Accurate tissue characterization in low-dose CT imaging with pure iterative reconstruction

Area of Science:

• Radiology
• Medical Imaging
• Quantitative CT

Background:

• Accurate Hounsfield unit (HU)-determined tissue characterization is crucial in CT imaging.
• Low-dose CT protocols are increasingly used to reduce radiation exposure.
• Assessing the impact of different iterative reconstruction (IR) techniques on quantitative accuracy is important.

Purpose of the Study:

• To evaluate the accuracy of low-dose hybrid iterative reconstruction (IR) and pure model-based IR (MBIR) for Hounsfield unit (HU)-determined tissue characterization.
• To compare the performance of different reconstruction algorithms on low-dose CT data for quantitative analysis.

Main Methods:

• Standard-protocol (SP) and low-dose modified-protocol (MP) CTs were acquired in 34 Crohn's disease patients.
• Images were reconstructed using filtered back projection (FBP), Adaptive Statistical iterative Reconstruction (ASiR), and MBIR.
• Tissue volumes (fat, soft tissue, bone/contrast) were measured using HU thresholds.

Main Results:

• Inter-observer agreement for tissue volume measurements was strong (rs > 0.9).
• MP-MBIR reconstructions demonstrated superior tissue volume measurement compared to other MP reconstructions.
• MP-MBIR results closely correlated with reference SP-ASiR40 images, particularly for fat volume estimation.

Conclusions:

• Hounsfield unit-determined tissue volume calculations from MP-MBIR images are comparable to standard-protocol ASiR40.
• Pure IR (MBIR) offers superior accuracy for quantitative tissue volume estimation on low-dose CT compared to hybrid ASiR methods.
• Optimal accuracy for tissue volume estimation on low-dose CT is achieved with pure iterative reconstruction.