Search
Close this search box.

Quantitate evaluation of photogrammetry with CT scanning for orbital defect – Scientific Reports

The object of study

The study was performed in gypsum model of maxillofacial defects as shown in Fig. 1. The gypsum model was obtained by the conventional method of taking an impression. As the area of interest of the study were intact eye and orbital defects. So, this part of the face is only considered at the time of taking impression conventionally while another part like the area below the nose is excluded. The approval for the study is taken from institutional ethics committee (Institutional ethical committee of Data Meghe Institute of Higher Education and Research with IEC No. DMIMS(DU)/IEC 2022/781 and and registered in Clinical Trial registry of India, CTRI Number CTRI/2022/08/044524 on date 01/08/2022. All methods were carried out in accordance with relevant guidelines and regulations and informed consent was obtained from all subjects and their legal guardian(s).

Figure 1
figure 1

Gypsum model of maxillofacial defect.

Data acquisition and processing

Two different methods that are 3d scanning with white structured light and photogrammetry were used in acquiring the 3d data of the gypsum model.

Zeiss scanner whose was employed as white structured light for creating 3d model. The setup for data acquisition with 3d scanner is shown in Fig. 2, the gypsum model was placed over the turntable, and scans were acquired at 8 angular steps. Once the scans were acquired these scans were aligned automatically in scanner software. The generated 3d models were then further processed into the software for removing noise and closing the holes and the final 3d models were exported in STL format.

Figure 2
figure 2

Zeiss (Comet 3D) scanner setup.

The photogrammetry method is based on structure from a motion algorithm where multiple high images taken from different locations are converted into 3d model (Sapirstein, 2016). Nikon d5300 DSLR, and Google pixel 2xl were used for capturing High-quality 2d images of models,. Nikon 5300 DSLR is dedicated to photography itself and to acquire quality images one needs to have some understanding about operating DSLR whereas the other Google pixel 2xl is smartphone which is easy to use for non-professionals. The setup for obtaining the photographs of the gypsum model is shown in Fig. 3.

Figure 3
figure 3

Setup for photogrammetry data acquisition.

The models were placed on the turntable and a continuous source of light is used for producing an evenly distributed light on the gypsum model. Precaution has been taken to avoid the shadows on the object. The devices for image acquisition were attached to the tripod for reducing the motion blur. To acquire every detail of the defect area of the gypsum model, photographs were taken from three different elevations directing towards the lower, mid, and upper area of the defect, camera positions are shown in Fig. 4.A photograph number calibration study was performed to find out the effect of a number of photos on the quality of details of 3d model obtained. 90p, 60p, and 30p were taken for each gypsum model for photograph calibration study.

Figure 4
figure 4

Camera positions for photogrammetry.

Autodesk Recap photo as photogrammetry software was used for creating the 3d mesh models of each gypsum model. Autodesk recap is web-based software so the user has to upload only images and other procedures will be done by software only and the final 3d model will be available. The generated 3d models were then further processed into the software itself for removing noise and closing the holes. And finally, all the models were exported in high-quality stl format. The reference 3d models were obtained by CT scan method.

Data analysis

All the 3d models generated by different methods described in the previous section were evaluated on 4 parameters which are total time (Acquisition time + Processing time), accuracy, completeness, and resolution.

Acquisition time is defined as the time required to take scans or images. Processing time is considered as the time required to the alignment of the scan to final 3d model creation. The accuracy of all the model was calculated against the gold standard CT scanned model. Accuracy is defined as the surface to the surface distance calculated at each point of 3d mesh model. Surface to surface comparisons were performed in cloud compare opensource software. 3d mesh models of reference 3d model and model generated by other study method were imported in to Cloud Compare (CC) environment. First the two models were aligned manually by 3-point method and then fine registration is done by using ICP option available in CC. Once the 3d model aligned completely then cloud to cloud comparison was done to found the surface deviation. Cloud to mesh method is used for surface comparison as it is robust compared to other. All the further data processing done in Cloud compare software.

Completeness of the 3d models is defined as the percentage number of points below 2 mm. This is also found by using the result obtained in cloud-to mesh comparison.

Resolution is considered as no of polygon per unit area of 3d model generated by each method.