Abstract Archives of the RSNA, 2004
SSM05-04
The Effect of Image Compression on Computer-assisted Nodule Volume Measurement
Scientific Papers
Presented on December 1, 2004
Presented as part of SSM05: Chest (Lung Nodules: Growth)
Jane P. Ko MD, Presenter: Nothing to Disclose
Jeffrey Chang MD, Abstract Co-Author: Nothing to Disclose
Elan Bomsztyk, Abstract Co-Author: Nothing to Disclose
James S Babb PhD, Abstract Co-Author: Nothing to Disclose
David Paul Naidich MD, Abstract Co-Author: Nothing to Disclose
Henry Rusinek PhD, Abstract Co-Author: Nothing to Disclose
To assess the effect of 2D JPEG lossy image compression on computer assisted nodule volume assessment.
Fifty-one nodules were selected from low-dose CT scans and compressed using a 2D JPEG lossy image compression method to 10:1, 20:1, 30:1, 40:1 levels. Nodules were characterized in terms of size, location and attenuation. Nodule volumes on original and non compressed images were obtained using three computer assisted methods. A mixed model analysis of variance was conducted to evaluate the impact of compression level, nodule characteristics, and measurement algorithm on volume estimates.
Mean nodule volumes were 388.0mm3 (34-3474 mm3). Three were calcified, 33 solid non-calcified, and 15 were solid (13 ground-glass (GG)) in attenuation. 32 were adjacent to the pleura. No significant difference between measurements obtained on original images and images compressed to 10:1 was identified. A significant difference between original images and those compressed to 20:1 and higher was noted (mean volume loss 8.3 mm3 for solid and 34.9 mm3 for GG nodules at 20:1, P<0.05). The level of image compression significantly interacted with nodule size, location, and attenuation (P<0.001). The magnitude of the detrimental effect of image compression was greater for GG nodules and central nodules (45.9 mm3 for central vs. 12 mm3 for peripheral nodules at 20:1, P<0.05).
Nodule volume measurements differed significantly at a 20:1 compression level, particularly for ground-glass nodules, when compared with measurements obtained from original non-compressed images and may lead to less accurate assessment of nodule growth.
Ko, J,
Chang, J,
Bomsztyk, E,
Babb, J,
Naidich, D,
Rusinek, H,
The Effect of Image Compression on Computer-assisted Nodule Volume Measurement. Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL.
http://archive.rsna.org/2004/4407205.html