Premium
TH‐CD‐207B‐02: An Automated Technique to Measure Spatial Resolution in Clinical CT Images: Application to Patient Data
Author(s) -
Sanders J,
Ding A,
Samei E
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4958208
Subject(s) - image resolution , iterative reconstruction , kernel (algebra) , mathematics , optical transfer function , projection (relational algebra) , medical imaging , artificial intelligence , algorithm , nuclear medicine , computer science , pattern recognition (psychology) , medicine , mathematical analysis , combinatorics
Purpose: To evaluate an automated technique for measuring spatial resolution across a database of clinical CT exams. Methods: A fully automated algorithm was developed to extract a CT resolution index (RI) analogous to the modulation transfer function from clinical CT images by measuring the edge‐spread function (ESF) across the patient's skin and converting the results into scalar values, frequency at 50% RI, f50. The program was previously validated against observer data. This algorithm was applied to a database of CT images from our hospital that were reconstructed with different reconstruction algorithms (filtered back‐projection and iterative) and reconstruction kernels (soft and hard). The results were analyzed in terms of mean and variability within reconstruction methods and kernel aspects of the protocols. Results: The automated algorithm successfully measured the RI index from all of the clinical datasets examined. The measured f50 values increased with harder kernels for both FBP and iterative reconstruction. The mean f50 was 0.30 ± 0.02 mm‐1 and 0.42 ± 0.03 mm‐1 for images reconstructed with soft and hard kernels, respectively, using filtered back‐projection. The corresponding values for iterative reconstructions were 0.34 ± 0.02 mm‐1 and 0.39 ± 0.03 mm‐1, respectively. Overall, there was more variability in the f50 measurements made on datasets reconstructed with a hard kernel. The differences were statistically significant (p<0.05). Conclusion: Clinically‐informed, patient‐specific spatial resolution can be measured from clinical datasets. The method is sufficiently sensitive to reflect changes in spatial resolution due to different reconstruction parameters. The method can be applied to automatically assess the spatial resolution of patient images and quantify dependencies that may not be captured in phantom data.