Head and Neck Auto Segmentation Challenge based on Non-Local Generative Models

A new patch based label fusion method based on generative approach is proposed for segmentation of mandible, brainstem, parotid and submandibular glands, optic nerves and the optic chiasm in head and neck CT images. The proposal constructs local classifiers from a dictionary of patches and weights their contribution using a generative probabilistic criterion. Also, a gaussian slide window is used to weight the multiples estimations of neighboring voxels. The proposed method was evaluated on a set of 15 CT images (10 off-site and 5 onsite) provided by the organizers of the Head and neck Auto-Segmentation challenge(MICCAI 2015), where the obtained results are comparable to many of the other methods used in the challenge.

Head and Neck Auto Segmentation Challenge based on Non-Local Generative Models

A new patch based label fusion method based on generative approach is proposed for segmentation of mandible, brainstem, parotid and submandibular glands, optic nerves and the optic chiasm in head and neck CT images. The proposal constructs local classifiers from a dictionary of patches and weights their contribution using a generative probabilistic criterion. Also, a gaussian slide window is used to weight the multiples estimations of neighboring voxels. The proposed method was evaluated on a set of 15 CT images (10 off-site and 5 onsite) provided by the organizers of the Head and neck Auto-Segmentation challenge(MICCAI 2015), where the obtained results are comparable to many of the other methods used in the challenge.

Effective Improvements to Overcome “Blind Spots” in a Corrosion Management Program and Practical Measures to Verify Effectiveness

The recent industry wide post-ILI pipeline ruptures due to external corrosion happened in a relatively short period of time after the ILI using high-resolution Magnetic Flux Leakage (MFL) technology. Failure investigations show that the critical defects that caused these pipeline ruptures are generally long and complex corrosion, which typically consist of a number of deep corrosion pits (i.e. localized metal-loss) within an overall shallower, but relatively large corrosion area (i.e. generalized metal-loss). This has led us to investigate the gaps and “blind spots” of the ILI-based corrosion management program particularly to find out why existing methods fail to effectively identify and remediate such critical defects before they fail. Learning from these post-ILI failures, TransCanada has developed many assessment methods and criteria for identifying challenging areas. The many types of criteria account for blind spots from different perspectives in a multi-faceted manner. The traditional ILI based corrosion management programs calculate a deterministic failure pressure ratio (FPR) and maximum anomaly depth and ensure these do not reach a limiting value. However, this strictly deterministic assessment does not acknowledge the uncertainties, particularly the significant uncertainties in the ILI measurements, assessment models, and material properties. When all uncertainties are accounted for and a probabilistic excavation criterion is used, the excavations reveal that certain anomalies are found to be near-critical in the field even though the deterministic FPR based criteria did not identify these. The probabilistic criteria identifies longer shallower anomalies, with non-critical ILI based FPR values, as anomalies that have a higher probability of exceeding the FPR criteria in-the-ditch (where the uncertainties are minimized). This is because the probabilistic criterion acknowledges that longer anomalies are more sensitive to the depth measurement error and have a higher probability of becoming critical in-the-ditch. This “blind spot” in the deterministic method was overcome by incorporating a probabilistic criterion into the corrosion management program. The effectiveness of these new measures is discussed by examining excavation results of this program and subsequent ILI results. This paper discusses the approach to corrosion management where new learning and knowledge as well as new-found uncertainties are readily accommodated. The approach is also transparent and documented; so that new information can be incorporated into the assessment and post-ILI failures can be prevented more effectively.