Qualitative and Quantitative Assessment of Nonlocal Means Reconstruction Algorithm in a Flexible PET Scanner

CT image denoising methods for image quality improvement and radiation dose reduction

With the ever-increasing use of computed tomography (CT), concerns about its radiation dose have become a significant public issue. To address the need for radiation dose reduction, CT denoising methods have been widely investigated and applied in low-dose CT images. Numerous noise reduction algorithms have emerged, such as iterative reconstruction and most recently, deep learning (DL)-based approaches. Given the rapid advancements in Artificial Intelligence techniques, we recognize the need for a comprehensive review that emphasizes the most recently developed methods. Hence, we have performed a thorough analysis of existing literature to provide such a review. Beyond directly comparing the performance, we focus on pivotal aspects, including model training, validation, testing, generalizability, vulnerability, and evaluation methods. This review is expected to raise awareness of the various facets involved in CT image denoising and the specific challenges in developing DL-based models.

Brain imaging of sequential acquisition using a flexible PET scanner and 3-T MRI: quantitative and qualitative assessment

OBJECTIVE

A mobile PET scanner termed flexible PET (fxPET) has been designed to fit existing MRI systems. The purpose of this study was to assess brain imaging with fxPET combined with 3-T MRI in comparison with conventional PET (cPET)/CT.

METHODS

In this prospective study, 29 subjects with no visible lesions except for mild leukoaraiosis on whole brain imaging underwent 2-deoxy-2-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG) cPET/CT followed by fxPET and MRI. The registration differences between fxPET and MRI and between cPET and CT were compared by measuring spatial coordinates. Three-dimensional magnetization-prepared rapid acquisition gradient-echo T1-weighted imaging (T1WI) was acquired. We applied two methods of attenuation correction to the fxPET images: MR-based attenuation correction, which yielded fxPET_MRAC; and CT-based attenuation correction, which yielded fxPET_CTAC. The three PET datasets were co-registered to the T1WI. Following subcortical segmentation and cortical parcellation, volumes of interest were placed in each PET image to assess physiological accumulation in the brain. SUV_mean was obtained and compared between the three datasets. We also visually evaluated image distortion and clarity of fxPET_MRAC.

RESULTS

Mean misregistration of fxPET/MRI was < 3 mm for each margin. Mean registration differences were significantly larger for fxPET/MRI than for cPET/CT except for the superior margin. There were high correlations between the three PET datasets regarding SUV_mean. On visual evaluation of image quality, the grade of distortion was comparable between fxPET_MRAC and cPET, and the grade of clarity was acceptable but inferior for fxPET_MRAC compared with cPET.

CONCLUSIONS

fxPET could successfully determine physiological [¹⁸F]FDG uptake; however, improved image clarity is desirable. In this study, fxPET/MRI at 3-T was feasible for brain imaging.

Application of a Flexible PET Scanner Combined with 3 T MRI Using Non-local Means Reconstruction: Qualitative and Quantitative Comparison with Whole-Body PET/CT

PURPOSE

Flexible positron emission tomography (fxPET) employing a non-local means reconstruction algorithm was designed to fit existing magnetic resonance imaging (MRI) systems. We aimed to compare the qualitative and quantitative performance of fxPET among fxPET with MR-based attenuation correction (MRAC), fxPET with CT-based attenuation correction (CTAC) using CT as a part of WB PET/CT, and whole-body (WB) PET/CT.

PROCEDURES

Sixteen patients with suspected head and neck cancer underwent 2-deoxy-2-[¹⁸F]fluoro-D-glucose WB PET/CT scans, followed by fxPET and 3 T MRI scans. Phantom data were compared among the three datasets. For registration accuracy, we measured the distance between the center of the tumor determined by fxPET and that in MRI. We compared image quality, detection rates, and quantitative values including maximal standardized uptake value (SUVmax), metabolic tumor volume (MTV), total lesion glycolysis (TLG), and tumor-to-muscle ratio (TMR) among the three datasets.

RESULTS

The phantom data in fxPET, except the percent contrast recoveries of 17-mm and 22-mm hot spheres, were inferior to those in WB PET/CT. The mean registration accuracy was 4.4 mm between fxPET using MRAC and MRI. The image quality was comparable between two fxPET datasets, but significantly inferior to WB PET/CT (p < 0.0001). In contrast, detection rates were comparable among the three datasets. SUVmax was significantly higher, and MTV and TLG were significantly lower in the two fxPET datasets compared with the WB PET/CT dataset (p < 0.005). There were no significant differences in SUVmax, MTV, and TLG between the two fxPET datasets or in TMR among the three datasets. All quantitative values had significantly positive correlations.

CONCLUSIONS

Compared with WB PET/CT, the phantom data and image quality were inferior in fxPET. However, the results of the detection rates and quantitative values suggested the clinical feasibility of fxPET.