Skip to main content

Extension Guide and Ownership

Extension Guide​

This section is a practical guide for changing the system without losing track of the main architectural seams.

Add a New Backend Study Field​

  1. Add the property in JsonFormatUtility.cs
  2. Map it in StudyMapper.cs
  3. Store it in StudyRuntimeSO.cs if it is needed after fetch
  4. Update any UI presenters that should display it

Add a New Modality​

  1. Ensure backend exposes the new asset as a VRDF-like asset
  2. Ensure StudyMapper can derive the modality code
  3. Add the modality entry to the dropdown UI if needed
  4. Ensure the VRDF file naming matches VolumeDVR.LoadVolumeByCodeAsync() expectations

Add a New Segmentation UI Behavior​

  1. Extend VolumeDVR label control methods if needed
  2. Update BrainMenuToggleToVolumeDVR
  3. Optionally add richer item behavior via XRMenuItemInteractable

Replace or Extend the AI Backend​

  1. Update StudyRuntimeSO or backend payloads if endpoint discovery changes
  2. Extend GeminiClient transport logic
  3. Update GeminiVoiceInterface response parsing and history loading rules
  4. Keep confidential-mode behavior aligned with the new backend contract

This document should be maintained whenever one of the following changes:

  • The startup pipeline changes
  • The backend API or websocket contract changes
  • The VRDF format changes
  • The shader/material contract changes
  • The voice/AI backend changes
  • Build profiles or Android security settings change

Suggested future companion documents:

  • docs/VRDF_FORMAT.md
  • docs/VOICE_AND_AI.md
  • docs/BUILD_AND_RELEASE.md
  • docs/SECURITY_NOTES.md

Summary​

CASSANDRA XR is a functional Unity XR application with a clear product focus:

  • backend-driven study selection
  • custom medical volume loading
  • mobile-aware URP volume rendering
  • immersive metadata and report viewing
  • voice-driven AI assistance

Its current architecture is practical and product-oriented, but still prototype-shaped:

  • scene-composed rather than strongly modular
  • operationally effective but lightly tested
  • performance-aware but not yet fully optimized
  • feature-rich but not fully hardened

The most important architectural center of gravity is the path:

StudyService -> SessionDataController -> StudyRuntimeSO -> VolumeDVR -> Shader

with the voice/AI system acting as a second major vertical slice:

PorcupineWakeWordListener -> GeminiVoiceInterface -> GeminiClient -> ChatManager

Together, those two slices define the core identity of the application.

Continue Reading​