computational Pathology teaching module for DSMU.&lt;/p&gt;
&lt;p&gt;The programme below is &lt;strong&gt;tentative&lt;/strong&gt; and will be updated as speakers are confirmed. Participation is free; please
by 23 November 2026. Places are limited and allocated in order of registration; we&amp;rsquo;ll confirm by email. Coffee and a light lunch will be offered to confirmed participants. A certificate of attendance will be issued to all attended participants.&lt;/p&gt;</description></item><item><title>INTERACT-CMIL: Interpretable multi-task grading of conjunctival melanocytic lesions</title><link>https://albarqouni.github.io/projects/interact-cmil/</link><pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate><guid>https://albarqouni.github.io/projects/interact-cmil/</guid><description>&lt;p&gt;&lt;strong&gt;Collaborators:&lt;/strong&gt; &lt;em&gt;University Hospital Tübingen, University Hospital Erlangen&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;This project addresses the grading of conjunctival melanocytic intraepithelial lesions (CMIL), rare precursors of conjunctival melanoma characterized by complex, interdependent histopathological criteria and substantial inter-observer variability.
Together with clinical partners, we curated a multi-center, expert-annotated histopathology dataset comprising 486 conjunctival biopsy image patches labeled along five clinically defined grading axes.
Methodologically, we developed &lt;em&gt;INTERACT-CMIL&lt;/em&gt;, a multi-task deep learning framework modeling inter-criterion dependencies via selective multi-head supervision and an inter-task consistency loss.&lt;/p&gt;
&lt;p&gt;The proposed approach outperformed CNN and pathology foundation-model baselines (macro-F1 up to 85%), producing coherent and interpretable predictions aligned with expert grading standards. Methodological development and evaluation were coordinated within my group in close collaboration with the clinical partners.&lt;/p&gt;
&lt;h3 id="perspective"&gt;Perspective&lt;/h3&gt;
&lt;p&gt;The project established a reproducible computational benchmark for CMIL grading and contributed to the standardization of digital ocular pathology. A related follow-up project on AI-assisted eyelid tumor analysis received a University of Bonn Prototype Grant, supporting translational development and clinical feasibility studies.&lt;/p&gt;</description></item></channel></rss>