Robust Algorithms for Object Localization

Model-based localization, the task of estimating an object's pose from sensed and corresponding model features, is a fundamental task in machine vision. Exact constant time localization algorithms have been developed for the case where the sensed features and the model features are the same type. Still, it is not uncommon for the sensed features and the model features to be of different types, i.e., sensed data points may correspond to model faces or edges. Previous localization approaches have handled different model and sensed features of different types via sampling and synthesizing virtual features to reduce the problem of matching features of dissimilar types to the problem of matching features of similar types. Unfortunately, these approaches may be suboptimal because they introduce artificial errors. Other localization approaches have reformulated object localization as a nonlinear least squares problem where the error is between the sensed data and model features in image coordinates (the Euclidean image error metric). Unfortunately, all of the previous approaches which minimized the Euclidean image error metric relied on gradient descent methods to find the global minima, and gradient descent methods may suffer from problems of local minima. In this paper, we describe an exact, efficient solution to the nonlinear least squares minimization problem based upon resultants, linear algebra, and numerical techniques. On a SPARC 20, our localization algorithm runs in a few microseconds for rectilinear polygonal models, a few milliseconds for generic polygonal models, and one second for generalized polygonal models (models composed of linear edges and circular arcs).     

Colon cancer cell vaccine prepared with replication-deficient vaccinia viruses encoding B7.1 and interleukin-2 induce antitumor response in syngeneic mice

The objective of this study was to evaluate the clinical efficacy and cost effectiveness of inpatient and outpatient laparoscopic lumbar diskectomy (LLD) compared with laminectomy (LAM) in the surgical treatment of disabling L5-S1 disk herniation. Sixty-two adults underwent surgery for herniated L5-S1 intervertebral disks (31 LLD and 31 LAM). Operative blood loss (EBL) (milliliters), operative time (ORT) (minutes), hospital stay (LOS), and rehabilitation time to normal activity (REHAB) (days), recurrent symptoms, postoperative morbidity, percent pain free, and hospital patient charges were calculated. Thirty LLD patients (97%) had immediate relief of disk pain. Morbidity after LLD included transient urinary retention (one) and rectus hematoma (one). One LAM patient had a pseudomeningocele. Among patients observed for > or =6 months, with a median follow up time of 34 months, 22 of 25 LLD patients (88%) returned to normal activity, while 12 of the LAM group (52%) were disabled (p = 0.004). Functional outcome was improved by LLD for workers compensation patients followed > or =6 months, with 86% LAM disabled, vs. 10% LLD (p = 0.001). Sixteen LLD patients (52%) and 18 (58%) of the LAM group needed postoperative physical therapy. Four LLD patients recurred; three required reoperation. Four LAM patients had surgery for recurrent disk herniation. ORT was longer for LLD than LAM (210 vs. 158 minutes, median, p < 0.05). EBL and REHAB time were significantly reduced with LLD, vs. LAM. With a median follow-up of 34 months, 58% of LLD and 39% of LAM patients followed > or =6 months were pain free. Outpatient LLD (n = 9) reduced LOS (1 day vs. 2 days and 4 days, p < 0.01) and lowered patient charges ($4,405 vs. $5,723 and $7,192, p < 0.01) compared with inpatient LLD (n = 23) and LAM, respectively. LLD is a safe, cost-effective, minimally invasive alternative to LAM for treating herniated L5-S1 disks. Compared with LAM, LLD reduces EBL, LOS, REHAB time, and patient charges, improves function, and increases long-term pain relief. Cost effectiveness is optimized when LLD is performed as outpatient surgery.     

The kinesin-related proteins, Kip2p and Kip3p, function differently in nuclear migration in yeast

The roles of two kinesin-related proteins, Kip2p and Kip3p, in microtubule function and nuclear migration were investigated. Deletion of either gene resulted in nuclear migration defects similar to those described for dynein and kar9 mutants. By indirect immunofluorescence, the cytoplasmic microtubules in kip2Delta were consistently short or absent throughout the cell cycle. In contrast, in kip3Delta strains, the cytoplasmic microtubules were significantly longer than wild type at telophase. Furthermore, in the kip3Delta cells with nuclear positioning defects, the cytoplasmic microtubules were misoriented and failed to extend into the bud. Localization studies found Kip2p exclusively on cytoplasmic microtubules throughout the cell cycle, whereas GFP-Kip3p localized to both spindle and cytoplasmic microtubules. Genetic analysis demonstrated that the kip2Delta kar9Delta double mutants were synthetically lethal, whereas kip3Delta kar9Delta double mutants were viable. Conversely, kip3Delta dhc1Delta double mutants were synthetically lethal, whereas kip2Delta dhc1Delta double mutants were viable. We suggest that the kinesin-related proteins, Kip2p and Kip3p, function in nuclear migration and that they do so by different mechanisms. We propose that Kip2p stabilizes microtubules and is required as part of the dynein-mediated pathway in nuclear migration. Furthermore, we propose that Kip3p functions, in part, by depolymerizing microtubules and is required for the Kar9p-dependent orientation of the cytoplasmic microtubules.     

