A Novel Hybrid Endoscopic Dataset for Evaluating Machine Learning-based Photometric Image Enhancement Models
This work addresses illumination issues in endoscopy for medical diagnosis and computer vision applications, but it is incremental as it focuses on dataset creation and method evaluation without a major breakthrough.
The paper tackled the problem of illumination artifacts in endoscopic images, which can cause misdiagnosis or hinder computer vision methods, by introducing a new synthetically generated dataset and evaluating image-enhancement methods, with the deep-learning-based LMSPEC method achieving the best quantitative results at around 7.6 fps.
Endoscopy is the most widely used medical technique for cancer and polyp detection inside hollow organs. However, images acquired by an endoscope are frequently affected by illumination artefacts due to the enlightenment source orientation. There exist two major issues when the endoscope's light source pose suddenly changes: overexposed and underexposed tissue areas are produced. These two scenarios can result in misdiagnosis due to the lack of information in the affected zones or hamper the performance of various computer vision methods (e.g., SLAM, structure from motion, optical flow) used during the non invasive examination. The aim of this work is two-fold: i) to introduce a new synthetically generated data-set generated by a generative adversarial techniques and ii) and to explore both shallow based and deep learning-based image-enhancement methods in overexposed and underexposed lighting conditions. Best quantitative results (i.e., metric based results), were obtained by the deep-learnnig-based LMSPEC method,besides a running time around 7.6 fps)