Tooth extraction difficulty diagnosis and complications prediction device and method
View Patent ↗The present invention provides a device including a region of interest setting unit that sets a region of interest including a third molar and a periodontal region, a parameter calculation unit that calculates at least one parameter for periodontal disease prognosis evaluation on the basis of image information within the region of interest, a tooth extraction difficulty evaluation unit that evaluates the tooth extraction difficulty for the third molar on the basis of a previously created deep learning algorithm with the parameter as an input, and a complications prediction unit that predicts complications depending on a degree of invasion between the third molar and an IAN, and the parameter calculation unit calculates a parameter among an impaction depth of the third molar, the distance between the third molar and a lower jawbone, an angle of the third molar, and the degree of invasion between the third molar and the IAN.
1 . A tooth image-based tooth extraction difficulty diagnosis and complications prediction device, comprising:
a region of interest setting unit configured to set a region of interest including a third molar and a periodontal region by:
resizing a panoramic radiological image to a resolution of 1056×512 pixels,
applying contrast limited adaptive histogram equalization (CLAHE) to the resized image,
detecting a lower third molar of the lower jaw in the panoramic radiological image by using a RetinaNet detection model having a ResNet-152 backbone,
obtaining a bounding box of the lower third molar from the detection model,
cropping the panoramic radiological image to 700×700 pixels based on the bounding box such that a cropped image includes the lower third molar, teeth adjacent to the lower third molar, and an inferior alveolar nerve (IAN) canal, and
setting the cropped image as the region of interest;
a parameter calculation unit configured to calculate, based on image information within the region of interest, a set of parameters that comprises an impaction depth of the third molar, a distance between the third molar and a lower jawbone, an angle of the third molar, and a degree of invasion between the third molar and the IAN;
a tooth extraction difficulty evaluation unit configured to classify a tooth extraction difficulty of the third molar into one of Vertical Eruption (VE), Soft Tissue Impaction (STI), Partial Bony Impaction (PBI), and Complete Bony Impaction (CBI) by using a hybrid vision transformer model R50+ViT-L/32 as a classifier based on the impaction depth of the third molar, the distance between the third molar and the lower jawbone, and the angle of the third molar; and
a complications prediction unit configured to classify a likelihood of IAN damage into three classes N1, N2, and N3 by using the hybrid vision transformer model R50+ViT-L/32 as a classifier based on the degree of invasion between the third molar and the IAN,
wherein N1 indicates that the lower third molar does not reach an IAN canal in a panoramic radiographic image, N2 indicates that the lower third molar invades one IAN canal line in the panoramic radiographic image, and N3 indicates that the lower third molar invades two IAN canal lines in the panoramic radiographic image.
2 . The tooth extraction difficulty diagnosis and complications prediction device of claim 1 , wherein the tooth extraction difficulty evaluation unit combines the impaction depth of the third molar, the distance between the third molar and the lower jawbone, and the angle of the third molar to evaluate tooth extraction difficulty into one of vertical eruption (VE), soft tissue impaction (STI), partial bony impaction (PBI), and complete bony impaction (CBI).
3 . The tooth extraction difficulty diagnosis and complications prediction device of claim 1 , wherein the complications prediction unit classifies complications prediction results into three classes depending on a degree of invasion between the third molar and the IAN.
4 . The tooth extraction difficulty diagnosis and complications prediction device of claim 1 , further comprising:
a result output unit configured to display the difficulty of the extraction of the third molar, an extraction method, and a level of a likelihood of occurrence of the complications together with performance of detection of the third molar while overlapping these on a third molar image.
5 . A tooth image-based tooth extraction difficulty diagnosis and complications prediction method, comprising:
setting a region of interest of a third molar including a periodontal region by:
resizing a panoramic radiological image to a resolution of 1056×512 pixels,
applying contrast limited adaptive histogram equalization (CLAHE) to the resized image,
detecting a lower third molar of the lower jaw in the panoramic radiological image by using a RetinaNet detection model having a ResNet-152 backbone,
obtaining a bounding box of the lower third molar from the detection model,
cropping the panoramic radiological image to 700×700 pixels based on the bounding box such that a cropped image includes the lower third molar, teeth adjacent to the lower third molar, and an inferior alveolar nerve (IAN) canal, and
setting the cropped image as the region of interest;
calculating, based on image information within the region of interest, a set of parameters that comprises an impaction depth of the third molar, a distance between the third molar and a lower jawbone, an angle of the third molar, and a degree of invasion between the third molar and the IAN;
classifying a tooth extraction difficulty of the third molar into one of Vertical Eruption (VE), Soft Tissue Impaction (STI), Partial Bony Impaction (PBI), and Complete Bony Impaction (CBI) by using a hybrid vision transformer model R50+ViT-L/32 as a classifier based on the impaction depth of the third molar, the distance between the third molar and the lower jawbone, and the angle of the third molar; and
classifying a likelihood of IAN damage into three classes N1, N2, and N3 by using the hybrid vision transformer model R50+ViT-L/32 as a classifier based on the degree of invasion between the third molar and the IAN,
wherein N1 indicates that the lower third molar does not reach an IAN canal in a panoramic radiographic image, N2 indicates that the lower third molar invades one IAN canal line in the panoramic radiographic image, and N3 indicates that the lower third molar invades two IAN canal lines in the panoramic radiographic image.
6 . The tooth extraction difficulty diagnosis and complications prediction method of claim 5 , wherein classifying the tooth extraction difficulty includes combining the impaction depth of the third molar, the distance between the third molar and the lower jawbone, and the angle of the third molar to evaluate tooth extraction difficulty into one of vertical eruption (VE), soft tissue impaction (STI), partial bony impaction (PBI), and complete bony impaction (CBI).
7 . The tooth extraction difficulty diagnosis and complications prediction method of claim 5 , further comprising:
displaying the difficulty of the extraction of the third molar, an extraction method, and a level of a likelihood of occurrence of the complications together with performance of detection of the third molar while overlapping these on a third molar image.