Supplementary Methods and Videos of OrbitalPSI-GPT

Description

This repository contains extended technical documentation and supplementary video materials associated with OrbitalPSI-GPT, a text-driven, clinician-in-the-loop framework for patient-specific orbital implant (PSI) design. These materials provide additional implementation details, demonstrate the system's interactive design workflow, and support research transparency and reproducibility.

File Descriptions

1. Extended Technical Methods for OrbitalPSI-GPT.pdf

This document provides detailed technical documentation of the OrbitalPSI-GPT framework, including the region-aware multiscale Gaussian process morphable model (GPMM), automated landmark detection (ALD), dataset preparation, training and evaluation protocols, and model configuration information. It also documents anatomical constraints, reconstruction procedures, and model-version metadata to facilitate technical understanding and reproducibility.

2. Supplementary Video 1 | PSI Design Process via OrbitalPSI-GPT (Narrated Version)

This video provides a concise, edited demonstration of the core PSI design workflow, accompanied by narration and voice-over. It highlights the key interactions within the OrbitalPSI-GPT interface, illustrating the progression from natural-language instructions to the final patient-specific implant design.

3. Supplementary Video 2 | PSI Design Process via OrbitalPSI-GPT (Unedited/Raw Version)

This video contains the complete, unedited screen recording of a PSI design session. It documents the actual interaction sequence, system response latency, iterative design adjustments, and final design output without post-processing. The recording provides a transparent demonstration of the system's operational performance under the recorded conditions.

Related Resources

Source code and software updates: https://github.com/GaryGaoYao/OrbitalPSI-GPT

Associated manuscript: A Text-Driven Clinician-in-the-Loop Approach to Accessible Patient-Specific Orbital Implant Design for Low-Resource Settings.