6.5 CNT Source Array Geometry Calibration
6.5.6 Discussion
In this section, we introduced an in-house geometry calibration procedure for the s-DCT system. The procedure utilizes a simple geometry calibration phantom and a statistical model based optimization method to analyze the data. Reconstructed images using the recovered coordinates of the focal spots using this procedure produces good image quality
Figure 6.47: An image of a metal bead used in the geometry calibration procedure. The blurred edge of the bead increases uncertainty in edge detection and feature recognition of the image.
and spatial resolution, as well as a lack of image artifacts, suggesting the procedure can provide an accurate geometry calibration for the s-DCT system. This procedure can be also applied to other imaging systems, such as s-DBT system.
The main source of error of this procedure comes from the uncertainty in identifying features from the projection images. Due to the finite size of the focal spot, instead of ideal point, and the fact that metal beads can not block all x-rays, the image of the beads typically have a blurred edge(Figure 6.47). The blurred edge increases uncertainty in finding the edges or features of the fiduciary markers in the image. However, it is crucial to find the center of the bead (or one focus of the ellipse) in the geometry calibration procedure.[81] A high accuracy algorithm to identify features and edges from the image is desired.
The introduced in this section gives a more accurate estimation of the focal spot positions. This method provides a better estimatestatistically, so it does not mean it is the best estimate every time.
CHAPTER 7: Evaluation of imaging geometries for the s-DCT system Section 7.1: Overview
The feasibility of stationary digital chest tomosynthesis (s-DCT) using a distributed carbon nanotube x-ray source array was demonstrated in Chapter 5. The technology has the potential to increase the imaging resolution and speed by eliminating source motion. In addition, the flexibility in the spatial configuration of the individual sources allows new tomosynthesis imaging geometries beyond the linear scanning mode used in conventional systems. In this chapter, we investigate the effects of tomosynthesis imaging geometry on the image quality. The study was performed using the bench-top s-DCT prototype. System MTF and ASF were used as a quantitative measurement of the in-plane and in-depth resolution, respectively.
In this chapter, imaging geometries with the x-ray sources arranged in linear, square, rectangular, and circular configurations were investigated using comparable imaging doses. Anthropomorphic chest phantom images were acquired and reconstructed for image quality assessment. It was found that wider angular coverage resulted in better in-depth resolution, while the angular span had little impact on the in-plane resolution in the linear geometry. Non-linear imaging geometry lead to more isotropic in-plane resolution and better in-depth resolution compared to a linear imaging geometry with comparable angular coverage.
Section 7.2: Purpose
In Chapter 5, we have demonstrated the feasibility of a CNT-based s-DCT system using the CNT source array. The CNT source array has been shown to deliver sufficient x-ray output for chest tomosynthesis applications.
The flexibility of the CNT source array also allows novel imaging configurations that conventional DCT systems cannot easily achieve. In current DCT systems, one x-ray source is moved in a linear path to acquire projection images.[1, 4] While the imaging configuration has been optimized for linear imaging,[5, 71, 30] the impact of non-linear imaging geometries on image quality has not been investigated, except in a few studies, due to the difficulties of managing tube momentum of a moving X-ray source.[82–84]
The purpose of this study is to investigate the relationship between source configuration and the tomosynthesis image quality using an s-DCT system.
Section 7.3: Methods
The same bench-top s-DCT system introduced in Chapter 5 was used for this study. Im- ages were acquired using different x-ray source configurations. The system resolution, char- acterized by reconstructed tomosynthesis modulation transfer function (MTF) and artifact spread function (ASF), were measured for all imaging geometry configurations. Anthropo- morphic chest phantom images were acquired and reconstructed for quality assessment.
7.3.1: System Descriptions
The bench-top s-DCT, as shown in Figure 7.1, consists of a CNT source array and a flat panel detector mounted on a translation stage. Only a subset of 75 focal spots in the source array were used to acquire projection images. The same CsI based detector (Model DRX-1C) with 139µm×139µm pixel size and 35cm×43cm field of view was used. The source-to-detector distance was 1100 mm.
7.3.2: Imaging Geometries
Since the source array used in this study has a relatively short tube length, the detector and phantom were mounted on a stage that could be translated simultaneously to simulate
Figure 7.1: The s-DCT system consists a linear CNT source array, a flat panel detector, a phantom and translation stages. The detector and phantom can be simultaneously translated to simulate various imaging geometries.
various imaging geometries. The SID was kept at 1100 mm. By translating the detector and phantom along both the x direction and y direction, an effective source matrix relative to the detector was emulated. Subsets of projection images from this effective source matrix were selected to simulate all imaging geometries. Figure 7.2 shows the imaging geometries studied. The effective source array of each imaging geometry was centered at the detector center. By translating the detector and phantom along the scanning direction (x-direction), three 1D linear geometries with different angular spans of 11.6◦, 23◦, and 34◦ were created. Each had 29 projections which were acquired at 80 kVp and 0.5 mAs per projection.
Translation of the detector perpendicular to the scanning direction (y-direction) allowed simulation of square, rectangular, and circular imaging geometries. All three 2D source array imaging geometries included 32 projections. Each projection image was acquired at 80 kVp and 0.5 mAs. All 6 geometries studied had comparable total imaging dose. The square geometry had a length of 192 mm and 10◦ angular coverage on each side. The rectangular geometry had a length of 256 mm (13.3◦ angular coverage) and a width of 192 mm (10◦ coverage) along x and y direction respectively. The circular geometry had a radius of 192 mm and 10◦ angular coverage.
−400 −300 −200 −100 0 100 200 300 400 −200 −100 0 100 200 X (mm) Y (mm)
Detector frame Square Linear Rectangular Circular
Figure 7.2: Imaging geometry configuration studied. All imaging configurations are using same SID of 1100 mm. A linear imaging geometry with angular span of 11.6◦, 23◦,and 34◦, a square geometry centered with detector center and 10◦ angular span on each side, a rectangular geometry with 13.3◦ and 10◦ angular coverage along x and y direction, and a circular geometry with 10◦ angular coverage are studied. The x-direction is along the source array scanning direction, which is also corresponding to the phantom spine direction, while the y-direction is defined as perpendicular to the scanning direction, as noted in both figures.
7.3.3: Image Analysis
System MTF and ASF were used to quantify the tomosynthesis in-plane and in-depth (z- axis) resolutions. They were measured using a 100µm diameter tungsten cross-wire phantom. Projection images of the phantom were acquired and reconstructed into tomosynthesis slices using commercial software from Real Time Tomography (Villanova, PA). The data set was reconstructed with 1 mm spacing.