Task-Based Assessment of Post-processed Image Quality in Full-Field Digital Mammography Using the Task Transfer Function
Related Experiment Video
Updated: Nov 5, 2026
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
C Murata1, T Hara1, A Urikura2
1Department of Medical Technology, Nakatsugawa Municipal General Hospital, 1522-1 Komanba, Nakatsugawa, Gifu 508-8502, Japan.
Introduction:
To develop a task-based method for evaluating the spatial resolution characteristics of post-processed full-field digital mammography (FFDM) images and to investigate the influence of image-processing algorithms on image quality.
Methods:
Post-processed images were acquired using a commercially available FFDM system. Spatial resolution was evaluated by calculating the task transfer function (TTF) from images of a 10-mm acrylic disk embedded in oil, simulating the contrast between a breast mass and the surrounding glandular tissue. TTFs obtained with six image-processing algorithms were compared with the presampled modulation transfer function (pMTF) derived from the raw image data. Additionally, the contrast-to-noise ratio (CNR) was calculated for each algorithm using the same phantom.
Results:
The TTFs of the post-processed images exceeded unity (>1.0) at low spatial frequencies and showed significant suppression at higher spatial frequencies. These characteristics varied according to the image-processing algorithm: The 50% TTF values of the post-processed images were approximately 20%-35% lower than those of the pMTF, and the TTF response decreased to below 40% at 5 cycles/mm. The CNR ranged from approximately 2.2 to 3.7 and showed no consistent relationship with the TTF results.
Conclusion:
TTF analysis based on the proposed method demonstrated that the spatial-frequency characteristics of post-processed FFDM images differed from those reflected by the pMTF derived from raw image data. Furthermore, no consistent correlation was observed between TTF and CNR across the evaluated image-processing algorithms. These findings suggest that TTF reflects the effective spatial-frequency characteristics of post-processed FFDM images and, when combined with CNR, provides complementary information relevant to task performance.
Implications For Practice:
TTF and CNR may provide complementary information for evaluating post-processed mammographic image quality in clinical practice.
Related Concept Videos
01:27Imaging Studies III: Computed Tomography
01:25Imaging Studies II: Positron Emission Tomography and Scintigraphy
Fundamental Principles of PET