| MS TSA Meeting

Name of the Speaker: Mr. GURRALA AJAY KUMAR (EE18S001)
Guide: Prof. Mohanasankar S
Online meeting link: https://meet.google.com/the-xrrm-rkz
Date/Time: 22nd February 2024 (Thursday), 04:00 PM
Title: Design, Development and Validation of a Bronchoscopy System.

Abstract

Medical endoscopy, utilizing endoscopes—a flexible or rigid tube with a camera and light source—allows visualization of internal structures through natural orifices or minimal incisions, through procedures like gastroscopy and bronchoscopy. CMOS image sensors employed in modern endoscopes suffer from vertical fixed pattern noise (VFPN), exacerbated by long signal cables and limited dynamic range. Dark frame subtraction is a common method to address VFPN, but it compromises image quality if sensor parameters like gain and black level clamp (BLC) are not properly considered. HDR imaging improves the visibility and clarity of images obtained during endoscopic procedures. Nevertheless, conventional HDR techniques are frequently intricate and impractical for real-time endoscopy, where optimal performance is paramount.

Here, we introduce two novel image processing methods developed as part of the design and development of a bronchoscopy system. The first method minimizes VFPN by adapting the dark frame subtraction technique to account for the impact of sensor gain and BLC. The second method attains HDR-like image quality in real-time by generating multiple low-pixel-bit-depth frames from a high-pixel-bit-depth frame and employing them to produce the high-quality image.

Our adapted dark frame subtraction technique shows a significant 30% enhancement in peak signal-to-noise ratio (PSNR) over conventional methods. Likewise, the HDR-like imaging approach provides enhanced visualization of subtler details in both dark cavities and bright areas, resulting in a uniformly exposed view and improved contrast among various tissue types. Validation via simulated bronchoscopy procedures confirms the efficacy of both methods in enhancing endoscopic imaging quality.

Furthermore, we offer practical insights into the design, development, and validation of the bronchoscopy system. Importantly, the system's adaptable design allows repurposing for diverse endoscopic procedures to significantly reduce non-recurring engineering (NRE) costs for the OEM and the total cost of ownership (TCO) for clinicians, which is very critical in resource constrained healthcare markets like India.