Improving 90Y PET Scan Image Quality Through Optimized Reconstruction Algorithms

Article Figures & Data

Download figure
Open in new tab
Download powerpoint
Download figure
Open in new tab
Download powerpoint
FIGURE 1.
Box plots comparing median quality assessment scores among algorithms. Results from Kruskal–Wallis rank sum test and Wilcoxon rank sum test are included.
Download figure
Open in new tab
Download powerpoint
FIGURE 2.
A 52-y-old man with rectal neuroendocrine cancer metastatic to liver underwent ⁹⁰Y-PRRT therapy. Reconstructed PET/CT using algorithms A (A), B (B), and C (C) managed to detect hepatic metastases (dotted arrows) seen on corresponding CT images (D). However, there was more visible noise within liver for PET using algorithms A and B than for PET using algorithm C (solid arrows). In addition. extrahepatic noise such as that in right adrenal gland and spleen (arrowheads) was less apparent using algorithm C. Right adrenal noise can potentially be mistaken as hepatic metastasis using algorithms A and B (arrowheads).

TABLE 1.

Parameters of the 3 Tested Algorithms

Parameter	Algorithm A	Algorithm B	Algorithm C
VUE Point	VUE Point HD (OSEM)	VUE Point FX (OSEM + TOF)	VUE Point HD (OSEM)
Gaussian filter cutoff	6.4 mm	6.0 mm	5.0 mm
Number of subsets	24	18	32
Sharp IR (point-spread function)	On	On	On
z-axis filter	Standard	Heavy	Standard
Number of iterations	2	3	1
Matrix	192 × 192	192 × 192	192 × 192
Minutes per bed position	30	30	30

TABLE 2.

Scoring Criteria for Image Quality Assessment

QS	Rating	Description
1	Nondiagnostic	Excessive noise or artifacts; delineation of tumor and background uptake mostly impossible
2	Barely diagnostic	Substantial noise and artifacts; delineation of tumor and background uptake difficult but possible
3	Fairly diagnostic	Somewhat noisy and artifacts that interfere with reading; delineation of tumor and background uptake feasible but not satisfactory
4	Diagnostically excellent	No interfering noise or artifacts; delineation of tumor and background uptake satisfactory

TABLE 3.

Patients’ Demographic Data

Participant	Age (y)	Sex	BMI (kg/m²)	Diagnosis	Therapy	Radiotracer dose (GBq)
1	52	M	20.6	Rectal NET	PRRT	3.70
2	58	M	24.9	Midgut NET	PRRT	3.70
3	39	M	19.2	Paraganglioma	PRRT	4.22
4	54	F	21.8	Midgut NET metastatic to liver	PRRT	3.03
5	41	M	19.7	Pancreatic NET metastatic to liver	SIRT	2.97
6	68	M	25.9	HCC	SIRT	1.30
7	69	M	26.1	HCC	SIRT	0.58
8	59	M	29.9	HCC	SIRT	2.50
9	96	M	25.9	HCC	SIRT	0.73
10	79	F	21.4	HCC	SIRT	3.00

BMI = body mass index; NET = neuroendocrine tumor; HCC = hepatocellular carcinoma.

TABLE 4.

Number and Percentage of Discrete Scores Rated by 10 Readers on 10 Patients’ Scans Reconstructed Using Algorithms A–C

Algorithm	Therapy	Score 1	Score 2	Score 3	Score 4	P
A	SIRT	0 (0.0%)	11 (18.3%)	38 (63.3%)	11 (18.3%)	<0.001
	PRRT	10 (25.0%)	24 (60.0%)	6 (15.0%)	0 (0.0%)
	SIRT + PRRT	10 (10.0%)	35 (35.0%)	44 (44.0%)	11 (11.0%)
B	SIRT	0 (0.0%)	28 (46.7%)	29 (48.3%)	3 (5.0%)	<0.001
	PRRT	14 (35.0%)	13 (32.5%)	13 (32.5%)	0 (0.0%)
	SIRT + PRRT	14 (14.0%)	41 (41.0%)	42 (42.0%)	3 (3.0%)
C	SIRT	0 (0.0%)	0 (0.0%)	15 (25.0%)	45 (75.0%)	<0.001
	PRRT	1 (2.5%)	8 (20.0%)	24 (60.0%)	7 (17.5%)
	SIRT + PRRT	1 (1.0%)	8 (8.0%)	39 (39.0%)	52 (52.0%)

TABLE 5.

Multivariate Analysis Comparing Quality Assessment Scores