|
|
|
Abstracts1) Invited Talks : Pr. Emmanuel Vanderpoorten Title: Surgical Robotics, Getting into Shape
Abstract: Minimally invasive surgery continues to drive the evolution of surgical robotics. The promise of faster recovery, smaller scars, and improved ergonomics is encouraging clinicians to integrate these technologies into routine practice. Beyond established indications such as prostatectomy and cholecystectomy, emerging techniques aim to further minimize access size and incision count. As instruments become smaller and more flexible, maintaining precise awareness of their configuration and position becomes increasingly critical for safe and intuitive operation. This talk highlights current applications and recent advances in sensing and control of instrument shape, enabling safer manipulation and supporting the ongoing trend toward ever smaller surgical access.
Pr. Nicolas Andreff (FEMTO-ST) Title: Minimal Impact Surgical Robotics Abstract: Miniaturisation is a historical occupation in Franche-Comté. This lead to the creation of a research activity in surgical microrobotics. A selection of results obtained at the FEMTO-ST institute over the last 15 years will be presented, targeting minimising the negative clinical impact of surgery. This will be followed by a prospective over the minimisation of the negative environmental impact of surgical robotics. Carole Lartizien, DR CNRS Laboratoire CREATIS - Univ Lyon, CNRS, Inserm, INSA Lyon, UCBL, CREATIS, UMR5220, U1294, F-69621, Villeurbanne, France Title: Fully and weakly supervised modeling of multiparametric MRI for automatic detection, segmentation and aggressivity characterization of prostate cancer. Abstract: Prostate cancer (PCa) is the most frequently diagnosed cancer in men in more than half the countries of the world. Multiparametric magnetic resonance imaging (mp-MRI) has shown excellent results in the detection of high-grade tumors and is now recommended by the European Association of Urology guidelines, to be performed prior to any biopsy in cases of clinical suspicion of PCa. However, characterizing focal prostate lesions in mp-MRI sequences is time demanding and challenging, even for experienced readers. Despite the use of the diagnostic criteria such as PI-RADS v2 scoring system, its inter-reader reproducibility remains moderate at best. This talk will focus on the research projects I have conducted over the past few years to develop deep learning models that automatically detect, segment PCa as well as characterize their aggressiveness. I will discuss fully supervised models as well as innovative weakly supervised architectures that have proven effective on private and public datasets.
2) PIB CAMI researchers Chenji Li Title: sEMG-Based Motion Recognition for Robotic Surgery Training Using Machine Learning and Variable-Length Sliding Windows abstract : Tbd Nassib Abdallah Title: Intraoperative speech analysis during Awake Brain Surgery : Towards on automatic deep-learning based approach Abstract: Awake brain surgery requires continuous monitoring of speech to preserve functional language areas during tumor resection. Various modalities such as EEG, EMG, fMRI, and audio can be used for monitoring; however, speech remains the least invasive and most interpretable signal, as detected impairments can be clinically verified with speech therapists.
The modeling of intraoperative audio presents major challenges related to noise, speaker overlap, and inter-patient variability. Artificial intelligence enables the extraction and classification of relevant acoustic patterns from clinical recordings, supporting objective assessment of speech impairments. Preliminary experiments conducted within the IMAGINE team at LaTIM, U1101, CHU Brest, using the Databrase corpus and the open-access TORGO dataset (recorded outside the surgical context but focused on speech impairment detection), demonstrate the feasibility of automatic classification of intraoperative speech disorders using deep learning models. Elijah Van Houten
|
