SutureFormer: Learning Surgical Trajectories via Goal-conditioned Offline RL in Pixel Space

TL;DR

SutureFormer is a goal-conditioned offline reinforcement learning framework that predicts surgical needle trajectories from endoscopic videos by modeling sequential pixel-wise motion, improving accuracy over existing methods.

Contribution

It introduces a novel formulation of needle trajectory prediction as a sequential decision-making problem using offline RL with dense rewards from sparse annotations.

Findings

Reduces Average Displacement Error by 58.6% compared to baselines.

Effectively models continuous, physically plausible needle motion in pixel space.

Demonstrates superior performance on a new kidney wound suturing dataset.

Abstract

Predicting surgical needle trajectories from endoscopic video is critical for robot-assisted suturing, enabling anticipatory planning, real-time guidance, and safer motion execution. Existing methods that directly learn motion distributions from visual observations tend to overlook the sequential dependency among adjacent motion steps. Moreover, sparse waypoint annotations often fail to provide sufficient supervision, further increasing the difficulty of supervised or imitation learning methods. To address these challenges, we formulate image-based needle trajectory prediction as a sequential decision-making problem, in which the needle tip is treated as an agent that moves step by step in pixel space. This formulation naturally captures the continuity of needle motion and enables the explicit modeling of physically plausible pixel-wise state transitions over time. From this perspective, we propose SutureFormer, a goal-conditioned offline reinforcement learning framework that leverages sparse annotations to dense reward signals via cubic spline interpolation, encouraging the policy to exploit limited expert guidance while exploring plausible future motion paths. SutureFormer encodes variable-length clips using an observation encoder to capture both local spatial cues and long-range temporal dynamics, and autoregressively predicts future waypoints through actions composed of discrete directions and continuous magnitudes. To enable stable offline policy optimization from expert demonstrations, we adopt Conservative Q-Learning with Behavioral Cloning regularization. Experiments on a new kidney wound suturing dataset containing 1,158 trajectories from 50 patients show that SutureFormer reduces Average Displacement Error by 58.6% compared with the strongest baseline, demonstrating the effectiveness of modeling needle trajectory prediction as pixel-level sequential action learning.