Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
PolyPose: Deformable 2D/3D Registration via Polyrigid Transformations
Authors: Vivek Gopalakrishnan, Neel Dey, Polina Golland
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Across extensive experiments on diverse datasets from orthopedic surgery and radiotherapy, we show that this strong inductive bias enables Poly Pose to successfully align the patient s preoperative volume to as few as two X-rays, thereby providing crucial 3D guidance in challenging sparse-view and limited-angle settings where current registration methods fail. |
| Researcher Affiliation | Academia | Vivek Gopalakrishnan MIT EMAIL Neel Dey MIT, MGH, and HMS EMAIL Polina Golland MIT EMAIL |
| Pseudocode | No | The paper describes the methodology in Section 3, including equations and workflow diagrams (Figure 3), but does not present a formal pseudocode block or an algorithm section. |
| Open Source Code | Yes | Additional visualizations, tutorials, and code are available at https://polypose.csail.mit.edu. |
| Open Datasets | Yes | We first perform experiments on a longitudinal dataset of CT scans of 31 patients undergoing radiotherapy for head and neck squamous cell carcinoma [58]... [58] Tatiana Bejarano, Mariluz De Ornelas-Couto, and Ivaylo B Mihaylov. Head-and-neck squamous cell carcinoma patients with CT taken during pre-treatment, mid-treatment, and post-treatment (HNSCC-3DCT-RT). https://doi.org/10.7937/K9/TCIA.2018.13upr2xf, 2018. ... Deep Fluoro. To measure performance on real X-ray images, we use Deep Fluoro, a cadaveric orthopedic surgery dataset of six preoperative CT volumes with associated intraoperative X-ray images (between 24-111 per subject) [60]... [60] Robert B Grupp, Mathias Unberath, Cong Gao, Rachel A Hegeman, Ryan J Murphy, Clayton P Alexander, Yoshito Otake, Benjamin A Mc Arthur, Mehran Armand, and Russell H Taylor. Automatic annotation of hip anatomy in fluoroscopy for robust and efficient 2D/3D registration. International Journal of Computer Assisted Radiology and Surgery, 15:759–769, 2020. |
| Dataset Splits | Yes | Head&Neck. We first perform experiments on a longitudinal dataset of CT scans of 31 patients undergoing radiotherapy for head and neck squamous cell carcinoma [58] using a 10/2/19 subject-wise training, validation, and testing split. |
| Hardware Specification | Yes | Poly Pose and all baseline methods were trained (if applicable) and evaluated using a single NVIDIA RTX A6000. |
| Software Dependencies | No | The paper mentions using 'Adam optimizer [57]' and 'open-source implementations of the rendering equation (3) as a series of vectorized tensor operations [50]', but does not provide specific version numbers for software dependencies or libraries. |
| Experiment Setup | Yes | For both camera and structure-specific pose estimation, we perform gradient-based optimization on rigid transforms parameterized in the tangent space se(3). Across all experiments, we use the Adam optimizer [57] with step sizes βrot = 10 2 and βxyz = 100 for the rotational and translational components of se(3), respectively. |