Adaptive statistical iterative reconstruction improves image quality without affecting perfusion CT quantitation in primary colorectal cancer

Abstract

Objectives: To determine the effect of Adaptive Statistical Iterative Reconstruction (ASIR) on perfusion CT (pCT) parameter quantitation and image quality in primary colorectal cancer.

Methods: Prospective observational study. Following institutional review board approval and informed consent, 32 patients with colorectal adenocarcinoma underwent pCT (100 kV, 150 mA, 120 s acquisition, axial mode). Tumour regional blood flow (BF), blood volume (BV), mean transit time (MTT) and permeability surface area product (PS) were determined using identical regions-of-interests for ASIR percentages of 0%, 20%, 40%, 60%, 80% and 100%. Image noise, contrast-to-noise ratio (CNR) and pCT parameters were assessed across ASIR percentages. Coefficients of variation (CV), repeated measures analysis of variance (rANOVA) and Spearman' rank order correlation were performed with statistical significance at 5%.

Results: With increasing ASIR percentages, image noise decreased by 33% while CNR increased by 61%; peak tumour CNR was greater than 1.5 with 60% ASIR and above. Mean BF, BV, MTT and PS differed by less than 1.8%, 2.9%, 2.5% and 2.6% across ASIR percentages. CV were 4.9%, 4.2%, 3.3% and 7.9%; rANOVA P values: 0.85, 0.62, 0.02 and 0.81 respectively.

Conclusions: ASIR improves image noise and CNR without altering pCT parameters substantially.

Keywords: Colorectal neoplasms; Computer-assisted image processing; Multidetector computed tomography; Perfusion imaging; Radiation dosage.

Fig. 1

Axial perfusion CT image reconstructed with 0% (A), 60% (B) and 100% ASIR (C). Parametric blood flow colour map (D).

Fig. 2

Box and whisker plots of perfusion CT parameters across increasing ASIR levels. Solid line in box represents median value, upper and lower bars represent first and third quartiles, whiskers represent 95% confidence intervals. P values of repeated measures ANOVA (with or without GG correction) are reported.

Fig. 3

Box and whisker plots of image noise and CNR across increasing ASIR levels. Solid line in box represents median value, upper and lower bars represent first and third quartiles, whiskers represent 95% confidence intervals.