TTDF: A Two-Stage Framework for Reliable Surgical Phase Transition Detection

Yushi Guo; Pietro Valdastri; Duygu Sarikaya

cs.CVSep 13, 2026

TTDF: A Two-Stage Framework for Reliable Surgical Phase Transition Detection

Authors: Yushi Guo, Pietro Valdastri, Duygu Sarikaya

Organizations: School of Computer Science, University of Leeds, Leeds, UK · STORM Lab UK · School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK

Abstract

Reliable workflow transition detection is important for context-aware surgical assistance and downstream decision support. However, online surgical phase recognizers primarily focus on frame-wise accuracy and temporal consistency, rather than the reliability of workflow transition events. Directly converting phase changes into events is unreliable: temporal jitter and workflow-illegal switches produce false or duplicate events, while persistent, workflow-consistent candidates may remain incorrect. To address this limitation, we formulate reliable workflow transition detection as a distinct event-level task operating on outputs of a frozen online phase recognizer. We propose the Two-Stage Transition Detection Framework (TTDF), a causal framework that progressively filters transition candidates. Transition Candidate Extraction (TCE) first applies a minimum-duration requirement and a workflow-graph constraint to remove false candidates caused by temporal jitter and phase transitions not allowed by the workflow graph. Specifically, a candidate is retained only if the predicted target phase persists for a minimum duration and the ordered phase pair belongs to the workflow graph's allowed transition set. TCE thereby produces a high-recall candidate set without additional training. Transition Candidate Verification (TCV) suppresses remaining false candidates using phase-posterior shifts and visual-change cues from frozen DINOv2 features. Events are assessed using a phase-pair-aware one-to-one matching protocol. Experiments on Cholec80 show that TTDF reduces false transition emissions while preserving recall and controlling decision delay.

Figures & tables

Figure 1: Overview of the proposed Two-Stage Transition Detection Framework (TTDF). Given the online phase predictions, the Transition Candidate Extraction (TCE) module first removes false candidates caused by temporal jitter and phase transitions not allowed by the workflow graph, thereby constructing a high-recall transition candidate set. The Transition Candidate Verification (TCV) module then verifies each candidate using candidate-centered temporal representations and local visual-change cues, producing the final transition decisions.

Figure 2: Detailed illustration of the Transition Candidate Verification (TCV) module. Posterior and visual-change cues from a causal candidate-centered window are concatenated and processed by a lightweight TCN with mean pooling for binary verification.

Method	Prec. $\uparrow$	Rec. $\uparrow$	F1 $\uparrow$	False/GT $\downarrow$	Dup/GT $\downarrow$	Delay 50 $\downarrow$
Raw changes	0.093	0.672	0.163	6.570	0.332	0.0
+ Dwell	0.209	0.643	0.316	2.430	0.072	5.0
+ Legality (TCE)	0.321	0.643	0.428	1.357	0.072	5.0
+ TCV (TTDF)	0.470	0.531	0.497	0.603	0.013	6.3

Table 1: Progressive event-level transition detection on Cholec80. TCE denotes Stage 1 with dwell persistence and workflow legality, while TTDF additionally includes Stage 2 TCV. Delay 50 is the median online decision delay.

Cue	Emitted Events $\downarrow$	F1 $\uparrow$	False/GT $\downarrow$	Dup/GT $\downarrow$
Posterior $\ell_{1}$	245.6 $\pm$ 61.0	0.475 $\pm$ 0.025	0.557 $\pm$ 0.173	0.013
Posterior JS	287.2 $\pm$ 76.4	0.484 $\pm$ 0.017	0.684 $\pm$ 0.245	0.020 $\pm$ 0.015
Posterior $\ell_{1}$ +JS	244.2 $\pm$ 18.0	0.479 $\pm$ 0.012	0.551 $\pm$ 0.047	0.013
Visual	468.0 $\pm$ 4.5	0.426 $\pm$ 0.005	1.354 $\pm$ 0.008	0.071 $\pm$ 0.002
Posterior + visual	266.6 $\pm$ 30.9	0.497 $\pm$ 0.006	0.603 $\pm$ 0.094	0.013

Table 2: Ablation of candidate-centered cues in TCV on Cholec80. Posterior denotes the combined

\ell_{1}

and JS posterior cues. Emitted Events indicates the number of Stage-1 candidates retained by TCV. Lower is better for Emitted Events, False/GT, and Dup/GT; higher is better for F1.

