3 matic

After CT scanning, image reconstruction was performed to convert the stacked CT slices to a 3D CAD geometry for CFD simulations (Mimics and 3-matic, Materialise, Inc.). The reconstruction process involved the following steps:(i)

Thresholding – From the CT images, the face, respirator, and the air-gap between them were tagged separately based on the variations in pixel intensities (in Hounsfield units).

(ii)

Region growing – The tagged pixels belonging to each of the three regions of interest (face, respirator, and the air-gap) were separated using a ‘region growing’ operation. Thresholding and region growing complete the segmentation portion of the image processing.

(iii)

Reconstruction – Following segmentation, the 2D image was reconstructed into a 3D stereolithographic (STL) volume representing the desired geometry.

(iv)

Post-processing – The geometry in the STL file was subjected to additional wrapping and smoothing operations while ensuring that the gaps and contact points between the face and respirator were unaffected during post-processing. An image of the reconstructed and post-processed 3D CAD model is shown in Fig. 2c.

Subsequently, total gap surface area normal to the airflow direction was quantified using the 3-matic software (Materialise, Inc.)²¹ .

To quantify cartilage growth and mineral length, measurements were taken from 3D representations of the hindlimb knee joints generated through OPT techniques optimized for embryo imaging (79 (link)). Before OPT, samples were dehydrated in ethanol, stained with 0.055% Alcian blue for 5 hours, cleared in 1% KOH for 2 hours, and then stained with 0.01% Alizarin red for 2 hours at room temperature, permitting selective visualization of the cartilaginous tissues and mineral, respectively. Following OPT scans obtained as previously described (79 (link)), image projections were reconstructed (NRecon, Micro Photonics Inc., USA) and segmented into 3D models (Mimics 19, Materialise, Belgium) of the cartilage and mineral. Eight knee joint cartilage features and two mineral lengths of the 3D models were measured (Fig. 2). For joint feature measurements, the distance between the apexes of anatomical landmarks was collected for each model (3-Matic, Materialise, Belgium). To assess joint feature shape, the cartilage joint models were rigidly registered using N-point registration (3-Matic, Materialise) and the contour of the joints extracted in the medial, frontal, and lateral views of the joints. Wilcoxon signed-rank tests with Bonferroni corrections were used to test for paired differences in quantitative cartilage growth variables and mineral length between contralateral limb samples.