Modeling and Workflow Recognition for the Anterior Approach in Total Hip Arthroplasty

BACKGROUND AND PROJECT RATIONALE

In Total Hip Arthroplasty (THA), a degenerated hip joint is replaced with an artificial acetabular and femoral component. In clinical practice, several approaches to accessing bone anatomy are employed. The direct anterior approach (DAA) has been adopted widely in recent years due to smaller incisions, fewer complications, faster recovery, and improved patient outcomes. However, the reduced space and limited access may cause more complexity in the execution of surgical steps.

Primary objective

The first goal of this project is to create a systematic and general description of the surgical process for DAA in THA. For this purpose, surgical process models (SPMs) provide a basis to manage, organize, and optimize the surgical process. For this purpose, a labeled dataset of THA surgery videos will be segmented into individual surgical steps and sub-steps using a systematic approach. SPMs are a simplified network of surgical or surgery-related activities and their relationships. These models can be used to compare several interventions, surgeons, or whole Operating Room teams. In addition, the SPMs can be used as input for workflow management systems.

• Recording of total hip arthroplasty (Other): We record total hip arthroplasties (THA) of hip surgeons at Balgrist University Hospital (BUH), Kantonsspital Baden (KSB), Kantonsspital Winterthur (KSW). Video \& audio data are stored locally at Balgrist-managed servers (patient-sensitive information removed). Data includes: Demographics (year of birth, sex, smoking status, body mass index) Diagnosis Surgery-specific data \& other surgical procedures during recording Standard preoperative Computed tomography of the hip/pelvis Standard pelvis x-rays pre- \& postoperative Standard axial hip x-rays pre- \& postoperative Western Ontario \& McMaster Universities Osteoarthritis Index \& Harris-Hip-Score To ensure the generated Surgical process model(SPM), evaluated using the recordings obtained from surgeries performed at BUH, generally describes the surgical process of THA, we apply our model to surgeries from the KSB and the KSW. There, 3 THAs are recorded and analyzed regarding commonalities and discrepancies in the SPM.

Primary outcomes

• labeled recordings of total hip arthroplasty [Up to 1 year]: Surgery recordings of Total Hip Arthroplasties (THA) with Direct Anterior Approach (DAA) performed at Balgrist University Hospital will be used to develop an in-depth systematic analysis of surgical workflows and activity. The comparison and evaluation of THA with DAA performed at two external hospitals (Kantonsspital Baden, Switzerland and Kantonsspital Winterthur, Switzerland) will allow to generalize and evaluate the developed models: Multicenter study.

Other outcomes

• automated deep learning-based approach for surgical workflow and activity recognition for total hip arthroplasty (THA) [Up to 1 year]: With surgical process modeling (SPM) as the basis, this study aims to develop an automated deep learning-based approach for surgical workflow and activity recognition for total hip arthroplasty (THA), which has not been proposed in previous work. The labels created during the creation of the Surgical Process Model will be used to train a state-of-the-art deep neural network for surgical phase recognition. The surgical process model and the trained deep learning model can be used in future work for the in-depth analysis of surgical processes, surgical training, and skill assessment. Furthermore, automated surgical workflow recognition systems can provide vital input to physicians in the form of early warnings in cases of deviations and anomalies and context-aware decision support (Padoy, 2019), as well as the automatic extraction of a surgery's protocol, which is crucial for archiving, educational and post-operative patient-monitoring purposes (Zisimopoulos, et al., 2018).

Sex: All, ages: 18 Years to any

Inclusion Criteria:

* Primary surgery for total hip arthroplasty (THA) at Balgrist University Hospital, Kantonsspital Baden and Kantonsspital Winterthur
* Complete radiological data (CT and Hip/Pelvis x-ray)
* Signed consent

Exclusion Criteria:

* Missing or incomplete consent
* Previous surgery of the hip at the site of the THA
* Fracture of the hip or pelvis at the site of the THA

• Carrillo F, Esfandiari H, Muller S, von Atzigen M, Massalimova A, Suter D, Laux CJ, Spirig JM, Farshad M, Furnstahl P. Surgical Process Modeling for Open Spinal Surgeries. Front Surg. 2022 Jan 25;8:776945. doi: 10.3389/fsurg.2021.776945. eCollection 2021. PMID 35145990
• Charney M, Paxton EW, Stradiotto R, Lee JJ, Hinman AD, Sheth DS, Prentice HA. A Comparison of Risk of Dislocation and Cause-Specific Revision Between Direct Anterior and Posterior Approach Following Elective Cementless Total Hip Arthroplasty. J Arthroplasty. 2020 Jun;35(6):1651-1657. doi: 10.1016/j.arth.2020.01.033. Epub 2020 Jan 22. PMID 32057597
• Haynes JA, Hopper RH Jr, Ho H, McDonald JF 3rd, Parks NL, Hamilton WG. Direct Anterior Approach for Primary Total Hip Arthroplasty Lowers the Risk of Dislocation Compared to the Posterior Approach: A Single Institution Experience. J Arthroplasty. 2022 Mar;37(3):495-500. doi: 10.1016/j.arth.2021.11.011. Epub 2021 Nov 10. PMID 34774686
• Neumuth T, Jannin P, Strauss G, Meixensberger J, Burgert O. Validation of knowledge acquisition for surgical process models. J Am Med Inform Assoc. 2009 Jan-Feb;16(1):72-80. doi: 10.1197/jamia.M2748. Epub 2008 Oct 24. PMID 18952942
• Patel NN, Shah JA, Erens GA. Current Trends in Clinical Practice for the Direct Anterior Approach Total Hip Arthroplasty. J Arthroplasty. 2019 Sep;34(9):1987-1993.e3. doi: 10.1016/j.arth.2019.04.025. Epub 2019 Apr 18. PMID 31076194
• Seibold, M., Hoch, A., Farshad, M., Navab, N., & Fürnstahl, P. (2022). Conditional generative data augmentation for clinical audio datasets. Medical Image Computing and Computer Assisted Intervention (MICCAI). Singapore.
• Zisimopoulos, O., Flouty, E., Luengo, I., Giataganas, P., Nehme, J., Chow, A., & Stoyanov, D. (2018). DeepPhase: Surgical Phase Recognition in CATARACTS Videos. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial, 11073, 265-275.

• Philipp Fürnstahl, PhD, Principal investigator, Balgrist University Hospital
• Patrick Zingg, MD, Study chair, Balgrist University Hospital