Correlation between Quantitative and Qualitative Analysis on Image Quality of Digital Dental X-ray Images

Abstract

Signal to Noise Ratio (SNR) is a quantitative method to measure the effectiveness of contrast enhancement methods. The question is whether this measurement is consistent with a qualitative evaluation of a medical expert. Biomedical Research, including dentistry cases, usually involves medical experts to do the interpretation, and finalize the diagnosis. However, the subjective evaluations create differences of interpretation. There are two factors contributing to the low quality of dental radiographs-the first is because of the nature of the image acquisition process, and the second is the low X-ray dose usage. Due to that, research related to contrast enhancement algorithms is an accepted image processing method to assist medical officers in doing interpretation that is more reliable. Thus, this work presents an analysis of contrast enhancement algorithms (CEA) applied to digital intraoral radiograph images. The enhancement methods used in this study are CEAs, namely: adaptive histogram equalization (AHE), contrast limited adaptive histogram equalization (CLAHE), and sharp contrast limited adaptive histogram equalization (SCLAHE). The objective of this work is to find a correlation between the dentist evaluation and SNR values in determining the image quality based on the rate of abnormalities detected. Fifty-six original intraoral digital dental X-ray images are collected from the Faculty of Dentistry, Universiti Teknologi MARA Malaysia. These images are processed with the mentioned CEAs, resulting in a total of onehundred sixty-six observations. These images are compared against the original images. The method of assessment is through questionnaires where all images are arranged randomly. The observations focused on identifying widened periodontal ligament space (widened PDLs), periapical radiolucency (PR) and loss of lamina dura (loss LD) abnormalities. These are among the main signs of periapical disease. The results show that the SNR value does not reflect the dentist's evaluation in terms of image quality. The SNR values of CLAHE are able to precede other methods. However, SCLAHE is better than the rest based on dentists' evaluations for widened PDLs abnormality.