SEARCH · PubMed Health

Results for “computational morphology”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 595 records · Page 33Linked to original sources

PubMed1999

Intra- and inter-laboratory variability in the assessment of sperm morphology by strict criteria: impact of semen preparation, staining techniques and manual versus computerized analysis.

We designed prospective studies to compare manual and computerized analysis of sperm morphology by strict criteria using different semen processing and staining techniques. A total of 54 semen samples were studied; slides were prepared from each subject from liquefied semen and after washing, and stained with Diff-Quik or Papanicolaou. An intra-laboratory, blind assessment was performed manually (two observers) and using a computerized analyser (two readings). This demonstrated a very good correlation between manual analysis of liquefied and washed samples with both staining techniques [intraclass coefficient (ICC) = 0.93 and 0.83]. Greater agreement was observed between computerized readings (washed samples) of Diff-Quik (ICC = 0.93) than of Papanicolaou-stained slides (ICC = 0.66). An excellent intra-laboratory correlation was observed for within-computer readings (ICC = 0.93). There was moderate agreement between inter-laboratory computer readings (two centres, ICC = 0.72). Although there was lower inter-laboratory agreement for manual and manual versus computer readings, overall results of all manual and computer analyses showed good agreement (ICC = 0.73). Diff-Quik staining is reliable for both manual (liquefied) and computer (washed) analysis of strict sperm morphology. Intra- and inter-computer analyses using this method reached satisfactory levels of agreement. There is still high inter-laboratory variability for the manual method.

Adult↗

PubMed2000

[Study on statistical method of distribution for erythrocyte morphological features by computerized image processing].

This study sought to develop a new statistic method for the semiautomatic analysis and classification of erythrocyte morphology based on the morphological features and shape analysis of erythrocytes by computer image processing. Shape factor as the description of the erythrocyte morphological features was used for the shape classification of erythrocytes. And the models and algorithms of erythrocytes image segmentation, cell body shape recognition and measure can be implemented through the VC++. The present method can efficiently and semi-automatically provide the statistical analysis of erythrocyte morphology, and can give the distribution of erythrocyte morphological features. The result showed that there was a significant difference between the distribution curves of the normal erythrocyte morphology (one apex) and hemolytic anemia's (two apices). By this way, it can be obtained the proportion data of different cell bodies' shapes. This method could provide some information for the study and diagnosis of the diseases (e.g. hemolytic anemia, pre-leukemia) related to erythrocyte morphology.

Algorithms↗

PubMed1999

Direct effect of alcohol on the motility and morphology of human spermatozoa.

Excessive alcohol consumption has been associated with impaired reproductive function by causing the inhibition of penile tumescence and ejaculatory capability. Alcohol intoxication has also been implicated in impaired spermatogenesis and an increase in sperm structural anomalies. The aim of this study was to determine the direct effects of alcohol on sperm motility and morphology in vitro. Semen samples from 67 subjects were prepared using density centrifugation. Ethanol was added, at concentrations in serum equivalent to social, moderate and heavy drinking, to the medium in which the spermatozoa were cultured. Sperm motility was assessed using computer assisted semen analysis and morphology was assessed by Tygerberg strict criteria after 0, 15, 30, 60, 120, 180 and 240 min exposure. Each concentration of ethanol produced significant decreases in the percentage progressive motility, straight line velocity and curvilinear velocity. The amplitude of lateral head displacement was also depressed by 300 and 500 mg dL-1 of ethanol. A significant decrease in the number of spermatozoa with normal morphology and an increase in irreversible tail defects were observed after exposure to 300 mg dL-1 ethanol. When alcohol is added directly to sperm, at concentrations equivalent to those in serum after moderate and heavy drinking, damaging effects are observed in both sperm motility and morphology.

Ethanol↗

PubMed1988

Laminar-specific distribution and infrastructural detail of amyloid in the Alzheimer disease cortex visualized by computer-enhanced imaging of epitopes recognized by monoclonal antibodies.

