No abstract

In the STARE project (STructured Analysis of the REtina) we are developing a system that will automatically diagnose images of the ocular fundus, compare sequential images for change, and make clinically significant measurements of lesions and anatomical structures in the images. Ophthalmologists need to compare color images, fluorescein angiograms, indocyanine green angiograms, and scanning laser ophthalmoscopy for onset of disease and changes in lesions. The images are made from fundus cameras of different manufacture and at different magnification. Consequently we designed our system to register automatically images ofdifferent magnification or appearance.We extracted retinal blood vessels from these images with a rotating matched filter. Potential corresponding points (tie points) in each image were identified from the crossing points and branching sites ofthe vascular tree. Features were measured for candidate tie points and these were ranked by similarity. A statistical correlation of image intensity in a region around each point was the main feature used to select good tie points. An affine image transformation was created by a linear least squares method using the top ranked candidate tie points. An iterative method identified and removed possible erroneous tie points. The resulting image transformation was applied to create a common coordinate space for both images. Bright and dark lesions were extracted from both images to obtain objects. Finally, we compared features, such as shape, size, and color, ofthe same object in both images for possible object change detection. INTRODUCTIONThere is a need for comparing images for the onset of disease and for changes in lesions or normal anatomic structures in color images over days to years. The identification of some lesion characteristics in fluorescein angiograms, such as leakage, depends on change in the appearance ofthe lesion in minutes. The test for adequate laser treatment for age-related macular degeneration relies on the comparison of images of completely different appearance, namely, color images and fluorescein angiograms. The images may be made from fundus cameras of different manufacture and at different magnification. Consequently we began with a model-based system of registration that uses the crossing or branching points of retinal blood vessels as tie points in the green channel of color images of the ocular flindus. METHODS Crossing point extractionThere are anatomic features that are present in all images of the ocular fundus. Predominant among these is the retinal blood vessel network. Ophthalmic image types include color images, fluorescein angiograms, scanning laser ophthalmoscopy, and indocyanine green angiograms. The blood vessel 94

Bit and bottomhole assembly (BHA) dynamics are a challenge when drilling in hard and interbedded formations. Dynamic conditions such as stick-slip and whirl can limit performance or in the worst case destroy downhole components. A new percussive mud hammer motor is being implemented in China to improve drilling performance by imparting a controlled impact to the bit while drilling. Conventional technologies to address these challenges may have limited improvement significant added costs: Conventional motors can generate large amounts of torque to the bit but can be difficult to control tool face and depth of cut as the bit transitions through various formations. This can result in stick-slip and bit bounce that can quickly damage or destroy the bit and BHA. Rotary steerable systems can control downhole dynamics with complex electronic, hydraulic and mechanical components but are very expensive, require higher system pressures and on-site supervision. Turbines with impregnated diamond bits can provide smooth drilling but can have limited rates of penetration in the softer portions of interbedded formations. In operation they also require higher pressure drops, have limited build rates and require on-site supervision by a turbine technician. These conventional system limitations created an opportunity for a new technology that is simpler and brings greater value to the drilling operation. The hammer motor is a conventional power section and drivetrain with the patent pending hammer mechanism fitted in place of the lower bearing. This assembly has the same geometry of a bent housing motor with all of its capability plus the impact of a rotary hammer. A series of field trials in China have recently been performed to validate the performance increases created by the hammer motor. These trials demonstrated significant cost savings, improved drilling performance, and simplicity of operation.

Percussive air hammer tools have been used for many years to increase drilling ROP (rate of penentration) in air drilling applications. Similar developments for mud hammer tools have not been as successful. The incompressible nature of drilling mud makes the percussive action much slower to actuate using the same design methodology, rendering the tool ineffective. Other developments have suffered from reliability issues which have limited their drilling hours, therefore making them economically unfeasible. A novel percussive mechanically-actuated Hammer Motor, suitable for either mud or air drilling applications, has changed the landscape. This unique hammer assembly is assembled into a standard mud motor, without affecting the bit to bend distance. The percussive action of the tool is designed such that the bit remains in contact with the formation, while the hammering takes place against the top of the drive mandrel, driving the bit into the formation. The percussive impacts serve to greatly increase the effectiveness of the roller cone bit in crushing the rock, thus significantly increasing ROP. This paper illustrates a case study from Brazil, where the Operator has been using turbine motors or conventional motors to drill vertical wells through hard rock formations. The Hammer Motor displayed significantly higher ROP than the benchmark established by the other motors, while also reducing bit costs. These improvements in drilling performance improve the economics of drilling these hard rock formations, and are also applicable to other drilling applications.