[Paper Review] An Elastic Image Registration Approach for Wireless Capsule Endoscope Localization

Isabel N. Figueiredo, Carlos Leal+3 more|arXiv (Cornell University)|Apr 23, 2015

Gastrointestinal Bleeding Diagnosis and Treatment26 references4 citations

TL;DR

This paper proposes a multiscale elastic image registration method to improve wireless capsule endoscope (WCE) localization by modeling both rigid-like motion and non-rigid intestinal deformations due to peristalsis. The approach outperforms affine-only registration in estimating WCE speed, location, and orientation, especially under elastic deformations, with mean orientation errors below 10⁻³ and scale errors below 10⁻⁵ in synthetic tests and superior performance on real WCE video frames.

ABSTRACT

Wireless Capsule Endoscope (WCE) is an innovative imaging device that permits physicians to examine all the areas of the Gastrointestinal (GI) tract. It is especially important for the small intestine, where traditional invasive endoscopies cannot reach. Although WCE represents an extremely important advance in medical imaging, a major drawback that remains unsolved is the WCE precise location in the human body during its operating time. This is mainly due to the complex physiological environment and the inherent capsule effects during its movement. When an abnormality is detected, in the WCE images, medical doctors do not know precisely where this abnormality is located relative to the intestine and therefore they can not proceed efficiently with the appropriate therapy. The primary objective of the present paper is to give a contribution to WCE localization, using image-based methods. The main focus of this work is on the description of a multiscale elastic image registration approach, its experimental application on WCE videos, and comparison with a multiscale affine registration. The proposed approach includes registrations that capture both rigid-like and non-rigid deformations, due respectively to the rigid-like WCE movement and the elastic deformation of the small intestine originated by the GI peristaltic movement. Under this approach a qualitative information about the WCE speed can be obtained, as well as the WCE location and orientation via projective geometry. The results of the experimental tests with real WCE video frames show the good performance of the proposed approach, when elastic deformations of the small intestine are involved in successive frames, and its superiority with respect to a multiscale affine image registration, which accounts for rigid-like deformations only and discards elastic deformations.

Motivation & Objective

• To address the critical lack of precise WCE localization during wireless capsule endoscopy, which hinders accurate diagnosis and treatment planning.
• To model both rigid-like capsule motion and non-rigid intestinal deformations caused by peristalsis in WCE video sequences.
• To develop an image-based method that estimates WCE speed, location, and orientation using elastic image registration.
• To compare the proposed elastic registration with multiscale affine registration in terms of accuracy and robustness under deformation.
• To provide a reliable, computationally efficient, and clinically useful tool for WCE motion tracking using only image data.

Proposed method

• The method employs a multiscale elastic image registration framework that captures both rigid-like and non-rigid deformations in consecutive WCE frames.
• It begins with affine pre-registration to handle global rigid motion, followed by non-rigid refinement using a multiresolution approach.
• The registration uses normalized cross-correlation (NCC) as the similarity measure and the normalized dissimilarity measure (NDM) to evaluate registration quality.
• A projective geometry model based on the pinhole camera model is used to estimate WCE location and orientation from the transformation parameters.
• The method is validated using both synthetic data (with rotation and scaling) and real WCE video frames from clinical data.
• The approach is compared against a multiscale parametric (affine) registration method, which only models rigid-like motion and ignores elastic deformations.

Experimental results

Research questions

• RQ1 Can a multiscale elastic image registration approach accurately estimate WCE motion, including both rigid-like and non-rigid deformations, in real WCE video sequences?
• RQ2 How does the performance of elastic registration compare to affine registration in terms of WCE location, orientation, and speed estimation under elastic intestinal deformations?
• RQ3 To what extent does the normalized dissimilarity measure (NDM) correlate with registration accuracy and WCE localization error?
• RQ4 Can the proposed method reliably estimate WCE speed and orientation from image data alone, even under significant image rotation and scaling?
• RQ5 Does the method maintain low error in scale and orientation estimation across a wide range of synthetic transformations, especially at extreme scales and rotation angles?

Key findings

• The proposed elastic registration achieved mean orientation errors of approximately 10⁻³ for rotation angles up to 40°, with a notable increase to 10⁻¹ only at 45°, outperforming prior methods.
• Scale estimation errors remained below 10⁻⁵ for most scale factors, with only moderate increases at extreme scales, demonstrating robustness.
• In synthetic tests involving simultaneous rotation and scaling, the method maintained mean absolute errors below 1.74 for rotation and below 10⁻⁵ for scale, showing high accuracy.
• On real WCE video frames, the method provided qualitatively useful WCE speed estimates through the NDM measure, which correlated strongly with localization accuracy.
• The elastic registration significantly outperformed the affine-only method when elastic deformations were present, confirming its necessity for realistic WCE imaging.
• The results suggest that NDM is a more reliable indicator of registration quality than MEIR, as lower NDM values corresponded to smaller localization errors.