Convolutional Neural Networks and Geometric Moments to Identify the Bilateral Symmetric Midplane in Facial Skeletons from CT Scans

Abstract

In reconstructive craniofacial surgery, the bilateral symmetry of the midplane of the facial skeleton plays an important role in surgical planning. Surgeons can take advantage of the intact side of the face as a template for the malformed side by accurately locating the midplane to assist in the preparation of the surgical procedure. However, despite its importance, the location of the midline is still a subjective procedure. The aim of this study was to present a 3D technique using a convolutional neural network and geometric moments to automatically calculate the craniofacial midline symmetry of the facial skeleton from CT scans. To perform this task, a total of 195 skull images were assessed to validate the proposed technique. In the symmetry planes, the technique was found to be reliable and provided good accuracy. However, further investigations to improve the results of asymmetric images may be carried out.

Development of a Convolutional Neural Network Based Skull Segmentation in MRI Using Standard Tesselation Language Models

“…The results found in this article reflect a long-standing search for the development of a technique for bone segmentation in MRI, however, the proposed method DSC (0.7826 ± 0.03) does not exceed the performance of current CT techniques, DSC of 0.9189 ± 0.0162 [40]), DSC of 0.9200 ± 0.0400 [25]), and DSC of 0.9800 ± 0.013 [50]).…”

“…The deep learning framework chosen in this paper was UNet which was introduced by Ronneberger et al [14]. This type of CNN was chosen because it works well with very few training images, yields more precise segmentation, and has been used in a number of recent biomedical image segmentation applications [5,[9][10][11]40]. This network allows for a large number of feature channels in the upsampling procedure, which contribute in the propagation of context information to the highest resolution layers.…”

“…This distinction may be attributed to the size of the dataset used. [40,50] reported a mean dice coefficient of 0.9189 and 0.9800 for CT skull segmentation using CNN when a dataset of 195 and 199 images was used, and [50] attributed this high DSC to the dataset size and image resolution when compared to other study ( [25]). Thus, a change in the size of the dataset may contributed to the improvement of the values of the DSCs.…”

Use of Advanced Artificial Intelligence in Forensic Medicine

“…The methodology chapter presents the novel clinical workflow based on the implementation of three-dimensional convolutional neural network (3D CNN) algorithms [8][9][10]. The input is full head cone-beam computer tomography scans (CBCT) in the Digital Imaging and Communications in Medicine format (DICOM) [10][11][12][13][14][15]. The methodology chapter describes technical data preparation for 3D CNN utilization in the following practical aspects from forensic medicine:…”

“…Automatized 3D cephalometric landmark annotation [44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61] 4. Soft-tissue face prediction from skull and in reverse [62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77][78][79][80][81] 5. Facial growth vectors prediction [14,62,[82][83][84][85][86][87][88][89][90][91][92][93]…”