A Simplified Onsite Image-Registration Approach for Radiosurgery by Partial CT

Project Enquiry:

Fields with * are mandatory

ABSTRACT

Stereotactic radiosurgery (SRS) is a kind of method of radiation therapy that uses three-dimensional computerized imaging to precisely guide a high-powered X-ray beam to deliver a concentrated dose of radiation to the abnormal area of the body. This therapy can efficiently  and successfully treat many different types of tumors, either benign or malignant.

Commonly  used radiosurgery treatments are applied by CyberKnife or Gamma Knife through Stereotactic mechanism. Stereotactic radiosurgery can be assisted by two and three dimensional imaging  during a course of radiation treatment. This is usually called Image-guided radiation therapy  (IGRT), which utilizes the imaging coordinates to direct radiation beams to follow the actual radiation treatment plan. The image-guided system is the essential item to enable superior  accuracies in the dose delivery by locating the patient’s instantaneous position during  radiation treatments.

In CyberKnife system current setup, two X-ray imaging sources and  cameras are orthogonally mounted around the patient allowing instantaneous X-ray images to be  obtained. This thesis presents a method that uses the technique of computed tomography, called partial CT, to guide the stereotactic surgery and significantly simplifies the image registration procedure by one set of X-ray sources and radiation detectors. The image  processing for the partial CT is based on mutual information algorithm, a technically successful method, which ensures the accuracy of the image guidance system.

All the experiments are demonstrated with respect to the CyberKnife system. The thesis includes:  1) an overview with detailed representations for stereotactic radiosurgery techniques and devices of image guidance system applying to the radiation therapy field; 2) a hypothesis of mutual-information-based method used in partial CT image registration,  demonstrations of the feasibility of the proposed method as a new image guidance system by our experiments; 3) the innovative design of the robotic couch with one X-ray camera to obtain  patient’s partial CT image in the treatment stage and experimental results.

LITERATURE REVIEW

 CyberKnife: A Robotic Frameless Stereotactic Radiosurgery System:

The CyberKnife radiosurgery system has been developed in recent years, as a frameless, robotic, image -guided stereotactic radiosurgery system by manipulating an X-band linear accelerator. This recent adaptation allows for a more flexible treatment both in terms of the ability to deliver the therapy without using a frame (making the experience more comfortable for the patient) as well as increasing the fractionation flexibility. In 1996, Murphy and Cox described the accuracy of the first-generation CyberKnife and found it comparable to that of existing frame-based systems.

Cyberknife System Includes X-ray Imaging System, Optical Camera, Linear Accelerator and Collimators, Robotic Manipulator and Robotic Couch.

Cyberknife System Includes X-ray Imaging System, Optical Camera, Linear Accelerator and Collimators, Robotic Manipulator and Robotic Couch.

2D/3D Image Registration in Robotic Stereotactic Radiosurgery:

Recently, 2D/3D image registration is getting more and more attention because of the development of computer-assistet surgery (CAS). These CAS systems need to match preoperative images and plans to the intraoperative situation, to determine the relative position of surgical tools and anatomy, and to accurately position and move surgical robots.

BACKGROUND OF IMAGE -GUIDED SYSTEM IN CYBERKNIFE

CT Scan and Digitally Reconstructed Radiographs:

IGRT makes use of many different imaging techniques, using modalities ranging from
planar imaging to fluoroscopy to cone-beam CT. CT can fully describe a volume of tissue by producing closely spaced axial slices of patient anatomy. As we introduced, the orthogonal-intraoperative-image pairs are acquired to register to digitally reconstructed radiographs (DDR) derived from the pretreatment CT data set by aligning implanted fiducials or bony anatomy. A digitally reconstructed radiograph (DRR) is the artificial version of an X-ray image.

Pencil-beam Ct Scan.the X-ray Beam Penetrates the Target With the Opposite Detector Measuring the Transmitted Beam Intensity. After One Parallel Scanning Procedure, the X-ray Tube and Detector Rotates to Another Angle.

Pencil-beam Ct Scan.the X-ray Beam Penetrates the Target With the Opposite Detector Measuring the Transmitted Beam Intensity. After One Parallel Scanning Procedure, the X-ray Tube and Detector Rotates to Another Angle.

System and MI Method Applying to Onsite Image-Registration:

As we discussed in Chapter 2, XSI system dominates the radiosurgery field. In order to register patient’s positions between the planning stage and the treatment stage, two orthogonally projected images from +45° and -45°of the patient under the CyberKnife are constantly generated during the treatment period. This pair of projection images is then compared with artificially generated projection images from the CT data during planning stage, but with various positions.

