Validating computer applications for calculating spatial resolution and noise property in CT using simulated images...

Area of Science:

• Medical Imaging Physics
• Radiological Technology Assessment

Background:

• Accurate image quality assessment in computed tomography (CT) is crucial for diagnostic performance.
• Software tools are widely used for calculating image quality metrics like modulation transfer function (MTF), task transfer function (TTF), and noise power spectrum (NPS).

Purpose of the Study:

• To evaluate the accuracy of computer applications used for calculating MTF, TTF, and NPS in CT.
• To assess the performance of specific algorithms within the 'CTmeasure' software using simulated images with known properties.

Main Methods:

• Utilized simulated CT images with known property values, generated using a roll-off filter function for radial symmetry.
• Evaluated three MTF (wire method), two TTF (circular edge/linear edge methods), and one NPS (2D Fourier transform) applications integrated in 'CTmeasure'.
• Compared calculated image quality metrics against true values to determine accuracy.

Main Results:

• MTF calculations using the wire method demonstrated high accuracy with <1.0% error.
• TTF calculations (CE and LE methods) and NPS calculations exhibited significant errors (up to 50% and 30%, respectively) at high frequencies.
• Identified measurement technique insufficiencies, rather than software algorithms, as the primary cause of TTF and NPS errors.

Conclusions:

• The evaluated CT image quality calculation techniques are fundamentally sound.
• Improving measurement conditions significantly reduced errors in TTF and NPS calculations.
• The study validates the calculation algorithms within 'CTmeasure' when appropriate measurement practices are employed.