ABSTRACT
PURPOSE: Stereo matching is a crucial technology in the binocular laparoscopic-based surgical navigation systems. In recent years, neural networks have been widely applied to stereo matching and demonstrated outstanding performance. however, this method heavily relies on manual feature engineering meaning that professionals must be involved in the feature extraction and matching. This process is both time-consuming and demands specific expertise. METHODS: This paper introduces a novel stereo matching framework DCStereo that realizes a fully automatic neural architecture design for the stereo matching of binocular laparoscopic images. The proposed framework utilizes a densely connected search space which enables a more flexible and diverse architecture composition. Furthermore, the proposed algorithm leverages the channel and path sampling strategies to reduce memory consumption during searching. RESULTS: Empirically, our searched DCStereo on the SCARED training dataset achieves a mean absolute error of 3.589 mm on the test dataset, which outperforms hand-crafted stereo matching methods and other approaches. Furthermore, when directly testing on the SERV-CT dataset, our DCStereo demonstrates better generalization ability than other methods. CONCLUSION: Our proposed approach leverages the neural architecture search technique and a densely connected search space for automatic neural architecture design in stereo matching of binocular laparoscopic images. Our method delivers advanced performance on the SCARED dataset and promising results on the SERV-CT dataset. These findings demonstrate the potential of our approach for improving clinical surgical navigation systems.
Subject(s)
Algorithms , Laparoscopy , Humans , Neural Networks, ComputerABSTRACT
PURPOSE: Flexible ureteroscopy (FURS) plays an important role in the diagnosis and treatment of urological diseases. However, manipulating a flexible ureteroscope to the target quickly and safely may be challenging because of the tortuous lumen or poor visibility. Thus, information on the shape of the anterior part of a flexible ureteroscope in addition to the real-time pose is needed to perform accurate maneuvering in the lumen with minimal impingement on the inner renal wall and resulting tissue damage in FURS. METHODS: An adaptive mixed-order Bézier curve fitting algorithm and electromagnetic tracking (EMT) technique were developed for shape estimation utilizing the length of the anterior part, kinematic constraints and the pose information provided by two electromagnetic (EM) sensors mounted at the tip and base of the anterior part. A series of experiments were performed to qualitatively and quantitatively verify the validity of our method. Moreover, algorithm threshold conditions with reference significance under various shape cases were studied. RESULTS: The performance of our method was evaluated based on 19 representative planar bending shapes that often appear in FURS and eight non-planar shapes, yielding an average error (AE) of 1.0 mm. Moreover, the experiments proved the feasibility of applying our method in cases in which large bending angles (near 270 degrees) occur. CONCLUSION: Based on data from two EM sensors mounted at the tip and base of the anterior part of a flexible ureteroscope, the proposed algorithm adaptively selects a cubic or quartic Bézier curve to fit the shape of the anterior part. Experimental results prove the feasibility of our shape estimation method over a broad bending range. The proposed method demonstrates significant potential for use in ureteroscopic navigation systems and robot-assisted surgery.
