Methodological Considerations for Assessing Automatic Brightness Control in Endoscopy: Experimental Study

Abstract

Endoscopy is a critical application that requires adaptable illumination to adjust to varying imaging conditions. Automatic brightness control (ABC) algorithms ensure optimal brightness throughout the image with rapid but smooth response and render the true colours of the biological tissue under examination. To achieve good image quality, high-quality ABC algorithms are necessary. In this study, we propose a three-assessment method approach for objectively evaluating ABC algorithms based on (1) image brightness and its homogeneity, (2) controller response and response time, and (3) colour rendition. We conducted an experimental study to assess the effectiveness of ABC algorithms in one commercial and two developmental endoscopy systems using the proposed methods. The results showed that the commercial system achieved good, homogeneous brightness within 0.4 s, and its damping ratio was 0.597, indicating a stable system, but its colour rendition was suboptimal. The developmental systems had control parameter values that resulted in either a sluggish response (over 1 s) or a fast (about 0.3 ms) but unstable response with damping ratios above 1, causing flickers. Our findings indicate that the interdependency among the proposed methods can establish tradeoffs in the overall ABC performance better than single-parameter approaches. The study establishes that comprehensive assessments using the proposed methods can contribute to designing new ABC algorithms and optimising already implemented ones for efficient performance in endoscopy systems.

Keywords: automatic brightness control; colour rendition; controller response; image brightness.

Figure 1

Representaion of the three methodological considerations to be made to assess the performance of the ABC algorithms.

Figure 2

Experimental setup. (a) Endoscope railing to allow the adjustment of the distance of endoscope tip from the test chart. (b) Endoscopy system. (c) Endoscope mounted on the railing. (d) Illumination box with standard D65 light. (e) Test chart.

Figure 3

(a) Scattering on a matte surface. (b) Scope tips of the endoscope systems. S0 has 3 sets of light sources around its camera, while S1 and S2 have only 2 sets of light sources adjacent to their cameras.

Figure 4

(a) Illumination profile depicted by clustering the bins of the histogram of the observed image frame. (b) Transient response of a second order system with different damping ratios when a step input is given. (c) Vectorscope designed with the NTSC graticule as reference.

Figure 5

Histogram of the frame after the response has settled and illumination profile of E0 in (a) S0, (b) S1, and (c) S2.

Figure 6

Image masks of S0 and S1, which are present in all of the endoscopy image frames.

Figure 7

Histogram of the frame after the response has settled and illumination profile of E1 in (a) S1, (b) S2 and E2, (c) S1, and (d) S2.

Figure 8

Response curve in E0 for (a) S0, (b) S1, and (c) S2.

Figure 9

Response curve in E1 for (a) S1 and (b) S2, in E2 for (c) S1 and (d) S2, and when flickers occur in E2 for (e) S1 (f) S2.

Figure 10

Rendition of white on vectorscope by S0, S1, and S2.

Figure 11

Rendition on the vectorscope. (a) Red (R) by S0, S1, and S2; (b) cyan (C) by S0, S1, and S2; (c) green (G) by S0, S1, and S2; (d) magenta (M) by S0, S1, and S2; (e) blue (B) by S0, S1, and S2; and (f) yellow (R) by S0, S1, and S2.

Figure 12

Lists of the endoscopy systems (S0, S1, and S2), the emulations (E0, E1, and E2), and the assessments (A1, A2, and A3) employed in this experimental study.