Algorithms for Real-Time Endoscopy Image Processing Pipeline in Clinical Decision Support Systems

Algorithms for Real-Time Endoscopy Image Processing Pipeline in Clinical Decision Support Systems

Alexandr A. Pozdeev (Saint Petersburg Electrotechnical University “LETI,” St. Petersburg, Russia), Nataliia A. Obukhova (Saint Petersburg Electrotechnical University “LETI,” St. Petersburg, Russia) and Alexandr A. Motyko (Saint Petersburg Electrotechnical University “LETI,” St. Petersburg, Russia)
DOI: 10.4018/IJERTCS.2019100103

Abstract

A set of algorithms, taking in account endoscopic image features and computational cost for real-time realization is proposed. A noise reduction algorithm is based on determining the level of detail in an image fragment. For fragments with a different level of detail, different noise reduction filters are used. The enhancement algorithm is based on nonlinear contrast enhancement which highlights the contrast of vessels relative to the background without significant noise stressing, which is one of the main disadvantages of nonlinear enhancement algorithms. The custom color correction algorithm takes into account user preferences and provides a mean error less than 0.5% for each color coordinate. The “mosaic” synthesis algorithm gets panoramic images of low detail images with a mean stitching error less than 0.75 pix. The software realization of algorithms allows processing 4K endoscopic video with a speed of about 30 fps.
Article Preview
Top

Introduction

Clinical decision support system (CDSS) represents a modern trend in the development of medical video systems. Systems of this type integrate automatic image analysis results with the results obtained by the physician, as well as use the information from the system database. This integration allows for higher sensitivity and specificity of the diagnosis compared to cases of diagnoses being made independently by a physician, or the system (Liedlgruber, & Uhl, 2011).

Thus, visual analysis conducted by the physician is an important part of CDSS. This demands for high image quality for the physician to conduct analysis and, as a consequence, enables to ensure high efficiency of visual analysis which requires to:

  • Develop the algorithms to improve the quality of medical images, providing for a high level of image ergonomics;

  • Develop the algorithms for the synthesis of medical images with new properties and with increased diagnostic value.

The problem definition in this research is to develop the algorithms for image improvement and visualization for real–time pipeline realization for CDSS in endoscopy (gastroscopy).

In this case the main requirements for the algorithms are the following:

  • 1.

    High effectiveness of the algorithm, taking in account the features of endoscopic images. The main features of endoscopic images which make popular algorithms of digital processing low effective for application in pipeline are: uneven contrast (there are very dark and very bright areas in the same image), large areas of image with very low amount of detail;

  • 2.

    Each algorithm must be designed as a step of the endoscopic pipeline, which implies high processing speed. The computational cost of algorithms must give possibility of real – time pipeline realization, also the result of one step must be improved by the next step.

Additionally, the designed algorithms and software have to be adapted to real time processing of endoscopic video data with high resolution (up to FHD, 4K) and frame rate (up to 30 – 60 fps).

The main pipeline steps can be divided in two groups. The first group must provide the obtainment of quality images from a sensor. The algorithms of the second group are aimed at removing artifacts caused by difficult observation conditions. Such conditions include the lack of light, complex shape of objects under observation, organ muscular breathing, movement of the sensor, which leads to the following artifacts: high presence of noise, nonlinear brightness and contrast characteristics, blur and low sharpness. The critical procedures of this group are: noise reduction and contrast enhancement. These tasks are especially difficult in endoscopy because of the fact the endoscopic images suffer from various types of degradation (Campiho & Kamel, 2014). Moreover, these algorithms are the most computationally expensive. Thus, the first part of our research includes the development of real-time algorithms of the second group, such as noise removing, brightness and contrast enhancement, taking into account the main features of endoscopic images.

The following new real-time algorithms for the endoscopic pipeline were proposed:

  • The algorithm of noise removing, which is based on experimental results of modern noise reduction algorithms assessment. The algorithm determines the level of high frequency in the image fragment (level of detail in the fragment). The median/KNN filtering and NLM were used for low detailed fragments and high detailed fragments respectively. The algorithm enables one to obtain high quality images with reasonable computational cost;

  • The algorithm of image enhancement. For image sharpening and contrast enhancement it is proposed to consider the local features of the image on the basis of the found functional relationship between the correction strength and the normalized brightness variance in the image fragment. The algorithm carries out contrast enhancement without significant stressing of the noise component, which is one of main disadvantages of modern non-linear enhancement algorithms, especially in low detail images;

  • The algorithm of special custom color correction, which enables the physician to adjust the picture according his or her own color preference.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 11: 4 Issues (2020): 1 Released, 3 Forthcoming
Volume 10: 4 Issues (2019)
Volume 9: 2 Issues (2018)
Volume 8: 2 Issues (2017)
Volume 7: 2 Issues (2016)
Volume 6: 2 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing