Search

Over the last five years or so, neural network (NN)-based approaches have been steadily gaining performance and popularity for a wide range of optical character recognition (OCR) problems, from isolated digit recognition to handprint recognition. We present an NN classification scheme based on an enhanced multilayer perceptron (MLP) and describe an end-to-end system for form-based handprint OCR applications designed by the National Institute of Standards and Technology (NIST) Visual Image Processing Group. The enhancements to the MLP are based on (i) neuron activations functions that reduce the occurrences of singular Jacobians; (ii) successive regularization to constrain the volume of the weight space; and (iii) Boltzmann pruning to constrain the dimension of the weight space. Performance characterization studies of NN systems evaluated at the first OCR systems conference and the NIST form-based handprint recognition system are also summarized.

We use an analysis of a simple model of recurrent network dynamics to gain qualitative insights into the training dynamics of feedforward multilayer perceptrons (MLPs) used for classification. These insights suggest changes to the training methods used for MLPs that improve network performance significantly. In previous work, the probabilistic neural network (PNN) was shown to provide better zero-reject error performance on character and fingerprint classification problems than radial basis function and MLP-based neural network methods. We will show that performance equal to or better than PNN can be achieved with a single three-layer MLP by making fundamental changes in the network optimization strategy. These changes are: 1) use of neuron activation functions, which reduce the probability of singular Jacobians; 2) use of successive regularization to constrain the volume of the minimized weight space; 3) use of Boltzmann pruning to constrain the dimension of the weight space; 4) use of Prior class probabilities to normalize all error calculations, so that statistically significant samples of rare but important classes can be included without distorting the error surface. All four of these changes are made in the inner loop of a conjugate gradient optimization iteration and are intended to simplify the training dynamics of the optimization. On handprinted digits and fingerprint classification problems these modifications improve error-reject performance by factors between 2 and 4, and reduce network size by 40-60%. Copyright 1997 Elsevier Science Ltd.

From 1981 through 1989, serum samples from 855 white-tailed deer (Odocoileus virginianus) from Ossabaw Island, Georgia (USA), were tested for antibodies to bluetongue virus (BTV) and epizootic hemorrhagic disease virus (EHDV). During this period, prevalence of precipitating antibodies to BTV and EHDV as determined by agar gel immunodiffusion (AGID) tests decreased from 74% to 3% and from 34% to 1%, respectively. Antibodies were detected in serum samples from 0.5-yr-old deer only during 1981, 1982, and 1983, and with few exceptions, positive serological results after 1983 were restricted to older age classes. A decrease in prevalence of precipitating antibodies to BTV and EHDV in age classes exposed during 1981 indicates that AGID results from white-tailed deer populations underestimate the extent of previous exposure to these viruses. Serum neutralization test results from AGID-positive deer indicated that BTV 11 was the principal serotype responsible for infections during 1981. Since 1983, this serotype has been replaced by BTV 13; however, there has been a low level of transmission within the herd. Infection with EHDV 2 appeared most prevalent during 1982; as with BTV 13, there has been limited transmission in this high density deer population since 1983.

From 1981 to 1989, sera were collected from 3,077 white-tailed deer (Odocoileus virginianus) in Georgia and from 1,749 deer from 12 additional states in the southeastern United States. In Georgia, prevalence of precipitating antibodies to epizootic hemorrhagic disease virus (EHDV) and bluetongue virus (BTV), as determined by agar gel immunodiffusion tests, was dependent on physiographic region, age, and year. Overall prevalence of antibodies to EHDV and/or BTV was 11, 33, 48, and 14% for the Mountain, Piedmont, Coastal Plain, and Barrier Island regions, respectively. Results suggested varying patterns of EHDV and BTV activity throughout the state. Serologic results from other southeastern states were consistent with the Georgia sample; prevalence estimates (EHDV and/or BTV) for corresponding physiographic regions deviated by less than 10%. Over this larger geographical area, antibody prevalence in deer appeared to increase with decreasing latitude.

A microneutralization test (MNT) system utilizing cytopathic effect end points was effective in determing neutralization indexes for infectious bronchitis virus antibodies. The system is reproducible within 1 index unit at the 95% level of probability. Comparison of the MNT to tests in eggs resulted in a positive correlation (B =0.81), which was significant (P greater than 0.01). The quantitative dose-response relationship of the MNT is linear (P greater than 0.005), with the 95% prediction limits fitting between one 10-fold dilution.

Severe enteritis attributed to parvovirus infection was diagnosed in raccoons (Procyon lotor) purchased from a commercial animal dealer for interstate translocation. Intestinal lesions included edema, hyperemia, and focal mucosal necrosis. Microscopically, intestinal villi were reduced markedly, and there was necrosis of the glandular crypts. Intranuclear inclusions were seen in many crypt cells. A parvovirus that was infective for Crandell feline kidney cells and primary canine kidney cells was isolated from the spleen of a sick raccoon. The raccoon virus was distinguished from feline panleukopenia virus by the former's ability to hemagglutinate swine erythrocytes at a pH of 7.2. Serum neutralization testing of infected raccoons indicated that the raccoon isolate differed from feline panleukopenia virus and may have been canine parvovirus.

A serological survey of 200 healthy, mature dogs was made to determine Brucella canis infectivity rates. The 9 per cent rate reported in the stray dogs was significantly higher (0.05 less than P greater than 0.025) than the 1 per cent rate found in pet dogs. These rates coupled with a predictably growing stray dog population have patent zoonotic implications.

A 4-year-old mixed-breed spayed bitch was euthanatized after 2 1/2 months of progressive neurologic disease. Prototheca organisms were identified by histopathology, culture, and electron microscopy. Specific fluorescent antibody procedures revealed two species--Prototheca wickerhamii and Prototheca zopfii. Organisms and pyogranulomatous lesions were found in the brain, spinal cord, right eye, kidneys, and heart.

A Brahman x Santa Gertrudis herd of cows bred to a Hereford bull was evaluated because of a 3-year history of several calves born with congenital blindness. Multiple congenital ocular anomalies in 2 calves included microphthalmos, microcornea, microcoria, heterochromia iridis, microlentia, cataracts, retinal dysplasia, retinal detachment, anterior segment dysgenesis, acorea, and proliferation of the anterior neuroectoderm. On the basis of the lack of environmental factors and persistence of an intermittent problem when breeding to a single bull, a genetic defect was diagnosed as the probable cause. Dominant inheritance with varied expressivity may have best explained the lack of obvious signs in the bull, with emergence of various anterior and posterior segment defects in offspring from unrelated cows.

In a serologic survey of stray and pet dog populations of Georgia, serums were screened for Brucella canis antibodies, using the slide agglutination test. If results were positive, B canis antibody titers were determined, using the standard tube agglutination test. The stray dogs had significantly (P less than 0.01) higher titers than did the pet dogs. The reactor rate was 58% higher for the slide agglutination test than for the tube agglutination test. The manufacturer's evaluation of the slide agglutination test was based on a comparison of the serologic results of that test with those of the tube agglutination test, using a comparative method that permitted the results to be interpreted as 99% agreement between the 2 tests. Reevaluation of the manufacturer's data by a different method indicated that the slide agglutination test is very accurate when the results are negative (99.7% specific) but less so when the results are positive (62.5% sensitive).

An enzyme-linked immunosorbent assay (ELISA) was developed for the serodiagnosis of swine toxoplasmosis. The results of the ELISA were interpreted as percentage (% ELISA) and ELISA titer and were compared with results of the dye test, indirect fluorescent antibody test, and indirect hemagglutination test. The results obtained from both the % ELISA and ELISA titer directly correlated with the dye test results. The ELISA is a rapid, sensitive, and readily performed serologic test that should be ideal for examining (screening) swine for toxoplasmosis.

From January, 1973, to December, 1975, the most prevalent microorganisms isolated from ear swab specimens taken from 323 dogs with otitis externa were Staphylococcus aureus, Pseudomonas aeruginosa, and Proteus spp. Comparison of the antibiograms of these isolants to those reported from England in 1961 and from the United States in 1970 suggested emergence of a greater proportion of bacterial strains with resistance to antibacterial agents commonly used to treat otitis externa of dogs.

From December 1983 to December 1984 a study on parasites, diseases and health status was conducted on sympatric populations of sambar deer (Cervus unicolor) and white-tailed deer (Odocoileus virginianus) from St. Vincent Island, Franklin County, Florida. Ten sambar and six white-tailed deer were examined. White-tailed deer had antibodies to epizootic hemorrhagic disease virus and bluetongue virus. Serologic tests for antibodies to the etiologic agents of bovine virus diarrhea, infectious bovine rhinotracheitis, vesicular stomatitis, parainfluenza 3, brucellosis, and leptospirosis were negative in both species of deer. White-tailed deer harbored 19 species of parasites; all were typical of the parasite fauna of this species in coastal regions of the southeastern United States. Sambar deer harbored 13 species of parasites, which apparently were derived largely from white-tailed deer. The only exception was Dermacentor variabilis which occurs frequently on wild swine on the island. The general health status of sambar deer appeared to be better than that of white-tailed deer. This was hypothesized to result from the sambar deer's utilization of food resources unavailable or unacceptable to white-tailed deer and to the absence and/or lower frequency of certain pathogens in sambar deer.

The frequencies of precipitating antibodies to bluetongue virus (BTV) and epizootic hemorrhagic disease virus (EHDV) in domestic ruminants and white-tailed deer (WTD) in Georgia were 36% and 32%, respectively (n = 2,200). The frequencies of seropositivity to BTV and EHDV were high among cattle (47% and 42%, respectively [n = 1,068]) and less so in WTD (36% and 34% [n = 414]). The frequencies among sheep were 34% for BTV and 29% for EHDV (n = 286), whereas among goats, seropositivity was 8% for BTV and 7% for EHDV (n = 433). Serum samples from northeastern Georgia (1 of the 4 regions in the survey) had the highest frequency of precipitating antibodies for BTV (45%) and EHDV (38%). The lowest frequency was in southeastern Georgia, with 29% seropositivity for BTV and 24% seropositivity for EHDV. Of the 175 farms or herds in the serosurvey, 70% included animals that had BTV-precipitating antibodies, and 67% included animals which had EHDV-precipitating antibodies. Seventeen viral isolates were obtained from individual animals on 9 different farms. Fifteen of the isolates were BTV--8 from cattle, 4 from sheep, and 3 from WTD; 8 of them were serotype 11, and 7 were serotype 17. Viral isolates from each of 2 WTD were identified as EHDV serotype 1 and serotype 2. Of the total 17 isolates, 11 were from clinically healthy ruminants, and 6 were from animals with clinical signs of BT or EHD. Five of the viral isolates originated from northeastern Georgia, 7 from the northwestern region, and 5 from the southwestern region; none was obtained from specimens from the southeastern region.

Experimentally induced lesions of cutaneous leishmaniasis and the effect of concurrent bacterial infection on the development of these lesions were studied in the golden hamster. Male outbred golden hamsters received intradermal injections at the base of the tail with approximately 10(7) promastigotes of Leishmania braziliensis panamensis, or promastigotes combined with Staphylococcus aureus or Pasteurella multocida or both, bacteria only, or sterile Eagle's minimal essential medium (MEME). The size of the resulting lesions was measured at least twice each week. Hamsters were killed at postinoculation Days 6, 13, 20, 27, 41, or 48, and each lesion was measured, aseptically excised, and bisected; half was used for bacteriologic culture and the other half was prepared for light microscopic examination. Lesions resulting from L. b. panamensis alone progressed from initial erythema to a granulomatous nodule and finally to a necrotic granuloma, often capped by a crateriform ulcer. Lesions resulting from a suspension of L. b. panamensis with added S. aureus or S. aureus and P. multocida, were initially larger, more erythemic and contained a greater proportion of neutrophils up to postinoculation Days 14-21 than did lesions resulting from L. b. panamensis alone. Concurrent infections with bacteria such as S. aureus and P. multocida had little effect on the development of ulcerating characteristics of lesions, but when S. aureus was present it appeared to enhance the severity of the early lesions. Between postinoculation Days 14-28, lesions produced by L. b. panamensis, with or without added bacteria had similar developmental progression of sufficient size for optimal testing of antileishmanial compounds.

The most prevalent microorganisms isolated from urine specimens of dogs (385) and cats (132) with clinical signs of urinary tract infections were Escherichia coli, Proteus spp., and Staphylococcus aureus. The results of quantitative urine-culturing methods showed 48.6% of the canine and 12.1% of the feline specimens had more than 10(5) organisms per ml of urine. The bacteria isolated appear to have a greater resistance to antibacterial agents than previously reported.

Virus was isolated from semen and fecal samples from a bull with orchitis, testicular degeneration, aspermatogenesis, and loss of libido. Both isolates were classified as picornavirus, bovine enterovirus serotype I, on the basis of physical, chemical, and serologic characteristics. Veterinary practitioners that may suspect viral infection as a cause of bovine infertility should submit both semen and fecal samples for virus isolation and identification.

Each of thirteen 6-week-old pigs was inoculated per os with 10,000 sporulated oocysts of Toxoplasma gondii. By postinoculation day (PID) 13, pigs were seropositive by the indirect fluorescent antibody test. Beginning on PID 13 and every 7 days thereafter through PID 97, 1 pig was killed and 6 tissues were examined for T gondii. Of the 13 pigs, 11 were infected, including the 1st pig killed on PID 13, although none of the pigs had gross lesions of toxoplasmosis. Tissues harboring T gondii most frequently were the heart and brain; organisms were detected less frequently in the longissimus muscles, diaphragm, and liver. Toxoplasma gondii was not detected in the bronchial lymph nodes. There was good correlation between antibody and presence of T gondii in these pigs. One additional pig, maintained as a noninfected control, remained seronegative and had no evidence of infection when killed on PID 97.

Swab specimens obtained from skin lesions of 45 cats were cultured bacteriologically for staphylococci. Thirty-two staphylococcal isolates were recovered from 30 cats and were biotyped, using biochemical tests contained in a staphylococcal identification system. Of 23 isolates considered coagulase-positive, 16 were identified as Staphylococcus aureus, 5 as S intermedius, and 2 as S hyicus. Of 9 isolates considered coagulase-negative, 6 were identified as S simulans, 2 as S epidermidis, and 1 as S xylosus. Antimicrobial susceptibility tests were done on all staphylococcal isolates, using a disk-diffusion method. Staphylococcal isolates were susceptible to clavulanic acid-amoxicillin, cloxacillin, cephalothin, chloramphenicol, gentamicin, erythromycin, and trimethoprim-sulfamethoxazole. Resistance to penicillin G, ampicillin, and tetracycline was frequent.

Virus titers from feline herpesvirus-infected feline tracheal organ cultures were higher than those from feline kidney cell cultures. The time for the development of peak viral yields was longer in organ culture as compared to cell cultures. Feline tracheal organ cultures infected with feline calicivirus produced minimal viral titers.

In August 1983, a study on parasites, diseases, and health status was conducted on sympatric populations of fallow deer (Dama dama) and white-tailed deer (Odocoileus virginianus) from Land Between The Lakes, Lyon and Trigg counties, Kentucky. Five adult deer of each species were studied. White-tailed deer had antibodies to epizootic hemorrhagic disease (EHD) virus and Leptospira interogans serovariety icterohemorrhagiae, and fallow deer had antibodies to bluetongue and EHD viruses. Serologic tests for bovine virus diarrhea virus, infectious bovine rhinotracheitis virus, parainfluenza3 virus, and Brucella spp. were negative. One white-tailed deer had an infectious cutaneous fibroma, and one fallow deer had pulmonary mucormycosis. White-tailed deer harbored 16 species of parasites, all of which are considered typical of the parasite fauna of this host in the southeastern United States. Fallow deer harbored nine species of parasites, including eight species known to occur in white-tailed deer on the area and one species (Spiculopteragia assymmetrica) that is not. All fallow deer had inflammatory lesions in the spinal cord and/or brain that were attributed to prior infection with meningeal worm (Parelaphostrongylus tenuis), indicating that P. tenuis infections are not always fatal for this species. The apparent high rate of exposure of Land Between The Lakes fallow deer to P. tenuis without a resultant high rate of clinical cerebrospinal parelaphostrongylosis is hypothesized to be due to a low prevalence and intensity of P. tenuis, partial innate resistance of fallow deer, and acquired immunity.

