Dataset - Benchmarking commercial depth sensors for intraoperative markerless registration in neurosurgery applications

Dataset - Benchmarking Commercial Depth Sensors for Intraoperative Markerless Registration in Neurosurgery Applications

Dataset Overview

This dataset includes a series of experiments for benchmarking commercial depth sensors in the context of intraoperative markerless registration for neurosurgery applications. The dataset contains calibration and registration data for multiple devices, focusing on 3D registration and transformation evaluation of point cloud data.

Devices and Data Included

Baseline data: Collected from authors and users, providing reference data for comparisons.

Depth Sensors:

Intel D405

Intel D435f

Intel D435f_no_proy (texture proyector off)

OakD

Photoneo-S

Zed-M

Zed-M_plus (neural network refinement)

The dataset includes both calibration data and results from registration experiments using these devices.

Directory Structure

The dataset is organized into the following directory structure for all the cameras:

├── Baseline
│ ├── Authors
│ ├── Authors_day_2
│ ├── Authors_day_3
│ └── User
├── D405
│ ├── Calibration
│ └── Register
├── D435f
│ ├── Calibration
│ └── Register
├── D435f_no_proy
│ ├── Calibration
│ └── Register├── ...

Directory Descriptions

Baseline/: Contains baseline data from registration experiments. This data is divided into subdirectories for each experiment type:

Authors/, Authors_day_2/, Authors_day_3/: Registration experiments performed by the authors on different days.

User/: Registration experiments involving different users.

D405/, D435f/, D435f_no_proy/, OakD/, PhotoneoS/, ZedM/, ZedM_plus/: Data related to specific Intel depth sensors. Each device folder contains:

Calibration/: Calibration files (in JSON and pickle formats) for the respective device.

Register/: Registration experiments for each device. These are divided into multiple folders (e.g., 1, 2, 3, etc.), with the following subdirectories:

FRE/: Transformation evaluation files, including camera parameters, images, point cloud data, and results.

TRE/: Transformation error evaluation files for different registration experiments (e.g., 1, 2, 3, etc.).

Data Description

Calibration Data

Each Calibration folder contains:

calibration.json: Calibration parameters for the respective device.

.pickle: Additional calibration data used during processing.

Registration Data

Each Register folder contains subdirectories:

FRE/: Files for transformation evaluation, including:

cam_params.json: Camera parameters used in registration.

images/: Images used for registration, including RGB captures, depth information, raw stereo images, and the bounding box of the phantom face.

mri.ply: MRI point cloud used for registration.

result_*.json: JSON file containing the results of the registration experiment.

scene_point_cloud_*.ply: Point cloud data for the scene, used to evaluate registration accuracy.

transformation_mri_reference_*.bin: Binary files containing transformation data used for registration.

TRE/: Evaluation files for transformation errors, organized into different subdirectories (1, 2, 3, etc.).

File Naming Convention

Files are named using timestamps for clarity, as shown in the following examples:

registration_experiment_20240715-151618.json: Registration experiment data recorded on July 15, 2024, at 15:16:18.

TRE folders: Numbered subdirectories (1, 2, 3, etc.) represent different transformation error evaluations.

License Information

This dataset is provided under the Creative Commons Attribution 4.0 International License (CC BY 4.0).