A Comparison of Inspector Performance Measures

Using experimental data from their signal-detection study, the authors examine traditional measures of inspector performance along with measures based on Bayes'rule. Results indicate that the inspector's decision process is affected by the a priori probability of nonconforming product in the inspection lot and that inspector decision error may result in a tighter OC curve for a sampling plan. The conclusions are that performance measures should be selected relative to performance evaluation objectives and that graphs describing acceptance sampling plans for ideal inspectors should be modified to reflect performance of real inspectors under actual conditions.     

The Utility of Signal-Detection Theory in the Analysis of Industrial Inspector Accuracy

In an experimental study of inspector accuracy utilizing industrial electronics inspectors as subjects, a visual, subject-paced task requiring a “yes-no” response was used. The results are thought to be generally applicable to visual inspection tasks, but may not be applicable in an inspection task requiring measurement, as in machine parts inspection. The major conclusion of the research is that signal-detection measures are more useful in performance evaluation than are other available measures, because they not only relate performance to payoff, but indicate the magnitude and direction of improvement required.     

Planning for Modular and Hybrid Fixtures

Fixturing encompasses the design and assembly of fixtures to locate and hold a workpiece during a manufacturing operation such as machining or assembly. We have implemented an automated design algorithm for a fixturing toolkit called the fixture vise; the fixture vise involves two fixture plates mounted on jaws of a vise and modular fixturing elements (pegs or flatted pegs). Generally, a fixture vise can handle appropriately sized prismatic workpieces in many different ways. The design algorithm runs in O(A) time, where A is the number of configurations achieving simultaneous contact between the modeled object and four fixture elements; since simultaneous contact is a necessary but not sufficient precondition for force closure, A provides an upper bound on the number of force closure fixture configurations. Received June 6, 1994; revised November 17, 1994, February 1, 1995, February 28, 1995, and July 11, 1995.     

Use of epidural anesthesia and spontaneous ventilation during transabdominal colon and rectal procedures in selected high-risk patient groups

Fibrodysplasia ossificans progressiva (FOP) is a rare genetic disorder characterized by congenital malformation of the great toes and progressive heterotopic ossification in distinct anatomic patterns. Early preosseous lesions in FOP are clinically and histologically indistinguishable from the lesions of aggressive juvenile fibromatosis (AJF). Although the genetic defect in FOP is unknown, bone morphogenetic proteins (BMPs) 2 and 4 are plausible candidates genes. To determine if there is a difference in BMP 2/4 expression in the early fibromatous lesions of the two conditions, we performed immunohistochemical studies with a monoclonal antibody to BMP 2/4 on the earliest detectable fibromatous lesions of FOP and compared them with histologically identical lesions resected from children who had AJF. Fibromatous cells from the early FOP lesions exhibited immunostaining for BMP 2/4, whereas histologically indistinguishable fibromatous cells from AJF lesions showed no evidence of BMP 2/4 immunostaining. It is incumbent on all physicians who treat patients with suspected fibromatosis to examine the toes to rule out FOP and to avoid unnecessary diagnostic biopsies because surgical trauma induces further bone formation in patients who have FOP. However, if diagnostic confusion still exists and a biopsy is performed, immunostaining with BMP 2/4 antibody may resolve the diagnostic dilemma between FOP and AJF before the appearance of heterotopic ossification is observed in the FOP lesions. Our data suggest that the BMP 2/4 subfamily of secreted proteins may be involved in the pathogenesis of the FOP lesions.     