Surgical Video Understanding

Surgical video understanding uses computer vision and multimodal AI to interpret procedures, instruments, actions, and events over time.

See all 118 papers

Shifted Window Temporal Transformer for YOLO-Guided Phase Recognition in Laparoscopic Cholecystectomy Surgery
Oct 26, 2026Sheekar Banerjee, Md. Shahabuddin, Md. Mahbub Hasan Akash +2Surgical Video Understanding Laparoscopic Surgery
Department of Computer Science and Engineering, IUBAT—International University of Business Agriculture and Technology, 4 Embankment Drive Road, Sector-10, Uttara Model Town, Dhaka, 1230, Bangladesh. · Boston Children’s Hospital, Harvard Medical School, 401 Park Dr, Boston, USA.
HTT-Net: Hierarchical Text-guided Transition Modeling for Surgical Video Phase Recognition
Jul 18, 2026Kunjie Deng, Jinghui Zhang, Weidong Chen +4Surgical Video Understanding Temporal Video Understanding
School of Information Science and Technology, University of Science and Technology of China, Hefei, China · Department of General Surgery, Beijing Friendship Hospital, Capital Medical University, Beijing 100050, China; State Key Lab of Digestive Health, Beijing 100050, China; and National Clinical Research Center for Digestive Diseases, Beijing 100050, China · Senior Department of Urology, Chinese PLA General Hospital, Beijing 100039, China
Stabilizing Temporal Inference Dynamics for Online Surgical Phase Recognition
May 11, 2026Yang Liu, Ning Zhu, Jingjing Peng +4Surgical Video Understanding Temporal Consistency
King’s College London, London, UK · University of Electronic Science and Technology of China, Chengdu, China · Mach Drive

Oct 26, 2026J/K move · Enter open · S save

Shifted Window Temporal Transformer for YOLO-Guided Phase Recognition in Laparoscopic Cholecystectomy Surgery

Sheekar Banerjee, Md. Shahabuddin, Md. Mahbub Hasan Akash, Nabilah+1

Department of Computer Science and Engineering, IUBAT—International University of Business Agriculture and Technology, 4 Embankment Drive Road, Sector-10, Uttara Model Town, Dhaka, 1230, Bangladesh. · Boston Children’s Hospital, Harvard Medical School, 401 Park Dr, Boston, USA.

Surgical Video Understanding Laparoscopic Surgery Spatiotemporal Transformers

Open paper AskPDF arXiv

Intelligent operating rooms require accurate awareness of surgical workflow for better intra-operative decisions, teamwork, and surgical documentation. Therefore, accurately recognizing surgical phase during surgery is crucial. However, previous artificial intelligence-based methods provide fluctuating predictions for the transitions, restricting their clinical use. To address this challenge, we propose a clinically meaningful framework that integrates a Swin Transformer and Temporal Transformer Encoder (TTE) for effective surgical phase recognition and a parallel YOLOv11s network for real-time surgical instrument localization. To avoid phase flickering and to produce stable workflow predictions, a causal temporal smoothing is applied. The proposed framework is tested on the Cholec80 dataset for the simultaneous recognition of the surgical phases and instruments. The overall phase recognition accuracy of this model was 92%, and the macro average F1-score was 0.86. This performance surpasses CNN based methods such as EndoNet (81.7%) and SV-RCNet (85.3%), as well as temporal convolutional and transformer-based approaches including TeCNO (88.6%), Trans-SVNet (90.9%), and LoViT (91.5%). Besides, the module of instrument detection achieved [email protected] = 0.863. The reliability of the results is confirmed, obtaining a high recognition rate of the workflows, and a high accuracy of the localization of the instruments. The proposed framework can serve as a foundation for intelligent surgical assistance, enriching situation awareness, and potential robotic surgery applications. Code will be made publicly available at: https://github.com/ac005sheekar/LapChol-Surgery-Swin-Cholec80

Abstract

Surgical Video Understanding

Explore similar work

Shifted Window Temporal Transformer for YOLO-Guided Phase Recognition in Laparoscopic Cholecystectomy Surgery

HTT-Net: Hierarchical Text-guided Transition Modeling for Surgical Video Phase Recognition

Stabilizing Temporal Inference Dynamics for Online Surgical Phase Recognition