Limits...
Image registration improves confidence and accuracy of image interpretation.

Erickson BJ, Patriarche J, Wood C, Campeau N, Lindell EP, Savcenko V, Arslanlar N, Wang L - Cancer Inform (2007)

View Article: PubMed Central - PubMed

Affiliation: Department of Radiology, Mayo Clinic, Rochester, MN, USA. bje@mayo.edu

AUTOMATICALLY GENERATED EXCERPT
Please rate it.

Sequential MR examinations of the brain are the primary method for clinical as well as research assessment of the effects of therapy on brain tumors... After completion, satisfactory alignment was visually confirmed using a ‘flicker test’ as well as linked cross-hairs (Fig 1)... For those series that did not align, interactive adjustment of algorithm parameters was performed to achieve satisfactory alignment... These 97 cases represented the final test set, and the consensus was the ‘correct’ answer... We found that image registration improved accuracy for 5 of the 6 readers, and the difference was statistically significant at the p < 0.03 level (see Figure 4)... Some have described a process for acquiring the MRI images in an aligned condition (Gedat, Braun et al.; )... This is likely to produce similar benefits... But it is unlikely to ever be as accurate as post-acquisition registration between high resolution scans... We and others have demonstrated improvements in inter-rater agreement as well as efficiency gains associated with automated alignment... Precise alignment is also a requisite for nearly every multiparametric image processing technique such as segmentation and classification We (Patriarche and Erickson) have described the use of precisely registered examinations for semi-automated detection of changes in brain tumors... We did not study issues of how the images should be optimally presented to a rater... We view this, as well as ‘in-place’ or flicker mode display option, as an important future direction of research on how to optimally interpret images... Displaying the images side-by-side is acceptable, though future research may demonstrate alternative display schemes that further increase the advantages image registration.

No MeSH data available.


Readers were more confident of their conclusions when comparing registered examinations.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC2666945&req=5

f5-cin-04-19: Readers were more confident of their conclusions when comparing registered examinations.

Mentions: We found that image registration improved accuracy for 5 of the 6 readers, and the difference was statistically significant at the p < 0.03 level (see Figure 4). There was also a higher confidence in ratings (Figure 5), though this was not statistically significant. Finally, there was a greater Az value for the registered exams as well: 0.87 vs 0.78 (see Figure 6). Anecdotally, there was also consensus that image registration made comparison more efficient and introduced no significant image degradation.


Image registration improves confidence and accuracy of image interpretation.

Erickson BJ, Patriarche J, Wood C, Campeau N, Lindell EP, Savcenko V, Arslanlar N, Wang L - Cancer Inform (2007)

Readers were more confident of their conclusions when comparing registered examinations.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC2666945&req=5

f5-cin-04-19: Readers were more confident of their conclusions when comparing registered examinations.
Mentions: We found that image registration improved accuracy for 5 of the 6 readers, and the difference was statistically significant at the p < 0.03 level (see Figure 4). There was also a higher confidence in ratings (Figure 5), though this was not statistically significant. Finally, there was a greater Az value for the registered exams as well: 0.87 vs 0.78 (see Figure 6). Anecdotally, there was also consensus that image registration made comparison more efficient and introduced no significant image degradation.

View Article: PubMed Central - PubMed

Affiliation: Department of Radiology, Mayo Clinic, Rochester, MN, USA. bje@mayo.edu

AUTOMATICALLY GENERATED EXCERPT
Please rate it.

Sequential MR examinations of the brain are the primary method for clinical as well as research assessment of the effects of therapy on brain tumors... After completion, satisfactory alignment was visually confirmed using a ‘flicker test’ as well as linked cross-hairs (Fig 1)... For those series that did not align, interactive adjustment of algorithm parameters was performed to achieve satisfactory alignment... These 97 cases represented the final test set, and the consensus was the ‘correct’ answer... We found that image registration improved accuracy for 5 of the 6 readers, and the difference was statistically significant at the p < 0.03 level (see Figure 4)... Some have described a process for acquiring the MRI images in an aligned condition (Gedat, Braun et al.; )... This is likely to produce similar benefits... But it is unlikely to ever be as accurate as post-acquisition registration between high resolution scans... We and others have demonstrated improvements in inter-rater agreement as well as efficiency gains associated with automated alignment... Precise alignment is also a requisite for nearly every multiparametric image processing technique such as segmentation and classification We (Patriarche and Erickson) have described the use of precisely registered examinations for semi-automated detection of changes in brain tumors... We did not study issues of how the images should be optimally presented to a rater... We view this, as well as ‘in-place’ or flicker mode display option, as an important future direction of research on how to optimally interpret images... Displaying the images side-by-side is acceptable, though future research may demonstrate alternative display schemes that further increase the advantages image registration.

No MeSH data available.