Monoclonal antibodies to the A4 amyloid polypeptide were used in immunocytochemical staining of the Alzheimer disease prefrontal cortex. Analysis of the resulting staining patterns allowed us to evaluate the amounts and distribution of amyloid-protein deposits exclusive of other senile-plaque components. Previously unappreciated infra-structural details of amyloid in the Alzheimer disease brain became accessible through computer-enhanced imaging procedures. Four discrete morphologic classes of amyloid deposits were observed and classified as punctate, macular, ring, and ring-with-core configurations. Computer imaging indicated that all four classes of immunostained deposits contain internal gradients of density. The classes were nonuniformly distributed with regard to size and location within cortical laminae. Our results support two separate but complementary hypotheses concerning the molecular neuropathology of Alzheimer disease in the prefrontal cortex. (i) Irrespective of cortical layer or morphology, density-gradient analyses suggest that amyloid deposits are elaborated through molecular and cellular events that may involve diffusion or coalescence of the A4 polypeptide. (ii) The distribution and morphology of prefrontal cortical amyloid deposits may be dependent upon underlying laminar-specific structures of the neocortex.

Alzheimer Disease↗

PubMed2001

Morphological algebraic models of the TU-wave patterns/in idiopathic long QT syndrome.

A computer-assisted analysis of the TU-complex morphology was employed to characterize repolarization abnormalities in LQTS and to assess arrhythmic risk. Electrocardiograms (ECGs) were collected from 14 idiopathic LQTS patients (seven without symptoms and seven with a history of syncope or cardiac arrest) and from 14 sex- and age-matched normal subjects. Digitized TU-wave patterns from V2-V6 precordial leads were analyzed. The morphologies of the T and U waves were modeled by an algebraic sum of differences between two pairs of action potential-like curves of different shape and duration so that the whole TU complex was approximated by (S1-S2)+(L1-L2). By finding the best fit model of the digitized TU-wave signal, the amplitude and duration of each decomposition curve were determined for each lead. The following 'secondary' parameters were then derived: (a) the ratio between the sum of the amplitudes of the two long (L1 and L2) and the two short (S1 and S2) decomposition curves (A-ratio), (b) the highest A-ratio found in V2 to V6 (A-ratio(max)), and (c) the model-derived durations of the T-wave, U-wave and TU-complex. Conventional measures of RR and QTc intervals and of QT dispersion did not differ between symptomatic and asymptomatic LQTS patients. Modeled QT interval was significantly longer in the symptomatic than in the asymptomatic LQTS patients and in asymptomatic LQTS patients than in the controls. In addition, symptomatic LQTS patients had a longer S2 and T-wave duration in most leads than normal subjects. Conversely, modeled QU interval and U-wave duration did not significantly differ between the three groups. Compared to normal subjects, the amplitudes of S1, S2, L1 and L2 in the LQTS patients were not significantly different in most leads. A-ratio and A-ratio(max) were greater in symptomatic than asymptomatic LQTS patients and in the latter than in controls. A cut-off value of 0.90 of A-ratio(max) separated all symptomatic (1.34+/-0.38) from all asymptomatic patients (0.60+/-0.21). Although the correlation between model parameters and cellular substrate is at present unclear, it is possible that the morphological alterations described by the model are related to the arrhythmogenic mechanism(s) of the idiopathic LQTS.

Adolescent↗

PubMed1998

Quantification of the morphological features of a full microvascular network.

Investigations into the changes that occur in microvasculature following the surgical procedure called delay have brought about the need for a computer system capable of quantifying the morphological features of a full microvascular network in terms of average vessel length, diameter, and tortuosity. Both the formulaic conventions that have been developed to measure these quantities as well as their implementation in the form of a HP-9000/UNIX based computer software system that we developed specifically for this purpose are discussed. Reliability studies performed using the final system to measure the microcirculatory network of a mouse latissmus dorsi muscle (LDM) showed 95% confidence intervals within 5% of means and coefficients of variability within 7% of means for all quantities measured in large (150-300 microns), medium (50-150 microns), and small (< 50 microns) diameter vessels. These variations were significantly smaller than the changes that were observed in a preliminary study comparing these microvascular network parameters before and after delay in the hairless mouse LDM, showing the proposed quantification methods to be well suited to the study of the microvascular changes following delay. It is hoped that the formulaic conventions, implementation process and reliability data will provide a useful comparison for other researchers interested in measuring similar features of microcirculatory networks.

Animals↗

PubMed1989

[Electrocardiographic findings in a school-age population using a computerized system: calculation of standard norms for heart rate and PR, QRS, and QT intervals].