Surgical adjuvant active specific immunotherapy for patients with stage III melanoma: the final analysis of data from a phase III, randomized, double-blind, multicenter vaccinia melanoma oncolysate trial

BACKGROUND: A phase III, randomized, double-blind, multicenter trial of active specific immunotherapy (ASI) using vaccinia melanoma oncolysate (VMO) was performed in patients with stage III (American Joint Commission on Cancer) melanoma to determine the efficacy of VMO to increase the disease-free interval (DFI) or overall survival (OS) in these patients. Two interim analyses of data from this trial were performed in May 1994 and June 1995. Although the results from these analyses showed no statistically significant improvement in DFI or OS in all patients using VMO, two subsets-men aged 44-57 years with one to five positive nodes and all patients with clinical stage I and pathologic stage II disease-showed an overall survival advantage with VMO therapy. A final analysis of data from this trial was performed in May 1996 and is reported here. The design of future melanoma vaccine trials is discussed based on information learned from this first randomized, multicenter trial of ASI therapy. STUDY DESIGN: A polyvalent VMO was prepared using melanoma cells derived from four melanoma cell lines and vaccinia vaccine virus (V). Patients were accrued from 11 United States institutions and were randomized by the Statistical Center at the University of Alabama, Birmingham. Two hundred fifty patients were randomized to treatment with either VMO (1 U containing 2 mg of total protein derived from 5 x 10(6) melanoma cells and 10(5.6) 50% tissue culture infectious dose of vaccinia virus) or control V (1 U containing 10(5.4) 50% tissue culture infectious dose of vaccinia virus) once a week for 13 weeks and then once every 2 weeks for a total of 12 months, or until recurrence. Patient data were collected by the Statistical Center and analyzed as of May 1996 for DFI and OS using Wilcoxon test and log-rank analysis. RESULTS: Two hundred seventeen patients were found to be eligible according to the inclusion criteria. Data from these patients were analyzed for DFI and OS after a median followup of 46.3 months (50.2 months for VMO and 41.3 months for V). This final analysis showed no statistically significant increase in either DFI (p = 0.61) or OS (p = 0.79) of patients treated with VMO (n = 104) compared with V (n = 113). At 2-, 3-, and 5-year intervals, 47.8%, 43.8%, and 41.7% of patients treated with VMO were disease-free, respectively, compared with 51.2%, 44.8%, and 40.4% of patients treated with V. At the same intervals, 70.0%, 60.0%, and 48.6% of patients treated with VMO survived, compared with 65.4%, 55.6%, and 48.2% of patients treated with V. In a retrospective subset analysis, male patients aged 44-57 years (n = 20) with one to five positive nodes showed 18.9%, 26.82%, and 21.3% improvement in survival at 2-, 3-, and 5-year intervals, respectively, after treatment with VMO when compared with V (n = 18) (p = 0.046). CONCLUSIONS: This study was a randomized, multicenter, placebo-controlled evaluation of an active specific immunotherapeutic agent to increase the DFI or OS of patients with stage III melanoma in a surgical adjuvant setting. In this trial, ASI with VMO when compared with V showed no difference in either DFI or OS. In a retrospective subset analysis, however, a subset of men with one to five positive nodes, between the ages of 44 and 57 years, showed a survival advantage with VMO. This result suggests that one must include a detailed subset analysis in the design of future trials of ASI for patients with American Joint Commission on Cancer stage III melanoma. An appropriate control arm also must be included in ASI trials.     

Psychiatric comorbidity among hospitalized AIDS patients vs. non-AIDS patients referred for psychiatric consultation

This study evaluated the clinical performance of a new resin modified glass ionomer cement, Geristore (Den-Mat Corp., Santa Maria, Calif.), for the bonding of orthodontic brackets and its effect on certain caries-associated microorganisms. This cement has been shown to possess increased mechanical properties and long-term fluoride release. There were 716 brackets bonded in 40 patients (17 males and 23 females), with a split-mouth technique and a composite resin, Phase II (Reliance, Itasca, III.), as a control. Bond failures were recorded up to 1 year. Plaque scores and plaque samples were taken from the area of the bonding adhesive in 20 patients, before, at 1 week, and 5 months after the placement of brackets. The plaque samples were investigated for the presence of Streptococcus mutans and lactobacilli. The overall bond failure rate was found to be 8.9% for Geristore and 3.1% for Phase II (p < 0.05). Labially, there was no significant difference (p > 0.05) in bond failure rate: 3.8% for Geristore and 1.7% for Phase II. The proportions of S. mutans and lactobacilli in plaque taken from around Geristore cement were reduced at 1 week and 5 months, when compared with Phase II resin, and this reduction was statistically significant (p < 0.05) at 1 week. Results of this study suggest that Geristore may be of use in the labial segments, especially in caries prone patients, in whom demineralization at debond may present an esthetic and restorative problem several years after treatment.