Abstract:3D volume construction of the maxillary sinus is important for understanding the 3D surface morphology of the maxillary sinuses and detecting changes or obstruction in sinuses. It is important to detect the pathological conditions affecting the sinuses and to determine the treatment outcomes. The cases of sinusitis and various other pathoses in maxillary sinuses are getting comparatively higher than in other sinuses. Therefore, analysis of maxillary sinus structure has some clinical importance. 3D volume construction from Computed Tomography (CT) can help in the volumetric study of the maxillary sinus. Nonetheless, CT imaging is expensive and has a higher radiation dose than X-ray imaging. Thus, the 3D construction from X-ray images is a challenging problem. This paper proposed a 3D construction method for maxillary sinus from lateral and Posterior-anterior (PA) 2D X-ray images. A novel 3D maxillary sinus construction model (2D-to-3D-MaxiDeform) is proposed which evolves the shape and size of the maxillary sinus providing the required volume, and other linear measurements in mesh format. The 2D-to-3DMaxiDeform uses a 3D point cloud template of the maxillary sinus and applies for registration on the template model. The deformation moves the 3D template points to the required positions of the X-ray image contours. The registration of the template model before deformation into the required shape makes sure the size, position, and angle of the template model are mapped with the input X-ray contours. The evaluation of the 2D-to-3DMaxiDeform achieves an average accuracy of 0.83, mDSC (mean Dice similarity coefficient) of 0.80, mIoU (mean Intersection over Union) of 0.67, recall of 0.90, precision of 0.74, specificity of 0.78 and RMSE of 2.3 mm. The proposed method is the first and novel approach for maxillary sinus 3D construction from X-ray images. The result shows that the proposed method could be a valuable tool for generating 3D models of the maxillary sinus to be used in a clinical setting.

Reconstructing Sinus Anatomy from Endoscopic Video -- Towards a Radiation-free Approach for Quantitative Longitudinal Assessment

A Quantitative Evaluation of Dense 3D Reconstruction of Sinus Anatomy from Monocular Endoscopic Video

Automatic semantic segmentation of the osseous structures of the paranasal sinuses

Self-supervised Dense 3D Reconstruction from Monocular Endoscopic Video

Construction and spatial directions analysis of digitized visible model of frontal sinus drainage pathway

Surface Reconstruction from Tracked Endoscopic Video Using the Structure from Motion Approach.

Automatic 3D reconstruction of an anatomically correct upper airway from endoscopic long range OCT images

EndoPerfect: A Hybrid NeRF-Stereo Vision Approach Pioneering Monocular Depth Estimation and 3D Reconstruction in Endoscopy

Machine learning framework for simulation of artifacts in paranasal sinuses diagnosis using CT images

2D-to-3DMaxiDeform: A computer-aided approach for 3D construction of maxillary sinus from PA and lateral X-ray images

EndoSurf: Neural Surface Reconstruction of Deformable Tissues with Stereo Endoscope Videos

Pilot Study of Applying Digital Visible Models for Endoscopic Nasal Surgery

Online 3D reconstruction and dense tracking in endoscopic videos

A 3D Reconstruction Method for Gastroscopic Minimally Invasive Surgery.

Machine Vision for Real-Time Intraoperative Anatomic Guidance: A Proof-of-Concept Study in Endoscopic Pituitary Surgery

A Review of 3D Reconstruction Techniques for Deformable Tissues in Robotic Surgery

Development of a Machine Learning-Enabled Virtual Reality Tool for Preoperative Planning of Functional Endoscopic Sinus Surgery

SurgicalGaussian: Deformable 3D Gaussians for High-Fidelity Surgical Scene Reconstruction

Computer-aided superimposition of the frontal sinus via 3D reconstruction for comparative forensic identification

EndoSparse: Real-Time Sparse View Synthesis of Endoscopic Scenes using Gaussian Splatting

Endoscopic navigation in the absence of CT imaging