EP457 - Virtual Reality Enabled Segmentation for Volumetric Analysis of Response to Neoadjuvant Therapy for Rectal Cancer

ePoster Presenter(s)

• 

  Christine Rodhouse, MD (she/her/hers)

  Resident physician
  University of Florida
  University of Florida
  Gainesville, FL, United States

  Disclosure information not submitted.

Author(s)

• 

  Grace R. Thompson, MD

  Resident Physician, General Surgery
  University of Florida, College of Medicine, Department of Surgery
  UF College of Medicine, Department of Surgery
  Gainesville, FL, United States

  Disclosure information not submitted.
• JC

  Jennifer C. Cremer, n/a

  PhD Candidate
  University of Florida
  Gainesville, Florida, United States

  Disclosure information not submitted.
• 

  Pavel Mazirka, MD

  Surgery Resident
  University of Florida - Gainesville, FL
  University of Florida
  Gainesville, FL, United States

  Disclosure information not submitted.
• 

  Jeremy A. Balch, MD, PhD (he/him/his)

  House Officer
  University of Florida
  University of Florida
  Gainesville, Florida, United States

  Disclosure information not submitted.
• JN

  Johan Nordenstam, MD, PhD

  Associate Professor
  University of Florida, College of Medicine, Department of Surgery
  Gainesville, Florida, United States

  Disclosure information not submitted.
• 

  Thomas E. Read, MD

  Professor and Chief, Division of Gastrointestinal Surgery
  Dept of Surgery Univ of Hawaii
  University of Hawaii John A. Burns School of Medicine
  Honolulu, Hawaii, United States

  All of the relevant financial relationships listed below have been mitigated.
  This individual has no financial relationships with ineligible companies.
• JP

  Jorg Peters, PhD

  Professor
  University of Florida
  Gainesville, Florida, United States

  Disclosure information not submitted.

Senior Author(s)

• KT

  Krista P. Terracina, MD

  Assistant Professor
  University of Florida, College of Medicine, Department of Surgery
  Gainesville, Florida, United States

  Disclosure information not submitted.

Purpose/Background: Neoadjuvant approaches to rectal cancer offer the potential to avoid surgery in select patients. Current strategies for quantitating responses to therapy are limited to assessing length and stage of the tumor, with volumetric analysis relying on manual segmentation, stacking multiple 2D MRI images in sequence to approximate 3D structures. Virtual reality (VR) renders the same images in a 3D model that can be rapidly manipulated by clinicians. In the setting of volumetric analysis for rectal cancer, VR may be superior to traditional manual segmentation. This is a pilot study analyzing response to neoadjuvant therapy for rectal cancer using VR-enabled segmentation.

Methods/Interventions: Rectal cancer imaging data and clinical data were acquired from patients enrolled in our database at the University of Florida (IRB 202201077). Pre and post treatment MRI DICOM data were modeled in VR. First, a desktop interface using a traditional stack presentation of the images enabled annotation of initial tumor boundaries and selection of isovalues related to the tumor tissue. The annotated region of interest (ROI) and isovalues were used to generate a rapid voxelization of the tumor. Voxelation was assessed in the VR environment using a commercially available headset and controllers. The VR interface provided the T2 and DWI scans as pass-through image planes aligned to the voxel model to curate the tumor boundaries. Once finalized, the voxel representation was converted into a surface model using MeshLab's Alpha Shapes filter and geometrically evaluated to calculate a total volume for the object. Pre and post treatment were compared and are reported with pathological outcome when available and with the staff radiologist's MRI interpretation.

Results/Outcomes: MRI data from five patients with at least T2 tumors were analyzed. Patients were between 40-70 years old at time of data acquisition. None of the tumors selected had downstaging of the T stage in radiologist reads, though 3/5 had reduction in length of the tumor reported by radiology. Tumors that had length reduction reported by radiology all had volume reduction by our method. There was poor correlation with the final pathologic greatest dimension of the tumor for either method which is consistent with previous reports of tumor shrinkage during fixation.

Conclusion/Discussion: This report demonstrates the potential for VR enabled segmentation of rectal cancer tumor volume to assess response to treatment. VR enabled analysis of total tumor volume changes may provide a useful adjunct to guide treatment decisions.

Table 1