Hierarchical Decomposition of Handwriting Deformation Vector Field for Improving Recognition Accuracy

Author(s):  
Toru Wakahara ◽  
Seiichi Uchida
2010 ◽  
Vol 37 (3) ◽  
pp. 1117-1128 ◽  
Author(s):  
C. Yan ◽  
H. Zhong ◽  
M. Murphy ◽  
E. Weiss ◽  
J. V. Siebers

2019 ◽  
Vol 4 (35) ◽  
pp. 1076
Author(s):  
Alexandros-Stavros Iliopoulos ◽  
Abhishek Dubey ◽  
Xiaobai Sun

2020 ◽  
Vol 47 (7) ◽  
pp. 3000-3012 ◽  
Author(s):  
Xiaokun Huang ◽  
You Zhang ◽  
Liyuan Chen ◽  
Jing Wang

2015 ◽  
Vol 12 (1) ◽  
pp. 4-13 ◽  
Author(s):  
Vitor Mendes Pereira ◽  
Iris Smit-Ockeloen ◽  
Olivier Brina ◽  
Drazenko Babic ◽  
Marcel Breeuwer ◽  
...  

Abstract BACKGROUND Cerebrospinal fluid leakage and ventricular compression during open surgery may lead to brain deformation called brain shift. Brain shift may affect intraoperative navigation that is based on image-based preoperative planning. Tools to correct or predict these anatomic modifications can be important to maintain precision during open guided neurosurgery. OBJECTIVE To obtain a reliable intraoperative volumetric deformation vector field describing brain shift during intracranial neurosurgical procedures. METHODS We acquired preoperative and intraoperative cone-beam computed tomography enhanced with intravenous injection of iodine contrast. These data sets were preprocessed and elastically registered to obtain the volumetric brain shift deformation vector fields. RESULTS We obtained the brain shift deformation vector field in 9 cases. The deformation fields proved to be highly nonlinear, particularly around the ventricles. Interpatient variability was considerable, with a maximum deformation ranging from 8.1 to 26.6 mm and a standard deviation ranging from 0.9 to 4.9 mm. CONCLUSION Contrast-enhanced cone-beam computed tomography provides a feasible technique for intraoperatively determining brain shift deformation vector fields. This technique can be used perioperatively to adjust preoperative planning and coregistration during neurosurgical procedures.


1986 ◽  
Vol 29 (3) ◽  
pp. 420-424 ◽  
Author(s):  
Michael Dorman ◽  
Ingrid Cedar ◽  
Maureen Hannley ◽  
Marjorie Leek ◽  
Julie Mapes Lindholm

Computer synthesized vowels of 50- and 300-ms duration were presented to normal-hearing listeners at a moderate and high sound pressure level (SPL). Presentation at the high SPL resulted in poor recognition accuracy for vowels of a duration (50 ms) shorter than the latency of the acoustic stapedial reflex. Presentation level had no effect on recognition accuracy for vowels of sufficient duration (300 ms) to elicit the reflex. The poor recognition accuracy for the brief, high intensity vowels was significantly improved when the reflex was preactivated. These results demonstrate the importance of the acoustic reflex in extending the dynamic range of the auditory system for speech recognition.


Sign in / Sign up

Export Citation Format

Share Document