Toolbox Manual

1. Preamble

This toolbox was developed at the Computer Vision Laboratory (Laboratório de Visão Computacional - LAVI), at the University of São Paulo, São Carlos. The proposed toolbox was developed on MATLAB and is intended for academic usage, please review the license file. This is the first version of this manual, so if you have any concern, suggestion, bug report, please send me an email.

2. Introduction

This software was firstly presented at this paper, and further validated at this paper. If you use the toolbox, our group will be very grateful if you refer to any of these papers.

The toolbox was developed based on the cone-beam computed tomography (CBCT) code, and further extended to the digital breast tomosynthesis (DBT) geometry, as illustrated in this Figure:

Two commercial DBT systems were used as a reference for the geometry. The first one from General Electric (GE), model Senographe Essential, and the second from Hologic, model Selenia Dimensions. It is important to note that there is an angle movement on the latter system, while the former is stationary. Section 4 has some specific information for the implementation of detector rotation geometry.

Each code on this software has its own comments, input requirements and output variables. Also, each method has specific references. Make sure to read them as you go deep in the codes.

3. Configuration

The main code is named Reconstruction.m, and each specific reconstruction method has its proper name, e.g., SART.m. All other functions are located at the Functions folder, and the geometry configuration files are located at the Parameters folder. Follow these specific steps in order to perform a simple reconstruction:

1. First of all, it is mandatory to configure the geometry parameters. You can skip this step if you want to keep the geometry as default. Note that if you want to use the geometry of the two commercial systems mentioned, there is a specific file for each of them. Feel free to create a new one. The table below illustrates all settings parameters that must be set prior to the reconstruction process.

Group	Parameter	Meaning
Geometry type	type	Type of geometry used (GE, Hologic)
Voxel quantity	nx	Number of voxels on the X axis (columns)
	ny	Number of voxels on the Y axis (rows)
	nz	Number of slices
Detector element quantity	nu	Number of detectors on the x axis (columns)
	nv	Number of detectors on the y axis (rows)
Voxel and detector dimensions	dx	Single voxel real size on the X axis (mm)
	dy	Single voxel real size on the Y axis (mm)
	dz	Single voxel real size on the Z axis (mm)
	du	Single detector real size on the X axis (mm)
	dv	Single detector real size on the Y axis (mm)
Geometry distances	DSD	Distance from the x-ray source to the detector (mm)
	DSO	Distance from the x-ray source to the top of the object (mm)
	DDR	Distance from the detector to the pivot (mm)
	DSR	Distance from the x-ray source to the pivot (mm)
	DAG	Size of Air Gap (mm)
Geometry movement	nProj	Number of x-ray projections
	tubeAngle	Tube angle span (Degree)
	tubeDeg	Tube angles on each projection
	detAngle	Detector angle span (Degree)
	detectorDeg	Detector angles on each projection
Optional	sliceRange	The range of slices to be saved
	iROI	The range of rows to be saved
	jROI	The range of columns to be saved
	bitDepth	Number of quantization bits

2. There are two options on the toolbox. You will be asked for them as soon as you run the main code:

.

The first option is to use real DBT images (projections) with DICOM format. The second option creates a three dimensional (3D) virtual Shepp-Logan phantom, and its projections based on the specified geometry.

If you skipped step 1, you probably want to use this virtual phantom. In case you choose to use real images, select the folder that contains the image files. If the DBT projections are from the Hologic equipment, make sure to organize the image files name as follow: xxx\_0.dcm to xxx\_14.dcm, as illustrated in the figure above. For the GE equipment, the software takes the image number from the DICOM header.

In this step, you have to choose the reconstruction method that you want to use. There are actually five methods implemented:

• Back-projection (BP),
• Filtered Back-projection (FBP),
• Simultaneous Algebraic Reconstruction Technique (SART),
• Simultaneous Iterative Reconstruction Technique (SIRT) and
• Maximum-Likelihood Expectation-Maximization (MLEM).

They are all written in the main code, so you just have to uncomment the one that you want to use. Note that the default method is FBP

4. Geometry

The whole geometry was designed to reconstruct the right breast. In our system, the X and Y coordinates refer to the Chest Wall-Nipple direction and Lateral profile respectively, as shown in the figure below. The slices are reconstructed parallel to the detector along the Z coordinate.

The global coordinate system origin (X=0, Y=0, Z=0) is located at the detector plane, next to the chest wall and along with the lateral profile, as shown in the figure below.

The detector coordinate system (x=0, y=0), and the image coordinate system (i=1,j=1) are also located at the detector plane, as demonstrates the figure below. The origin of the image coordinate system refers to the cell 1 of MATLAB.

Note that if you want to implement geometries, which the detector rotates, e.g. Hologic, you must use the projectionDD.m and backprojectionDD.m. These codes implement this rotation.

Please, report any bug on this manual or in the codes.