Advancing Arthroscopic Navigation: Evaluating SLAM Algorithms for Enhanced Spatial Orientation

19.03.2025, Abschlussarbeiten, Bachelor- und Masterarbeiten

This project aims to evaluate and benchmark existing Simultaneous Localization and Mapping (SLAM) algorithms for arthroscopic surgery, addressing challenges such as spatial orientation and navigation within the joint space. The study will focus on testing state-of-the-art camera-pose estimation techniques on arthroscopy data, analyzing their performance in terms of accuracy and robustness.

The Human-Centered Computing and Extended Reality Lab of the Professorship for Machine Intelligence in Orthopedics seeks applicants for Bachelor/Master Thesis for the Summer Semester 2025.

Background & Motivation

One of the key challenges in arthroscopic surgery (and endoscopic surgery in general) is maintaining spatial orientation and accurate navigation within the joint space (Desai et al., 2016). This is a particular problem for inexperienced or novice surgeons, often resulting in longer operative times and higher complication rates (Konan et al., 2011). Addressing these challenges through advanced navigation techniques can significantly enhance surgical efficiency and safety.

Student’s Task

A number of approaches have been proposed to enable navigation during endoscopic/arthroscopic procedures (Fu et al., 2021). Camera-pose estimation and tracking has mainly been based on Visual Odometry (VO) and Simultaneous Localization and Mapping (SLAM). Previous approaches include keypoint-detectors (Marmol et al., 2017, Karaoglu,et al., 2024), and SLAM adaptations (Marmol et al., 2019; Ozyoruk, 2021.) For example, Dense-ArthroSLAM (Marmol et al., 2019) is tailored to address the specific challenges of arthroscopy such as featureless homogeneous tissues and complex lighting conditions. The student should test existing SLAM algorithms and evaluate their performance on arthroscopy data.

• Review and analyze state-of-the-art SLAM algorithms for the specific use-case of arthroscopic navigation.
• Implement and test selected algorithms on available arthroscopy datasets.
• Evaluate algorithm performance based on accuracy, robustness, and computational efficiency.

Technical Prerequisites

The student should have some experience in data processing and deep learning. Knowledge in SLAM is beneficial.

Please send your transcript of records, CV and motivation to: Constantin Kleinbeck (constantin.kleinbeck@tum.de) with CC to hex-thesis.ortho@mh.tum.de

Literatur

Literature: Desai, M. J., Mithani, S. K., Lodha, S. J., Richard, M. J., Leversedge, F. J., & Ruch, D. S. (2016). Major peripheral nerve injuries after elbow arthroscopy. Arthroscopy: The Journal of Arthroscopic & Related Surgery, 32(6), 999-1002.
Konan, S., Rhee, S. J., & Haddad, F. S. (2011). Hip arthroscopy: analysis of a single surgeon’s learning experience. JBJS, 93(Supplement_2), 52-56.
Orgiu, A., Karkazan, B., Cannell, S., Dechaumet, L., Bennani, Y., & Gregory, T. (2024). Enhancing wrist arthroscopy: artificial intelligence applications for bone structure recognition using machine learning. Hand Surgery and Rehabilitation, 101717.
Fu, Z., Jin, Z., Zhang, C., He, Z., Zha, Z., Hu, C., ... & Ye, X. (2021). The future of endoscopic navigation: a review of advanced endoscopic vision technology. IEEE Access, 9, 41144-41167.
Marmol, A., Peynot, T., Eriksson, A., Jaiprakash, A., Roberts, J., & Crawford, R. (2017). Evaluation of keypoint detectors and descriptors in arthroscopic images for feature-based matching applications. IEEE Robotics and Automation Letters, 2(4), 2135-2142.
Marmol, A., Banach, A., & Peynot, T. (2019). Dense-ArthroSLAM: Dense intra-articular 3-D reconstruction with robust localization prior for arthroscopy. IEEE Robotics and Automation Letters, 4(2), 918-925.
Karaoglu, M. A., Markova, V., Navab, N., Busam, B., & Ladikos, A. (2023). RIDE: Self-Supervised Learning of Rotation-Equivariant Keypoint Detection and Invariant Description for Endoscopy. arXiv preprint arXiv:2309.09563.
Ozyoruk, K. B., Gokceler, G. I., Bobrow, T. L., Coskun, G., Incetan, K., Almalioglu, Y., ... & Turan, M. (2021). EndoSLAM dataset and an unsupervised monocular visual odometry and depth estimation approach for endoscopic videos. Medical image analysis, 71, 102058.

Kontakt: hex-thesis.ortho@mh.tum.de, hannah.schieber@tum.de