The research group focuses on researching and developing techniques for extracting information from manuals and textual content in the field of robotic surgery. Of particular interest to the group is the extraction of procedural knowledge from descriptions of robotic surgery procedures found in domain manuals and textbooks. The extracted procedural knowledge represents valuable content that can be exploited in many clinical applications: automation of surgical procedures by autonomous robots; construction/extension of surgical knowledge bases to support decision-making during surgical interventions; creation of question-answering style applications, usable for example during the learning phase by medical students.
The group, which involves the University of Verona (with three departments: Computer Science, Engineering for Innovation Medicine, and Foreign Languages and Literatures) and the University of Mannheim (Germany), publishes research results in top international journals and conferences in the field, and makes available to the research community the developed resources (e.g., datasets, pre-trained models), such as:
- Surgical Procedural Knowledge Sentences (SPKS) dataset: https://gitlab.com/altairLab/spks-dataset
- Robotic Surgical Propositional Bank (RSPB): https://gitlab.com/altairLab/robotic-surgery-propositional-bank
- SurgicBERTa: https://huggingface.co/marcobombieri/surgicberta
Main publications:
- Marco Bombieri, Marco Rospocher, Diego Dall'Alba and Paolo Fiorini. Automatic detection of procedural knowledge in robotic-assisted surgical texts. International Journal of Computer Assisted Radiology and Surgery, vol. 16, pp. 1287–1295. ISBN: 1861-6429 (2021)
- Marco Bombieri, Daniele Meli, Diego Dall'Alba, Marco Rospocher and Paolo Fiorini. Mapping natural language procedures descriptions to linear temporal logic templates: an application in the surgical robotic domain. Applied Intelligence, vol. 53, no. 22, pp. 26351–26363. ISBN: 1573-7497 (2023)
- Marco Bombieri, Marco Rospocher, Simone Paolo Ponzetto and Paolo Fiorini. Machine understanding surgical actions from intervention procedure textbooks. Computers in Biology and Medicine, vol. 152, pp. 106415. ISSN: 0010-4825 (2023)
- Marco Bombieri, Marco Rospocher, Simone Paolo Ponzetto and Paolo Fiorini. SurgicBERTa: A pre-trained language model for procedural surgical language. International Journal of Data Science and Analytics, vol. 18, no. 1, pp. 69–81 (2024)
- Marco Bombieri, Marco Rospocher, Simone Paolo Ponzetto and Paolo Fiorini. The Robotic-Surgery Propositional Bank. Language Resources and Evaluation, vol. 58, no. 3, pp. 1043–1071 (2024)
University of Verona:
- Marco Bombieri (dottorando, Dip. Informatica)
- Paolo Fiorini (Dip. Ingegneria per la medicina di innovazione)
- Marco Rospocher (Dip. Lingue e Letterature Straniere)
University of Mannheim:
