AutoSNAP: Automatically Learning Neural Architectures for Instrument Pose Estimation

TL;DR

AutoSNAP is an automatic framework that learns neural architectures specifically for instrument pose estimation in computer-assisted interventions, outperforming existing methods by tailoring architectures to the task.

Contribution

It introduces a novel architecture search method using Symbolic Neural Architecture Patterns (SNAPs) tailored for pose estimation in CAI, and demonstrates superior performance over existing architectures.

Findings

AutoSNAP discovers architectures that outperform hand-engineered models.

SNAPNet surpasses i3PosNet and DARTS in accuracy.

The framework enables efficient architecture optimization for CAI tasks.

Abstract

Despite recent successes, the advances in Deep Learning have not yet been fully translated to Computer Assisted Intervention (CAI) problems such as pose estimation of surgical instruments. Currently, neural architectures for classification and segmentation tasks are adopted ignoring significant discrepancies between CAI and these tasks. We propose an automatic framework (AutoSNAP) for instrument pose estimation problems, which discovers and learns the architectures for neural networks. We introduce 1)~an efficient testing environment for pose estimation, 2)~a powerful architecture representation based on novel Symbolic Neural Architecture Patterns (SNAPs), and 3)~an optimization of the architecture using an efficient search scheme. Using AutoSNAP, we discover an improved architecture (SNAPNet) which outperforms both the hand-engineered i3PosNet and the state-of-the-art architecture search method DARTS.