Search

International audience; BACKGROUND: With the rapid development of deep learning technology, deep neural networks can effectively enhance the performance of computed tomography (CT) reconstructions. One kind of commonly used method to construct CT reconstruction networks is to unroll the conventional iterative reconstruction (IR) methods to convolutional neural networks (CNNs). However, most unrolling methods primarily unroll the fidelity term of IR methods to CNNs, without unrolling the prior terms. The prior terms are always directly replaced by neural networks. PURPOSE: In conventional IR methods, the prior terms play a vital role in improving the visual quality of reconstructed images. Unrolling the hand-crafted prior terms to CNNs may provide a more specialized unrolling approach to further improve the performance of CT reconstruction. In this work, a primal-dual network (PD-Net) was proposed by unrolling both the data fidelity term and the total variation (TV) prior term, which effectively preserves the image edges and textures in the reconstructed images. METHODS: By further deriving the Chambolle-Pock (CP) algorithm instance for CT reconstruction, we discovered that the TV prior updates the reconstructed images with its divergences in each iteration of the solution process. Based on this discovery, CNNs were applied to yield the divergences of the feature maps for the reconstructed image generated in each iteration. Additionally, a loss function was applied to the predicted divergences of the reconstructed image to guarantee that the CNNs’ results were the divergences of the corresponding feature maps in the iteration. In this manner, the proposed CNNs seem to play the same roles in the PD-Net as the TV prior in the IR methods. Thus, the TV prior in the CP algorithm instance can be directly unrolled to CNNs. RESULTS: The datasets from the Low-Dose CT Image and Projection Data and the Piglet dataset were employed to assess the effectiveness of our proposed PD-Net. Compared with conventional CT reconstruction methods, our proposed method effectively preserves the structural and textural information in reference to ground truth. CONCLUSIONS: The experimental results show that our proposed PD-Net framework is feasible for the implementation of CT reconstruction tasks. Owing to the promising results yielded by our proposed neural network, this study is intended to inspire further development of unrolling approaches by enabling the direct unrolling of hand-crafted prior terms to CNNs.

SUMMARY: Palatal rugae is an irregular soft tissue, which is located in the front third of the hard palate, and is asymmetrically distributed from the middle suture to the sides. The difference, stability and extensive characteristics of palatal rugae morphology have gradually make it a characteristic indicator of forensic identification. However, a mature digital palatal rugae identification system has not yet been established at present. Feature extraction is the premise of palatal rugae image recognition. In order to obtain palatal rugae feature information in all directions and improve the reliability of forensic identification, it is necessary to collect palatal rugae images from a plurality of different angles. When the collected images are sent to the recognition system, the diversity of angles will often cause problems such as error recognition. If the tilted images are not rotated properly, it will make the forensic identification face many difficulties. To solve the problem of image skew caused by the diversity of acquisition angle, an algorithm based on orientation vector to correct the tilted palatal rugae images was proposed in this paper. Firstly, the criteria for standard palatal rugae image and the selection rules for feature points were set; Secondly, characterizing feature points according to the rules, and fitting two lines and find their direction vector; Finally, to obtain the corrected images, the tilted images were rotated by the angle determined by the two direction vectors. Simulation results show that the proposed algorithm can correct the tilted palatal rugae images collected from different angles and has strong robustness. RESUMEN: Las rugas palatinas son tejidos blandos irregulares, que se ubican en el tercio frontal del paladar duro y se distribuyen asimétricamente desde la sutura mediana hacia los lados. La diferencia, la estabilidad y las características extensivas de la morfología de las rugas palatinas la han convertido gradualmente en un indicador característico de la identificación forense. Sin embargo, un sistema de identificación de rugas palatinas digitales maduras todavía no se ha establecido en la actualidad. La extracción de características es la premisa del reconocimiento de imágenes de las rugas palatinas. Para obtener información sobre las características de las rugas palatinas en todas las direcciones, y mejorar la confiabilidad de la identificación forense, es necesario recopilar imágenes de las rugas palatinas desde una pluralidad de ángulos diferentes. Cuando las imágenes recogidas se envían al sistema de reconocimiento, la diversidad de ángulos a menudo causará problemas como el reconocimiento de errores. Si las imágenes inclinadas no se giran correctamente, la identificación forense se enfrentará a muchas dificultades. Para resolver el problema del sesgo de la imagen causado por la diversidad del ángulo de adquisición, en este documento se propuso un algoritmo basado en el vector de orientación para corregir las imágenes de las arrugas palatinas inclinadas. En primer lugar, se establecieron los criterios para la imagen de las rugas palatinas estándar, y las reglas de selección para los puntos de características. En segundo lugar, se determinaron puntos de características según las reglas, y se ajustaron dos líneas y encontrar la dirección del vector. Finalmente, para obtener las imágenes corregidas, las imágenes inclinadas se giraron según el ángulo determinado por la dirección de dos vectores. Los resultados de la simulación muestran que el algoritmo propuesto puede corregir las imágenes de rugas palatinas inclinadas recopiladas desde diferentes ángulos y tiene una gran robustez.