Implementing a deep learning model for automatic tongue tumour segmentation in ex-vivo 3-dimensional ultrasound volumes.

Three-dimensional (3D) ultrasound can assess the margins of resected tongue carcinoma during surgery. Manual segmentation (MS) is time-consuming, labour-intensive, and subject to operator variability. This study aims to investigate use of a 3D deep learning model for fast intraoperative segmentation of tongue carcinoma in 3D ultrasound volumes. Additionally, it investigates the clinical effect of automatic segmentation. A 3D No New U-Net (nnUNet) was trained on 113 manually annotated ultrasound volumes of resected tongue carcinoma. The model was implemented on a mobile workstation and clinically validated on 16 prospectively included tongue carcinoma patients. Different prediction settings were investigated. Automatic segmentations with multiple islands were adjusted by selecting the best-representing island. The final margin status (FMS) based on automatic, semi-automatic, and manual segmentation was computed and compared with the histopathological margin. The standard 3D nnUNet resulted in the best-performing automatic segmentation with a mean (SD) Dice volumetric score of 0.65 (0.30), Dice surface score of 0.73 (0.26), average surface distance of 0.44 (0.61) mm, Hausdorff distance of 6.65 (8.84) mm, and prediction time of 8 seconds. FMS based on automatic segmentation had a low correlation with histopathology (r = 0.12, p = 0.67); MS resulted in a moderate but insignificant correlation with histopathology (r = 0.4, p = 0.12, n = 16). Implementing the 3D nnUNet yielded fast, automatic segmentation of tongue carcinoma in 3D ultrasound volumes. Correlation between FMS and histopathology obtained from these segmentations was lower than the moderate correlation between MS and histopathology.

Comparison of image quality of 3D ultrasound: motorized acquisition versus freehand navigated acquisition, a phantom study.

PURPOSE: Intra-operative assessment of resection margins during oncological surgery is a field that needs improvement. Ultrasound (US) shows the potential to fulfill this need, but this imaging technique is highly operator-dependent. A 3D US image of the whole specimen may remedy the operator dependence. This study aims to compare and evaluate the image quality of 3D US between freehand acquisition (FA) and motorized acquisition (MA). METHODS: Multiple 3D US volumes of a commercial phantom were acquired in motorized and freehand fashion. FA images were collected with electromagnetic navigation. An integrated algorithm reconstructed the FA images. MA images were stacked into a 3D volume. The image quality is evaluated following the metrics: contrast resolution, axial and elevation resolution, axial and elevation distance calibration, stability, inter-operator variability, and intra-operator variability. A linear mixed model determined statistical differences between FA and MA for these metrics. RESULTS: The MA results in a statistically significant lower error of axial distance calibration (p < 0.0001) and higher stability (p < 0.0001) than FA. On the other hand, the FA has a better elevation resolution (p < 0.003) than the MA. CONCLUSION: MA results in better image quality of 3D US than the FA method based on axial distance calibration, stability, and variability. This study suggests acquiring 3D US volumes for intra-operative ex vivo margin assessment in a motorized fashion.