skip to main content
Idioma:

Contrast-enhanced ultrasound in optimization of treatment plans for diabetic nephropathy patients based on deep learning

Sun, Xiaoying ; Lu, Qiaoli

The Journal of supercomputing, 2022-02, Vol.78 (3), p.3539-3560 [Periódico revisado por pares]

New York: Springer US

Texto completo disponível

Citações Citado por
  • Título:
    Contrast-enhanced ultrasound in optimization of treatment plans for diabetic nephropathy patients based on deep learning
  • Autor: Sun, Xiaoying ; Lu, Qiaoli
  • Assuntos: Algorithms ; Artificial intelligence ; Blood flow ; Blood vessels ; Compilers ; Computer Science ; Deep Learning ; Diabetes ; Filtration ; Hemodynamics ; Hemorrhage ; Image quality ; Image reconstruction ; Image transmission ; Interpreters ; Kidneys ; Machine learning ; Medical imaging ; Optimization ; Parallel Computing in Biomed Sciences & Healthcare ; Processor Architectures ; Programming Languages ; Renal function ; Ultrasonic imaging
  • É parte de: The Journal of supercomputing, 2022-02, Vol.78 (3), p.3539-3560
  • Descrição: To evaluate the efficacy of atorvastatin in the treatment of diabetic nephropathy, contrast-enhanced ultrasound (CEUS) based on Poisson three-dimensional (3D) reconstruction algorithm was used. Poisson 3D reconstruction algorithm was optimized based on Gaussian filtering deep learning. Then, the running time, occupied space, and number of triangular blocks of Poisson 3D reconstruction algorithm before and after optimization were analyzed and compared through simulation experiments. One hundred and fifty-six DN patients were divided into an experimental group (conventional treatment + atorvastatin) and a control group (conventional treatment). Ultrasound examinations were performed on all patients before and after treatment, and the artificial intelligence (AI) “Doctor You” system was adopted for image transmission and diagnosis. The kidney volume ( V ) and hemodynamic parameters, including the maximal kidney volume (Vmax), minimal kidney volume (Vmin), and resistance index of all patients were detected and recorded before and after the treatment. The end-point events of patients were tracked. The results showed that the running time of the optimized Poisson 3D reconstruction algorithm was notably shorter, and it occupied more space compared with the pre-optimized Poisson 3D reconstruction algorithm, and the difference was remarkable ( P  < 0.05). The V (136.07 ± 22.16 cm 3 ) in experimental group after the treatment was smaller in contrast to the control group (159.11 ± 31.79 cm 3 ) ( P  < 0.05). The Vmax and Vmin of the renal artery in experimental group after treatment were obviously higher than those in control group, while the RI index of experimental group was relatively lower ( P  < 0.05). The 3D reconstructed image showed that the arterial blood flow was dendritic after treatment, the blood signal was abundant, and the distance among the peripheral blood vessels of the kidney and the renal capsule became small. In addition, the incidence of cerebral hemorrhage, transient cerebral ischemia, and new cerebral infarction in the experimental group was visibly lower than those in control group ( P  < 0.05). In short, the Poisson 3D reconstruction algorithm optimized by Gaussian filtering deep learning technology can not only guarantee the image quality, but also effectively shorten the running time of CEUS reconstruction. In addition, the atorvastatin could effectively improve the renal function of DN patients and could reduce the incidence of clinical cerebrovascular events, showing safe and feasible effect.
  • Editor: New York: Springer US
  • Idioma: Inglês
Voltar para lista de resultados

  • Realização: ABCD-USP USP   Logos de Redes Sociais: Freepik

Buscando em bases de dados remotas. Favor aguardar.