Pseudohyperchloremia and a negative anion gap were detected in goats with bromide intoxication. Bromide interferes with ion-specific electrodes, resulting in a falsely increased serum chloride concentrations. Bromide intoxication should be considered in animals with progressive neurologic signs, a high serum chloride value, and a low or negative anion gap.

A 3-year-old Limousin cow was admitted to the University of Georgia Teaching Hospital with a history of chronic weight loss and diarrhea of more than 1 year's duration. Clinical laboratory findings included proteinuria and hypoproteinemia and were suggestive of a nephrotic-like syndrome. Necropsy findings included generalized edema of the visceral organs and diffuse red glistening foci on the capsular and cut surfaces of the cortex of both kidneys. Microscopically, the kidney had hypercellular and hypersegmented glomeruli, focal glomerulosclerosis, periglomerular fibrosis, and tubular nephrosis. The glomerular hypercellularity was confirmed by electron microscopy to be due to increased mesangial cells.

The in-vitro effect of EDTA-Tris-lysozyme solution on 16 pathogenic bacteria of medical or veterinary importance was determined. Marked decreases in bacterial count occurred with Pseudomonas aeruginosa, Escherichia coli, Moraxella osloensis and Campylobacter fetus, and smaller decreses with Salmonella typhimurium, Shigella boydii, Aeromonas hydrophila, proteus mirabilis, Listeria monocytogenes and Erysipelothrix insidiosa. The test solution had no effect on Klebsiella ozaenae, Brucella canis, Cornynebacterium pyogenes, Coryne, renale, Streptococcus equi and staphylococcus aureus.

Myospherulosis was diagnosed in a mature pony. Several parent bodies containing many spherules were observed microscopically in biopsy material from an area of cellulitis. The spherules are altered red blood cells that form as the result of prolonged contact with necrotic fat or petrolatum-based ointments. These structures must be differentiated from fungal elements. The recommended treatment is surgical excision.

The purpose of this study was to compare virulence factors and antibiotic resistance profiles of Escherichia coli strains isolated from dogs and humans with urinary tract infections. Factors studied included resistance to antibiotics and the transferability of R-plasmids to a recipient E. coli; production of colicins, hemolysins, beta-lactamase, and urease; hemagglutination of erythrocytes; and fermentation of dulcitol. The canine E. coli isolates had a wider range of antibiotic resistance and a higher R-plasmid transmissibility rate. A higher percentage of the canine isolates produced colicins (40% vs. 24%), hemolysins (44% vs. 16%), beta-lactamase (52% vs. 4%), and fermented dulcitol (84% vs. 80%) as compared with the human isolates. The human isolates had a greater ability to hemagglutinate erythrocytes as compared with the canine isolates (24% vs. 8%). None of the isolates produced urease.

An attempt was made to reverse the antibiotic resistance of Proteus mirabilis used to induce experimental cystitis in cats and dogs. Results of in vitro studies on an oxytetracycline-resistant strain of P mirabilis, utilizing standard plate counts, indicated the organism became susceptible to oxytetracycline when it was treated with a solution of EDTA-tromethamine. In vivo experiments were conducted with cats and dogs in which 50% of the experimentally infected animals were treated with bladder lavage twice daily with EDTA-tromethamine solution and 50% with tromethamine buffer alone; all were given standard doses of oxytetracycline. Results of bacteriologic cultural examinations of urinary samples and urinalyses done each day, at the end of the experiments, indicated all animals had cystitis. Thus, although the organism's resistance to oxytetracycline was changed by EDTA-tromethamine in vitro, change in bacterial resistance did not occur in vivo in cats and dogs during clinical treatment.

A protruding nasal mass in a domestic shorthair cat with nasal discharge and recurrent fever was determined to be caused by infection with the fungus Trichosporon pullulans, as determined by cytologic, histologic, and fungal culture methods. Initially, the cat was treated orally with ketoconazole and the mass decreased in size. When signs of hepatotoxicosis developed, treatment was reduced to an alternate-day basis. However, the nasal mass increased in size with this regimen, and therapy with amphotericin B and 5-flucytosine was begun. It also was unsuccessful.