CT and MRI image fusion

Software Registration Basics

The process of software image registration plays a key role in modern oncology. Automated algorithms are used to overlay one image onto another, although manual correction by a specialist may be required in certain clinical situations.

Objective factors often hinder the ideal superposition of structures. The main causes of inaccuracies include:

• Patient movement or internal organ shift between different scanning stages.
• Differences in resolution between CT and MRI scanners.
• Presence of spatial distortions.
• Use of tomographic slices of unequal thickness.

To minimize these errors, deformable image registration (deformable registration) programs are being actively implemented. The essence of this technology is the targeted distortion of one of the images to achieve its most precise match with the anatomical structures on the other scan.

Image-Guided Radiation Therapy (IGRT)

Image-guided radiation therapy techniques aim to precisely deliver a dose to the tumor while protecting healthy structures. They are classified by the dimensionality of the acquired data:

Planar (2D) imaging is based on acquiring two static X-ray images, typically at a 90-degree angle to each other (in anteroposterior and lateral projections). This method helps assess the position of bony structures or a visible tumor using conventional X-ray radiation. The main problem with this method is that using linear accelerator therapeutic energies (6–18 MV) yields extremely low contrast between air, bone, and soft tissues compared to low-energy kilovoltage (kV) imaging.

Volumetric (3D) imaging makes it possible to identify the target organ and surrounding healthy tissues in the treatment position directly on the linear accelerator (before or during therapy). The acquired data are overlaid onto the dose distribution map of the planning CT.

Four main methods are used to obtain volumetric images on the accelerator:

1. Cone-beam CT (CBCT): Utilizes a kV tube mounted on the gantry at a 90-degree angle to the treatment head, rotating around the patient.
2. Megavoltage CT: Uses a fan beam of megavoltage energy to generate a verification volumetric image (helical imaging).
3. CT-on-rails: The scanner is located in the same room as the accelerator, and the treatment couch is rotated 180 degrees to move the patient into the scanner.
4. Ultrasound: Used for 3D target positioning control in patients with prostate or bladder cancer.

From Static to 4D Imaging

Four-dimensional imaging accounts for the motion of the tumor and adjacent organs over time.

In the classical approach, the gross tumor volume was expanded to the clinical target volume to cover areas of microscopic disease spread. An additional margin was then added to compensate for patient setup errors and any anatomical shifts, forming the final planning target volume.

The modern approach (with 4D) uses kilovoltage fluoroscopy or CBCT to precisely and quantitatively measure tumor mobility:

• The gross tumor volume is determined at the mid-position and extreme positions of the motion amplitude.
• Extended clinical volumes are created based on them.
• Combining these volumes forms a single internal target volume.
• Only a small margin for potential geometric setup errors is added to it.

An alternative option is tumor tracking directly during the irradiation session using 4D kV or MV imaging. To improve target visibility when contours are indistinct, implanted fiducial markers (fiducial markers) are placed in the tissue. An example of complex tracking is the CyberKnife system, which combines X-ray verification with optical motion tracking.

Practical Application of Position Verification

In practice, the linear accelerator's onboard imaging system is used before irradiation begins.

Two-dimensional control generates digitally reconstructed radiographs based on the planning CT, on which bony landmarks are contoured. These contours are overlaid onto current kilovoltage images of the patient lying on the treatment couch. If a discrepancy is detected (e.g., the anteroposterior image shows that the mandible and spinous processes are 3–4 mm higher than planned), patient repositioning or field adjustment is required. Such shifts are common in head and neck tumors due to patient weight loss—the immobilizing thermoplastic mask begins to fit loosely, allowing head mobility.

A more advanced level of control is provided by cone-beam CT. For instance, in a T4 laryngeal tumor, the device performs a scan right on the treatment couch before a fraction. The image is reconstructed in sagittal, coronal, and transverse planes, and then registered with the planning CT. The registration area is best visualized on coronal and sagittal slices because CBCT covers a shorter craniocaudal length. Unlike 2D scans, CBCT visualizes soft tissues excellently, enabling three-dimensional position correction. This ensures adequate tumor dose coverage and protection of normal tissues.

Class and exam questions

What else to study on this topic

• Deformable registration algorithms
• GTV, CTV, and PTV delineation in classical and 4D planning
• Application of cone-beam CT (CBCT) on linear accelerators
• Use of fiducial markers and tumor tracking
• Patient positioning errors in head and neck radiation therapy

Other topics in the "Diagnostic Radiology in Oncology" section

Prepared by WhiteCoat Pro faculty in accordance with the medical school curriculum. For educational purposes only.