OBJECTIVES: In this manuscript, the Robotic Facial Recognition System using the Compound Classifier (RERS-CC) is introduced to improve the recognition rate of human faces. The process is differentiated into classification, detection, and recognition phases that employ principal component analysis based learning. In this learning process, the errors in image processing based on the extracted different features are used for error classification and accuracy improvements.
RESULTS: The performance of the proposed RERS-CC is validated experimentally using the input image dataset in MATLAB tool. The performance results show that the proposed method improves detection and recognition accuracy with fewer errors and processing time.
CONCLUSION: The input image is processed with the knowledge of the features and errors that are observed with different orientations and time instances. With the help of matching dataset and the similarity index verification, the proposed method identifies precise human face with augmented true positives and recognition rate.
METHODS: Semi-structured interviews were conducted among 31 medical doctors in three Malaysian government hospitals on the implementation of the Total Hospital Information System (THIS) between March and May 2015. A thematic qualitative analysis was performed on the resultant data to deduce the relevant themes.
RESULTS: Five themes emerged as the factors influencing workarounds to the HIS: (a) typing skills, (b) system usability, (c) computer resources, (d) workload, and (e) time.
CONCLUSIONS: This study provided the key factors as to why doctors were involved in workarounds during the implementation of the HIS. It is important to understand these factors in order to help mitigate work practices that can pose a threat to patient safety.
PURPOSE: The purpose of this simulation study was to establish a reference percentage value that can be used to effectively reduce the size and polygons of the 3D mesh without drastically affecting the dimensions of the prosthesis itself.
MATERIAL AND METHODS: Fifteen different maxillary palatal defects were simulated on a dental cast and scanned to create 3D casts. Digital bulbs were fabricated from the casts. Conventional bulbs for the defects were fabricated, scanned, and compared with the digital bulb to serve as a control. The polygon parameters of digital bulbs were then reduced by different percentages (75%, 50%, 25%, 10%, 5%, and 1% of the original mesh) which created a total of 105 meshes across 7 mesh groups. The reduced mesh files were compared individually with the original design in an open-source point cloud comparison software program. The parameters of comparison used in this study were Hausdorff distance (HD), Dice similarity coefficient (DSC), and volume.
RESULTS: The reduction in file size was directly proportional to the amount of mesh reduction. There were minute yet insignificant differences in volume (P>.05) across all mesh groups, with significant differences (P
MATERIALS AND METHODS: An auricular prosthesis, a complete denture, and anterior and posterior crowns were constructed using conventional methods and laser scanned to create computerized 3D meshes. The meshes were optimized independently by four computer-aided design software (Meshmixer, Meshlab, Blender, and SculptGL) to 100%, 90%, 75%, 50%, and 25% levels of original file size. Upon optimization, the following parameters were virtually evaluated and compared; mesh vertices, file size, mesh surface area (SA), mesh volume (V), interpoint discrepancies (geometric similarity based on virtual point overlapping), and spatial similarity (volumetric similarity based on shape overlapping). The influence of software and optimization on surface area and volume of each prosthesis was evaluated independently using multiple linear regression.
RESULTS: There were clear observable differences in vertices, file size, surface area, and volume. The choice of software significantly influenced the overall virtual parameters of auricular prosthesis [SA: F(4,15) = 12.93, R2 = 0.67, p < 0.001. V: F(4,15) = 9.33, R2 = 0.64, p < 0.001] and complete denture [SA: F(4,15) = 10.81, R2 = 0.67, p < 0.001. V: F(4,15) = 3.50, R2 = 0.34, p = 0.030] across optimization levels. Interpoint discrepancies were however limited to <0.1mm and volumetric similarity was >97%.
CONCLUSION: Open-source mesh optimization of smaller dental prostheses in this study produced minimal loss of geometric and volumetric details. SculptGL models were most influenced by the amount of optimization performed.
DESIGN: Part 1 involved electroacoustic measurement and biological calibration of a laptop-earphone pair used for the computer-based audiometry (CBA). Part 2 compared CBA thresholds obtained without a sound booth with those measured using the gold-standard clinical audiometry.
STUDY SAMPLE: 17 young normal-hearing volunteers (Part 1) and 43 normal and hearing loss subjects (Part 2) recruited from an audiology clinic via convenience sampling.
RESULTS: The transducer-device combination produced outputs suitable for measuring thresholds down to 0 dB HL. Threshold pairs obtained from the CBA and clinical audiometry were highly correlated (Spearman's correlation coefficient, ρ = 0.92, p 25 dB HL.
CONCLUSIONS: The use of a computer-based audiometer application with consumer insert phone-earmuff combination can offer a cost-effective solution for boothless screening audiometry.