We evaluated some ECG parameters (HR, P-R, Q-R-S, Q-T) in a healthy school-age population. One-thousand-eight-hundred and ninety children ranging in age from 5 to 12 coming from different zones of the city of Naples were studied. On physical examination all subjects were free from cardiac disease. A computerized ECG (Muse 12SL System Marquette) was performed on every subject. One-hundred and seventy-four of the 1890 children were excluded from statistical analysis because of ectopic rhythm (junctional rhythm or wandering pacemaker) or poor quality of the recording. Of the 1716 children included in the study, 837 were male and 879 were female. The computerized ECG 12SL System Marquette registers an ECG record consisting of all 12 classical ECG leads acquired simultaneously over a 10 second period. Each individual complex can be analyzed in all leads by the computer. An interpretation using this extended record, along with an ECG record of conventional length, is presented to the physician for review. The first step in computerized ECG analysis is Q-R-S identification, then P wave identification, beat classification, rhythm analysis, morphology analysis, complex alignment and computation of median complex. All parameters were divided for sex and age and gathered into tables. The variability of P-R, Q-R-S, Q-T versus HR were also evaluated. The following conclusions were drawn: 1) Sex is a very important variable in the parameters examined. Males have a much slower HR, greater Q-R-S duration, and longer Q-T interval when compared to females. 2) As age increases, HR slowly decreases, while P-R, Q-R-S and Q-T intervals increase.(ABSTRACT TRUNCATED AT 250 WORDS)

Child↗

PubMed1991

Cytotoxicity of various crystalloid solutions to the endothelial cells of autologous grafts.

Harvesting and storage may influence the early failure of autologous grafts. We studied the endothelial layer of human saphenous veins (HSV) and bovine internal mammary arteries (BMA) after a two-hour storage period in comparison to vessel segments prepared immediately after harvesting. The storage solutions were 0.92% NaCl, 0.9% NaCL heparinized, Ringer's Lactate (RL), heparinized blood or Bretschneider's cardioplegic solution. To visualize the endothelial cell damage, specimens were stained with silver nitrate. The surface covered by morphologically intact endothelial cells was computed in percent. In a perfusion model using the carotid artery of rabbits, the increase of the permeability of the endothelium after two hours' perfusion was investigated (n = 5 each). In native HSV-segments 70.8 +/- 4.5% of the endothelium were found to be morphologically intact. The figures for stored grafts were: 14.7 +/- 4.4% (NaCl), 0.5 +/- 0.3% (hep. NaCl), 16.0 +/- 4.6% (RL), 37.2 +/- 5.3% (hep. blood) and 50.0 +/- 4.2% (Bretschneider). Comparable results were found with BMA specimens. BMA: 98.3 +/- 2.1% (native), 39.2 +/- 3.3% (NaCl), 8.4 +/- 3.5% (hep. NaCl), 11.2 +/- 4.8% (RL). 67.0 +/- 6.1% (hep. blood) and 78.1 +/- 4.7% (Bretschneider). In accordance with these results the increase of the permeability of the endothelial cell layer of the carotid artery of rabbits was significantly less after perfusion with Bretschneider's solution. The condition of the endothelial cells of autologous grafts at the time of implantation was found to be clearly dependent on the type of solutions used for perfusion and immersion. In comparison to all crystalloids studied in this series, Bretschneider's cardioplegic solution proved to be superior.

Adolescent↗

PubMed1985

Changes in the normal corneal endothelial cellular pattern as a function of age.

Human endothelial morphologic changes were quantitated by specular microscopy and computer-assisted morphometry to establish normal baselines of various morphologic parameters. Cellular polymegethism and cellular pleomorphism increases with age, and normal baseline parameters are detailed. Furthermore, no significant difference in any morphologic parameters between the right and left eye and between central and peripheral endothelium was detected in the normal corneas examined. These normal morphologic baselines can possibly be utilized to detect early corneal endothelial pathology and/or cell loss nondetectable by cell density measurement.

Adolescent↗

PubMed2005

No long-term effects of ultrasound therapy on bioabsorbable screw-fixed lateral malleolar fracture.

BACKGROUND AND AIMS: The present study was initiated to evaluate the long-term effects of low-intensity ultrasound therapy on bioabsorbable screw-fixed lateral malleolar fractures, which has not been studied earlier. PATIENTS AND METHODS: The study design was prospective, randomized, double-blinded, and placebo-controlled. Sixteen dislocated lateral malleolar fractures were fixed with one bioabsorbable self-reinforced poly-L-lactide screw. The patients used an ultrasound device 20 minutes daily for six weeks without knowing it was active (eight patients) or inactive (eight patients). The follow-up time was 18 months. The radiological bone morphology was assessed by multidetector computed tomography (MDCT) scans, the bone mineral density by dual-energy X-ray absorptiometry scans, and the clinical outcome by Olerud-Molander scoring and clinical examination of the ankle. RESULTS: The MDCT scans revealed that all fractures were fully healed, and no differences were observed in radiological bone morphology at the fracture site. The bone mineral density of the fractured lateral malleolus tended to increase slightly during the 18-month follow-up, the increase being symmetrical in both groups. No differences were observed in the clinical outcome or Olerud-Molander scores. CONCLUSIONS: The six-week low-intensity ultrasound therapy had no effect on radiological bone morphology, bone mineral density or clinical outcome in bioabsorbable screw-fixed lateral malleolar fractures 18 months after the injury.

Absorbable Implants↗

PubMedDate not supplied

Comparison of two staining and evaluation methods used for computerized human sperm morphology evaluations.

The purpose of the study was to analyse the agreement between computer analysed (Hamilton Thorne, IVOS Dimensions Version 3) normal sperm morphology and values obtained from 97 slides stained according to the Papanicolaou and Diff-Quik method. Liquefied semen samples were washed once by centrifugation and air dried smears on slides were made, which were stained according to the Papanicolaou and Diff-Quik method and analysed by computer. The paired t-test was used to assess whether any bias existed between the two methods. The limits of agreement were calculated using the Bland and Altman approach and a modification of this approach (mean-dependent limits). A significant bias of 1.6% was obtained in favour of higher normal sperm morphology percentages when using the Diff-Quik method. The standard limits of agreement were -13.4% to 16.6%, whereas the mean-dependent limits of agreement were 1.6% [5.8 + 0.6 (mean percentage normal morphology)]. Statistically, the Diff-Quik and Papanicolaou staining methods produce different normal sperm morphology profiles. These inherent differences may, therefore, require the establishment of new normal sperm morphology thresholds for male fertility, based on clinical data, when using the Diff-Quik staining method in conjunction with computerized analysis.

Azure Stains↗

PubMed2006

Role of mechanical factors in the morphology of the primate cerebral cortex.

The convoluted cortex of primates is instantly recognizable in its principal morphologic features, yet puzzling in its complex finer structure. Various hypotheses have been proposed about the mechanisms of its formation. Based on the analysis of databases of quantitative architectonic and connection data for primate prefrontal cortices, we offer support for the hypothesis that tension exerted by corticocortical connections is a significant factor in shaping the cerebral cortical landscape. Moreover, forces generated by cortical folding influence laminar morphology, and appear to have a previously unsuspected impact on cellular migration during cortical development. The evidence for a significant role of mechanical factors in cortical morphology opens the possibility of constructing computational models of cortical development based on physical principles. Such models are particularly relevant for understanding the relationship of cortical morphology to the connectivity of normal brains, and structurally altered brains in diseases of developmental origin, such as schizophrenia and autism.

Algorithms↗

PubMed1996

Computer simulations of voltage clamping retinal ganglion cells through whole-cell electrodes in the soma.

1. Computer simulations of voltage-clamp experiments in retinal ganglion cells were implemented to better understand the insights that can be obtained with this physiological approach. 2. Simulation studies of voltage clamping were based on the contemporary approach of using whole-cell recordings with low resistance electrodes attached to the soma. Realistic ganglion cell morphologies were provided by cell staining experiments in the mudpuppy retina; selected cells included small-, medium-, and large-field neurons whose morphologies were entered into a computer through a neuron tracing program. 3. Values for the specific membrane resistance (Rm) varied from 5,000 to 100,000 omega/cm2 to conform to the range of Rm values obtained with intracellular sharp electrodes and whole-cell recordings. 4. Synaptic input currents were simulated by injecting current with and without an underlying conductance change into different regions of the dendritic tree. The time-variant waveform of the current included a combined transient and sustained component similar to the waveform of ON-bipolar activation. 5. Simulations were base on 1) intact structures, which included the soma and the entire dendritic tree, and 2) a more limited cell geometry that included representation of the soma, but only part of the dendritic tree, to represent the restricted morphology that might be rendered after cutting the retina into 150-microns cross sections for retinal slice experiments. 6. The results of this study indicate that voltage clamping from the soma, with optimal, low resistance electrodes and series resistance compensation, provides an error-free voltage clamp for slow signals that are generated within a small electrotonic distance from the soma (approximately 0.1 lambda). 7. The ideal voltage-clamp conditions are optimized when synaptic conductances are small and nonlinear membrane elements are minimally activated: small-field neurons best approximate these conditions, but clamping errors are evident in these cells when more distal branches are activated. The degree of error in voltage clamping was much greater when medium-and large-field neurons were evaluated. 8. It was not possible to clamp action potentials (nonpropagating) even when they were generated near the soma in any of the three model cells examined. 9. Experimental paradigms were developed to demonstrate that inadequate voltage clamping can lead to errors in the interpretation of experimental data when relevant variables are not taken into consideration. Suggestions are made for determining and optimizing favorable clamp conditions.

Action Potentials↗

PubMed2005

Morphometry characterisation of European eel spermatozoa with computer-assisted spermatozoa analysis and scanning electron microscopy.

The aim of the present study was to characterise European eel spermatozoa morphometrically, as a basis for future studies on the morphological effects of methods for sperm cryopreservation and sperm quality. This characterisation was carried out measuring several spermatozoa morphology parameters (head length, width, area and perimeter) by scanning electron microscopy (SEM), in comparison with measurements developed in European eel spermatozoa with computer-assisted morphology analysis (ASMA). Spermatozoa head morphology showed differences in width (1.15+/-0.01 microm versus 1.12+/-0.01 microm), perimeter (14.68+/-0.13 microm versus 13.72+/-0.19 microm) and area (5.36+/-0.06 microm2 versus 1.12+/-0.01 microm2) for ASMA and SEM, respectively. When head length was evaluated, significant differences were found, being higher for SEM methodology (5.09+/-0.04 microm versus 4.29+/-0.03 microm). The curved and elongated spermatozoa head in eels means a problem for the ASMA system (Sperm Class Analyser), Morfo Version 1.1, Imagesp, Barcelona, Spain), causing an error in the length measurements. However, similar results were obtained by both techniques when spermatozoa head length was considered as the greater length between two points within the object (4.29+/-0.03 microm versus 4.31+/-0.04 microm for ASMA and SEM, respectively). In conclusion, this is one of the first applications of ASMA in fish and the first in this species, and confirms this system as a useful tool with wide applications in future fish spermatozoa studies. Width, perimeter and area could be used as parameters for the spermatozoa morphology evaluation, whereas the length requires a new programming of the Imagesp software.

Animals↗

PubMed1999

Computer-assisted anthropometry for outcome assessment of cleft lip.

Anthropometry and clinical examination best evaluate the morphology of repaired cleft lip and nose. An original, accurate, and practical image analysis of the lip and nose, which takes advantage of the mathematic, geometric, and organizational capabilities of public domain NIH-Image software (http://rsb.info.nih.gov/nih-image/), has been developed and tested over the past 6 years. A modified structured physical examination form that complements this analysis is under study. Accuracy of NIH-Image-based anthropometry was compared with direct measurements of 22 linear distances on the lip and nose. Twenty-five sets of direct measurements were taken, prospectively, on 15 children with repaired cleft lip over a 6-year period. The results were submitted to regression analysis. Then, relevant lip and nasal tip aesthetics were evaluated by the measuring capabilities of NIH-Image to create a quantitative assessment tool. For each episode, 15 possible faults were weighted, according to aesthetics and deformity, to provide an adverse score. The sum of the 5 lip scores, 10 nose scores, and combination gave respective grades. The analysis was modified to stratify congenital deformity to relate severity of disease to outcome. This analysis was applied to digitized images of 19 consecutive children, immediately prior to repair of complete unilateral cleft lip and nose, at the time of palate repair, and annually from the age of 3 to 6 years. There were 19 NIH-Image-based measurements of the congenital deformity and 35 measurements of surgical results; four children had three sets of records, eight had two sets, and seven had one set Descriptive statistics were applied. Following 556 paired direct and computer-assisted measurements, exceptional linear correlation was shown with a Pearson R coefficient of 0.96. The best correlation was lines within the plane of the camera lens, with the average difference ranging between 0.025 and 0.997 mm. Visual inspection of frontal and submental photographs of excellent, good, and poor results substantiates the ability of this analysis to quantify and grade a spectrum of relevant cleft lip and nasal anatomy. For these 19 patients, there was a broad range of performance scores, approximating a normal distribution. The mean of the NIH-Image-based analysis scores, 16.91, was a (very) good grade. A single standard deviation of 6.88 extended up into excellent and down to fair. The congenital analysis indicated a range of deformity. Comparing deformity with outcome, simple regression analysis had a coefficient of determination (R2) of 0.223, indicative of a weak positive relationship. An accurate and practical morphologic computer-assisted outcome assessment of repaired cleft lip and nasal deformity has been developed. There is a weak direct correlation between severity of deformity and outcome. Testing in multiple clinics is warranted.

Anthropometry↗

PubMed2001

Suitability for endovascular aneurysm repair in an unselected population.

BACKGROUND: Tertiary referral centres report that up to 60 per cent of patients may be suitable for endovascular repair of abdominal aortic aneurysm (EVAR). The aim of this study was to determine the percentage of abdominal aortic aneurysms (AAAs) presenting to a county-wide vascular service that were suitable for EVAR, and to examine the outcome of subsequent AAA repair in relation to aneurysm morphology. PATIENTS AND METHODS: All patients being assessed for AAA repair between January 1998 and December 1999 underwent spiral computed tomography angiography to determine aneurysm morphology and suitability for EVAR. Subsequent outcome for all patients in the study was recorded in a prospective vascular database. RESULTS: A total of 115 patients was assessed. Sixty-three aneurysms (55 per cent) had one or more absolute contraindications to EVAR, a further 13 (11 per cent) had at least one relative contraindication, and 39 (34 per cent) had no contraindication. Of patients with no absolute contraindication to EVAR, ten underwent successful EVAR, five did not meet recognized criteria for surgery, one awaits EVAR, four remain under observation, one awaits open repair, and 31 underwent open repair without death. CONCLUSION: Only 30 per cent of unselected AAAs presenting to a vascular service are entirely suitable for EVAR; most of these patients can safely undergo open AAA repair. These data suggest that increased use of EVAR is only possible by deploying devices in suboptimal morphology, and in treating patients who would not normally be considered for open AAA repair.

Aged↗

PubMed1996

[Simultaneous digital indocyanine green and fluorescein angiography].

BACKGROUND: The simultaneous recording of indocyanine green and fluorescein angiography is since more than twenty years possible. The real-time digital recording of angiograms is about since 10 years possible. Because of the poor temporal resolution due to technical limitations, both of the above methods were only seldom used routinely. Graphics workstations can digitize and store multiple video sources real-time with full video framerate. Comparing to conventional videoangiography, the quality can be improved significantly. MATERIALS AND METHODS: Simultaneous ICG and Fluorescein (SIF) angiography images of a two-channel Scanning Laser Ophthalmoscope were digitized and stored real-time at a rate of 28-50 images/sec using a graphics workstation. The digitized image sequence could be replayed as a motion picture and/or single images of the recorded sequence could be analyzed separately. RESULTS: The simultaneous digital recording of angiograms is easier, than using two video tape recorders. It was possible to enhance the quality of the stored images using standard digital image processing procedures. Morphological measurements and densitometry were done with software developed for the UNIX workstation, or with personal computers using standard software packages. CONCLUSION: Simultaneous digital ICG and fluorescein angiography has several advantages over conventional video-angiographies: the image quality is better, data retrieval is easier, it is easier to perform than conventional video angiography, image can be accessed through a computer network, it is easy to perform morphological and densitometrical measurements, etc.

Angiography↗

PubMed1999

Morphological bidirectional associative memories.

The theory of artificial neural networks has been successfully applied to a wide variety of pattern recognition problems. In this theory, the first step in computing the next state of a neuron or in performing the next layer neural network computation involves the linear operation of multiplying neural values by their synaptic strengths and adding the results. Thresholding usually follows the linear operation in order to provide for nonlinearity of the network. In this paper we discuss a novel class of artificial neural networks, called morphological neural networks, in which the operations of multiplication and addition are replaced by addition and maximum (or minimum), respectively. By taking the maximum (or minimum) of sums instead of the sum of products, morphological network computation is nonlinear before thresholding. As a consequence, the properties of morphological neural networks are drastically different from those of traditional neural network models. The main emphasis of the research presented here is on morphological bidirectional associative memories (MBAMs). In particular, we establish a mathematical theory for MBAMs and provide conditions that guarantee perfect bidirectional recall for corrupted patterns. Some examples that illustrate performance differences between the morphological model and the traditional semilinear model are also given.

Journal Article↗