L Koenig1, E Parks, M Analoui, G Eckert. 1. Oral Medicine and Oral Radiology, Marquette University School of Dentistry, 366, PO Box 1881, Milwaukee, WI 53201-1881, USA. lisa.koenig@marquette.edu
Abstract
OBJECTIVES: To test the hypothesis that there is no significant difference in the detectability of chemically-induced periapical lesions between a non-compressed image and one subjected to a Joint Photographic Experts Group (JPEG) lossy compression technique at a ratio of 23:1 or less. METHOD: Chemically-induced periapical lesions were created by placing a solution of 70% perchloric acid at the apex of extracted teeth in 13 human jaw cadaver specimens. Acid was applied in seven incremental time periods from 0-32 h. Extracted teeth were replaced in the socket and images were made using the Schick CDR digital sensor. Using a JPEG lossy compression algorithm, five compression ratios of 2:1, 14:1, 23:1, 28:1 and 47:1 were applied to the images. Images were viewed three times by three observers who ranked the presence or absence of a lesion at three sites, the mesial area, distal area and apex of the tooth, on a 5-point confidence scale. Intraobserver and interobserver agreement and agreement between the compressed and the original images were assessed with intraclass correlation coefficients (ICCs). RESULTS: Overall ICCs for measuring intraobserver agreement using all images were 0.77, 0.84, and 0.50 for the three observers, respectively. The overall ICC for assessing agreement between observers was 0.57. There was no significant difference (P>0.05) between compressed and original images for any site at compression ratios of 2:1, 23:1 and 28:1. There were significant differences for a compression ratio of 47:1. CONCLUSION: JPEG compression does not impact detectability of artificial periapical lesions at low and moderate compression ratios up to and including 28:1.
OBJECTIVES: To test the hypothesis that there is no significant difference in the detectability of chemically-induced periapical lesions between a non-compressed image and one subjected to a Joint Photographic Experts Group (JPEG) lossy compression technique at a ratio of 23:1 or less. METHOD: Chemically-induced periapical lesions were created by placing a solution of 70% perchloric acid at the apex of extracted teeth in 13 human jaw cadaver specimens. Acid was applied in seven incremental time periods from 0-32 h. Extracted teeth were replaced in the socket and images were made using the Schick CDR digital sensor. Using a JPEG lossy compression algorithm, five compression ratios of 2:1, 14:1, 23:1, 28:1 and 47:1 were applied to the images. Images were viewed three times by three observers who ranked the presence or absence of a lesion at three sites, the mesial area, distal area and apex of the tooth, on a 5-point confidence scale. Intraobserver and interobserver agreement and agreement between the compressed and the original images were assessed with intraclass correlation coefficients (ICCs). RESULTS: Overall ICCs for measuring intraobserver agreement using all images were 0.77, 0.84, and 0.50 for the three observers, respectively. The overall ICC for assessing agreement between observers was 0.57. There was no significant difference (P>0.05) between compressed and original images for any site at compression ratios of 2:1, 23:1 and 28:1. There were significant differences for a compression ratio of 47:1. CONCLUSION: JPEG compression does not impact detectability of artificial periapical lesions at low and moderate compression ratios up to and including 28:1.