No abstract

We present new pulmonary nodule segmentation algorithms for computed tomography (CT). These include a fully-automated (FA) system, a semi-automated (SA) system, and a hybrid system. Like most traditional systems, the new FA system requires only a single user-supplied cue point. On the other hand, the SA system represents a new algorithm class requiring 8 user-supplied control points. This does increase the burden on the user, but we show that the resulting system is highly robust and can handle a variety of challenging cases. The proposed hybrid system starts with the FA system. If improved segmentation results are needed, the SA system is then deployed. The FA segmentation engine has 2 free parameters, and the SA system has 3. These parameters are adaptively determined for each nodule in a search process guided by a regression neural network (RNN). The RNN uses a number of features computed for each candidate segmentation. We train and test our systems using the new Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) data. To the best of our knowledge, this is one of the first nodule-specific performance benchmarks using the new LIDC-IDRI dataset. We also compare the performance of the proposed methods with several previously reported results on the same data used by those other methods. Our results suggest that the proposed FA system improves upon the state-of-the-art, and the SA system offers a considerable boost over the FA system.

Robot kinematic calibration is the process of enhancing the positioning accuracy of a given manipulator and must be performed after robot manufacture and assembly or during periodical maintenance. This dissertation presents new computationally efficient and robust kinematic calibration algorithms for industrial robots that make use of partial measurements. These include a calibration method that requires the supply of Cartesian coordinates of the calibration points (3DCAL) and another calibration technique that only requires the radial measurements from the calibration points to some reference (1DCAL). Neither method requires orientation measurements nor the explicit knowledge of the whereabout of a reference frame. Contrary to most other similar works, both methods make use of a simplified version of the original Denavit-Hartenberg (DH) kinematic model. The simplified DH(-) model has not only proven to be robust and effective in calibrating industrial manipulators but it is also favored from a computational efficiency viewpoint since it consists of comparatively fewer error parameters. We present a conceptual approach I also would like to thank the director of Yaskawa Motoman Robotics, Inc. Mr. Wade Hickle, Senior Engineer Mr. Eric Marcil and Chief Engineer Mr. George Sutton for their guidance and for making the much needed materials available to me. I appreciate Dr. John Loomis and Dr. Ruihua Liu for participating in my committee. Thanks go to Mr. Tom Schenck of Measurement Specialties, Inc. for allowing me to conduct experiments in his workplace. Finally, I would like to give my special thanks to my dearest friends for all their help and encouragement. vi

No abstract