Medical-grade Sterilizable Target for Fluid-immersed Fetoscope Optical Distortion Calibration

Sandblasting is an important step in the fabrication process because the raw metal surface prominently reflects the endoscope light, making it impossible for the circles to be detected. It is difficult to distinguish the circles even with the naked eye (see Figure 5). Note that the surface of the target shown was already etched with a laser. However, this does not diminish light reflection.

Figure 5: Calibration target with no sandblasting applied. As seen from the endoscope view on the left, the glare from the endoscope light on the material surface makes it difficult even for the naked eye to distinguish the circles (there is a circle just to the southeast of the large reflection). Note that the surface of this target (i.e., the "background") was already etched, but this is not helpful in the absence of sandblasting. Please click here to view a larger version of this figure.

Prior to pattern etching, it is also important to etch the surface of the whole sample. This is necessary because the sandblasted surface has many specular reflections (see Figure 6), which interfere with blob detection.

Figure 6: Sandblasted surface with no etching. Although not as prominent as the raw metal surface, the relatively small specular reflections (some of which are highlighted with yellow arrows) are still sufficient to prevent blob detection from succeeding, so no calibration can be performed with this target. Please click here to view a larger version of this figure.

Applying the laser at different speeds gives different background colors. The background color plays a significant role in the contrast between the circles and the background. Hence, it is vital to determine the optimal background color. For this purpose, a plate with circles etched against a set of different backgrounds was created (see Figure 7). The backgrounds were tested using the feature detection module of OpenCV²³, which is used in the OpenCV camera calibration module¹⁷. In this work, the target was made of stainless steel, as it is the most common and reliable material used in clinics for medical devices. This material is freely available, not expensive, robust, and easy to sterilize. Other materials could potentially be used for the calibration target, such as aluminum or iodized metals, but this is the scope of future work.

Figure 7: Stainless steel plate featuring a palette of different background colors etched with the laser. Practical experiments were conducted in conjunction with the OpenCV feature detection module to determine which background color gives the optimal result in terms of blob-to-background contrast²³. The endoscope view on the left shows the plate. The moderate background colors (i.e., those other that the darkest and lightest ones) in this palette yield better blob detection. Please click here to view a larger version of this figure.

One of the advantages of this work is that performing a calibration using the fabricated target takes 2-3 min. Most of the effort goes to manually stabilizing the endoscope to obtain decent views of the calibration pattern. Using a custom-built endoscope holder could eliminate the need for manual stabilization, which in turn could significantly shorten calibration time.

Video 1: Video showing how optical distortion calibration can be performed using the developed calibration target together with the endocal software. Please click here to view this video. (Right-click to download.)

An advantage of our work compared to the work of Wengert et al.¹⁵ is that the OpenCV camera calibration module¹⁷ can be used as is for calibration, without requiring any modification or custom parameterization. Because OpenCV is a well-established and well-maintained software package and is very popular in the computer vision community, using it eliminates the need for writing and maintaining custom software. For the reader's convenience, a compact GUI application is provided¹⁶, which the reader can easily install and use to test new calibration targets. One disadvantage of our method compared to Wengert et al.¹⁵ is that their method is more robust to occlusions of the pattern, as it does not require the detection of all blobs.

Initially, a calibration target with a checkerboard pattern was fabricated for this work. However, this type of calibration target proved to be unsuitable in experiments due to the difficulty of detecting the corners of the checkerboard squares. Corner detection relies on histogram-based image binarization (see the OpenCV source code²⁴). This implies the need for a clear color contrast between the dark and light squares, which could not be guaranteed with our checkerboard pattern, partially due to specular reflections, like the ones shown in Figure 6. Such specular reflections are present even after background etching; however, the detection of the circles seems to be less sensitive to this shortcoming.

In the current setup, only perpendicular views of the calibration target allow for successful blob detection. This is due to the specular reflections from the target surface hampering blob detection at oblique angles. We are working to further improve the target so as to allow for the acquisition of views at a wider range of angles, which could potentially improve the quality of performed calibrations²⁰.

In the real-time placental mosaicing pipeline that was previously proposed¹¹, the computation of the transformation that maps image pairs relies on the successful detection and grouping of features. Optical distortions, on the other hand, cause a group of features with a rigid geometry to appear different across images. As a consequence, this difference leads to inaccuracies in the computed transformations, which cause drifts in the resulting image mosaics. Because the most prominent optical distortions are present towards the edges, endoscopic images are currently cropped to their innermost regions. A good correction for optical distortions would potentially allow for the incorporation of a larger part of each image into the mosaicing process. The advantage of this method is two-fold. First, it would increase the number of detected features in each image, potentially improving the computation of the image transformations. Second, it would allow for the whole target anatomical surface to be reconstructed in a shorter time.