Publications by authors named "Thuy Nuong Tran"
Article Synopsis
- - Colorectal cancer (CRC) is a leading cause of death globally, and early detection of polyps is crucial for reducing mortality and improving diagnostic efficiency.
- - This study introduces a complete validation framework and evaluates various techniques for detecting, segmenting, and classifying polyps, finding that most methods perform well in detection and segmentation but struggle with classification.
- - The research emphasizes the need for further advancements in polyp classification to support clinicians effectively during procedures, proposing a standardized method to assess and compare different approaches in the field.
Article Synopsis
- - Formalizing surgical activities as triplets of instruments, actions, and target anatomies helps enhance the understanding of tool-tissue interactions, improving AI assistance in image-guided surgeries.
- - The CholecTriplet2022 challenge expands the previous work by adding weakly-supervised localization of surgical tools and modeling their activities as ‹instrument, verb, target› triplets.
- - The paper outlines a baseline method and presents 10 new deep learning algorithms, while also comparing their effectiveness and analyzing results to provide insights for future surgical research.
Article Synopsis
- Image-based tracking of medical instruments is crucial for enhancing surgical data science, but existing methods struggle with difficult images and lack generalizability.
- The Heidelberg Colorectal (HeiCo) dataset is introduced as the first publicly available resource for testing detection and segmentation algorithms, focusing on robustness and adaptability.
- This dataset features 30 laparoscopic videos, sensor data, and detailed annotations for over 10,000 frames, aiding in organizing global competitions like the Endoscopic Vision Challenges.
Intraoperative tracking of laparoscopic instruments is often a prerequisite for computer and robotic-assisted interventions. While numerous methods for detecting, segmenting and tracking of medical instruments based on endoscopic video images have been proposed in the literature, key limitations remain to be addressed: Firstly, robustness, that is, the reliable performance of state-of-the-art methods when run on challenging images (e.g.
View Article and Find Full Text PDF