Biomedical Image Processing

Biomedical Engineering at KMITL is well recognized for its research in the Biomedical Image and Signal processing area.  In the past decade, approximately ten Ph.D. candidates have graduated in this discipline. Some of the research highlights in Biomedical Image and Signal processing are provided below.

Face Recognition: In this research 3D Gaussian curvature on 3D facial surface was computed and used as geometric feature that is invariance to geometric transformation. The geometric features are the zero-curvature points residing on the parabolic contour (zero-Gaussian curvature contour). The features were arranged in an conformal order before being used to align the two 3D facial data. The error was then computed as used as criterion to identify person.

 

C. Pintavirooj, Fernand  S. Cohen, P. Torsanon, “3D Face Alignment and Registration in the Presence of Facial Expression Differences”, IEEJ transactions on electrical and electronic engineering,  Vol 8. no. 4, 2013

3D Eye Motion: Stereo cameras were installed on the goggle to capture the 2D image of the eye. The 3D coordinate was detected and used to move the 3D model of eyeball corresponding to the patient eyeball movement. The application of the system is used to diagnose patient with dizziness vertigo.

 

*S. Wibirama, S. Tungjitkusolmun,C. Pintavirooj, “Dual-Camera Acquisition for Accurate Measurement of Three-Dimensional Eye Movements”, IEEJ transactions on electrical and electronic engineering,  Vol 8. no. 3, 2012

Ultrasonic Computed Tomography: Our system consists of the ultrasonic transmitter and receiver aligned in the opposite direction in the water tank. The ultrasonic signal was sent traversing the phantom and collected with the receiver. The series of collected signal served as projection data and can be used to reconstruct the cross section image of the phantoms.

 

*A. Sunpanich, K. Hamamoto and C. Pintavirooj, “An Investigation on Attenuation UCT with Wave Paths Enhancement for Breast Ultrasound”, IEEJ transactions on electrical and electronic engineering,  vol. 7, 2013

EMG Classification: A matrix of emg surface electrods were attached on the muscle. The magnitude of EMG from each electrode was collected and mapped on the 2D data. The pattern of 2D data for each muscular contraction is unique and can be classified the type of contraction. The system can be used to control robot arm installed on an amputated patient.

Apnea Diagnostic and treatment System: We developed the diagnostic and treatment tool for patient with apnea. The high-speed computed Tomography was collected from the patient having problem with apnea. The onset of apnea was used as a trigger to start the CT scan. The data was used for 3D model and flow analyze in the naso-laryngeal cavity.

Fingerprint Indentification: The minutae points were extracted from the fingerprint image. The affine-invariant geometric features , after ordered in an comformal order, were used as an landmark to align the two finger images ; one from reference, the other from quary. The error was computed and used to identify person.

*C. Pintavirooj, Fernand  S. Cohen, W. Torsanon, “Fingerprint Verification and Identification Based on Local Geometric Invariants Constructed from Minutiae Points and AugmentedWith Global Directional Filterbank”, IEICE ieice transactions on information and systems