Disputation
Integrating domain priors in rule-based and deep learning algorithms for biomedical image analysis
Trina De, PhD student, CASUS Young Investigator Group “Machine Learning for Infection and Disease”
Abstract of the talk// Trina argues with her thesis that domain-specific priors should be encoded explicitly as constraints or inductive biases in both model design and evaluation. In contrast to generic computer-vision pipelines, she developed biomedical image analysis methods that are computationally effective while remaining aligned with biomedical semantics and structural plausibility. The work targets microscopy-driven diagnostics, where modality variability, high-throughput demands, and spatially nested entities create mathematical and algorithmic challenges.
CV// Trina De is a doctoral candidate at Technische Universität Dresden. She completed her bachelor’s degree at St. Xavier’s College (Autonomous) in Kolkata and her master’s degree at the Chennai Mathematical Institute in Chennai (both India). After internships at Morgan Stanley and Fortiate she began her career in industry as a data scientist and postgraduate trainee at Larsen & Toubro Infotech (2020–2022). Since 2022, she has been conducting doctoral research at CASUS, an institzte of Helmholtz-Zentrum Dresden-Rossendorf, applying data science and machine learning methods to scientific and biomedical problems.
Trina will be talking live in Dresden. However, as the event is organized in a hybrid format that includes a videoconferencing tool by Zoom Inc., people not present in Dresden and interested in the topic have the chance to also join the talk by using the following link (meeting ID: 696 9180 6390 / passcode: Jd3C&5#G).
Seminar Room (top floor)
Center for Systems Biology Dresden (CSBD)
Pfotenhauerstraße 108
01307 Dresden
16 March 2026, 10 am