Z Grelewicz*, K Tatebe, Rush University Medical Center, Chicago, IL

• 
  Z Grelewicz

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  

PO-GePV-M-204 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: Advanced MRI techniques such as Diffusion Tensor Imaging (DTI) tractography can provide useful information for delineating disease and normal tissue for cranial radiotherapy. These images encode the components of the dominant eigenvector of the diffusion tensor onto red, green, and blue color channels, indicating the direction of nerve bundles. Because DTI tractography images are three channel color images, many commercial contouring systems are incapable of correctly processing them. We investigated the feasibility of a work-around to incorporate these advanced imaging techniques into the treatment planning workflow.

Methods: A MIM extension was developed to bring DTI tractography images into the MIM workspace. The extension parsed the raw DTI gradient images to calculate the diffusion tensors, and then independently calculated the tractography image. The tractography image was then down-sampled from true color format (24 bits) to 8-bit using a custom color palette. The color palette was derived from the combined images of five patients. The down-sampled 8-bit image and unified color palette were imported into the MIM workspace for contouring. As a pilot study, the extension was used on five patients to assist a board-certified radiation oncologist in contouring brainstem, with and without tractography images. The contour similarity was assessed in terms of Dice coefficient to determine how much the inclusion of tractography affects contouring decision making.

Results: The MIM extension was successful in generating tractography images capable of being used for registration and contouring in the MIM workspace. For the five patients investigated, the average Dice coefficient between brainstems contoured with and without tractography was 0.91, with standard deviation 0.02.

Conclusion: A work-around exists for the problem of using derived true color images in commercially available contouring and registration systems. This work-around may applied to additional advanced imaging series, such as perfusion imaging, in the future.

Keywords

Treatment Planning, MRI, Contour Extraction

Taxonomy

IM/TH- Image Analysis Skills (broad expertise across imaging modalities): Image processing

Contact Email