DESIGN OF APPROACH FOR IMAGE-GUIDED RADIATION THERAPY

Partial CT and Radon Transform:

Definition of Partial CT

Computerized Tomography (CT) has been widely used in hospitals and clinics. A CT scanner casts a set of X-rays to pass through tissues, bones, and whatever organs. The radiation intensities attenuate to what are measured at the exit by an array of detectors. The measurements form the so-called projection data. Filtered Back-Projection method based on Fourier Slice Theorem is typically adopted to reconstruct the projection data in frequency domain and recover the tomographic information.

Onsite Image -Registration Designs:

This section presents some designs for acquiring partial CT data based on one set of X-ray source and radiation detectors. Contrasting to conventional CT’s set up, where an X-ray source rotates around the subject, our partial CT device sets the X-ray source stand still or rotates slightly and uses a robot-assisted couch to collect projection data at the same time.

Cambered Detector Attaches to the Back Side of the Robotic Couch; X-ray Tube Stays Right Above the Couch Before the Radiosurgery Treatment.

Cambered Detector Attaches to the Back Side of the Robotic Couch; X-ray Tube Stays Right Above the Couch Before the Radiosurgery Treatment.

SIMULATION RESULTS

Simulated Experiment between XRI System and MI Method by MATLAB:

Application Programming Interfaces: MATLAB. To provide a convenient way of programming the graphics and generating DRR images, an application programming interface (MATLAB) is needed. MATLAB (matrix laboratory) is a high-level language and interactive environment for numerical computation, visualization, and programming.

MATLAB Interface.

MATLAB Interface.

Mapping Maximum Mutual Information between Partial CT Images:

In the first section of this Chapter, we have discussed that the limited-scanning-angle partial CT reconstruction images are hard to be recognized. However, we asked previously that if the partial CT reconstruction images acquired at the patient’s treatment stage, the images can match the partial CT data generated from the full-scanning CT images. We use the feasible method demonstrated in the previous section, mutual information, to find the most similar images by mapping all the intensities in the images.

Simplified Onsite Image-Registration Approach for Radiosurgery:

In this section, we are going to prepare more complicated planning stage matrixes to simulate the possibilities of patient’s rotation and translation. For the three images acquired from treatment stage, can we prove that mutual information is still working between these partial CT sinograms and reconstructions with very small variations? In order to illustrate the process as much as possible, the design has been chosen and highlighted as the basic design and the foundation of our theory.

CONCLUSION

The first Chapter of this thesis elaborates the development and application of stereotactic radiosurgery system including Gamma-knife, Linear accelerator (LINAC) and LINAC based CyberKnife system, which is the model we used to implement our experiments. Image-guided radiation therapy are also reviewed in this Chapter because our hypothesis and experiments are all around the field of image-guidance system.

In the following, we fully review the robotic frameless CyberKnife system based on LINAC system specifically in the field of 2D/3D image registration. Hardware and software serving in the current CyberKnife system are explained and recent 2D/3D image registration methods are studied. With the help of 2D/3D registration methods, surgical robots may be programmed by using a preoperative 3D dataset library and a set of intraoperative fluoroscopic or on-site X-ray projections.

 FUTURE WORKS

  • Error analysis: we cannot finish the error analysis job in this thesis because of the huge dataset take us a lot of time to conduct. In the future, we are going to do the error analysis of the value of mutual information between partial CT reconstructed images and partial CT sonograms; between wider partial CT scanning angles and narrower partial CT scanning angles; between partial CT reconstructed images and X-ray stereotactic image pairs, etc.. Calculating the accuracy of our design will play an important part in our future studies.
  • So far, our last experiment is on 4D only because the limitation of the time. However,
    the real simulation should go 6D for any rigid subject. In next studies, we are going to
    simulate 6D translation and rotation: x, y, z, yaw, pitch and roll, as the patient’s real
    displacement during radiosurgery treatment. These variable quantities in 6D should be more accurate with smaller variations. Also, the accuracy is the indispensable indicator in the next studies.
  • Graphical User Interface (GUI) is a best way to build an interface that allows users to
    interact with our experiments by using images of result. The goal of our GUI is to help users to understand and verify our experiments by choosing their own combination of patient’s intraoperative position through GUI interface and can observe the result compares with their own input data.

Source: University of Miami
Author: Wupeng Yin

Download Project

>> Image Processing Project Topics with Full Reports and Free Source Code

Project Enquiry:

Fields with * are mandatory

Leave a Comment

Your email address will not be published. Required fields are marked *