Synthetic MR images produced by subspace methods: a comparison.
Explore the source record for details and available documents.
Brain Neoplasms↗
Biomedical subjects
J J Sychra
Publications and source records attributed to J J Sychra.
Discriminant analysis on cells from developing squamous cancer of the respiratory tract.
Cytologic preparations made from the tracheobronchial tree taken by the Schreiber catheter have been scanned by three color microphotometry. The digitized cell images were processed by the analytical cytodiagnostic programs of the TICAS system. Cells were sorted into two control groups and five groups of increasing atypia ranging from normal epithelium to invasive squamous cell carcinoma. Standard statistical tests, including Wilk's Lambda, Rao's V, and the Kruskal-Wallis tests are performed on these subsets of cell image features. This study demonstrates that discriminant analyses permit differentiation between normal cells and those from marked atypia or carcinoma and that the classification achieves a high degree of agreement with visual assignment.
Animals↗
Differentiation by TICAS analysis of cell populations of tracheal aspirates from hamsters with squamous-cell carcinoma.
Individual cells from the tracheal aspirates of hamsters exposed to benzo-a-pyrene were scanned at .5 mum in three colors. Features relating to size, shape, and color were extracted and calculated by computer. The single cells were then classified by these features into separate populations with varying degrees of atypia, extending up to frank cancer cells. A high degree of accuracy was attained in classification by these methods.
Animals↗
Radionuclide synthetic Fourier images of cardiac wall motion abnormalities.
At present, the phase images together with amplitude images are used in nuclear medicine to aid the diagnosis of cardiac regional wall motion abnormalities (RWMA). The phase images represent the spatial distribution of the relative phase of the first harmonic fit of pixel time activity curves, and the amplitude images represent the distribution of the amplitude of the fit. These images contain only part of the information present in the original radionuclide images, and have to be mentally integrated with other known information to obtain a diagnosis. The proposed synthetic Fourier images overcome these deficiencies as their pixel intensity is a function of additional Fourier parameters of pixel time activity curves and of pixel location and are not limited to the first harmonic. But most importantly, their computation is based on "teaching" the computer by examples of previously diagnosed cases. The images offer direct and robust diagnosis which is superior to that derived from separate phase and amplitude images, especially for the detection of mild RWMA.
Fourier Analysis↗
Patient classification based on cytologic sample profiles. I. Basic measures for profile construction.
Past efforts in the field of automated cell recognition have focused upon the separation and classification of cell types. From these efforts, large data banks have been built and work in the field is now shifting towards the practical application of this information for clinical diagnoses. This paper presents the initial results of work on a system developed to undertake the reduction of the masses of data into diagnostically useful patient cytologic sample profiles.
Autoanalysis↗
Computer analysis of atypical urothelial cells. I. Classification by supervised learning algorithms.
Computer discrimination of atypical (ATY) urothelial cells from the urinary sediment by means of supervised learning algorithms discoled that these cells form a distinct, although ill-defined, family of cells which differs from normal (NEG) and malignant (POS) cell groups. The clinical significance of this observation must await long-term clinical follow-up. The possibility of issuing computer displays on individual patients with possible diagnostic and prognostic implications is discussed.
Aged↗
Computer analysis of atypical urothelial cells. II. Classification by unsupervised learning algorithms.
Explore the source record for details and available documents.
Aged↗
Dimensionality reducing displays in cell image analysis.
The paper discusses the problem of representation of multivariate cell data sets in two dimensions such that the essence of the situation as represented by the multivariate feature space is preserved. A corresponding projectional technique has been developed and illustrated on a set of five different cell types of ectocervical cells, including normal and carcinoma cells.
Computers↗
Computer recognition of ectocervical cells. Classification accuracy and spatial resolution.
The effect of a relaxation of the spatial resolution of images of ectocervical cells on classification accuracy by computer has been studied. The discriminatory capabilities of optical density, texture and shape features have been evaluated. Comparable computer classifications have been derived. The differences in the discriminating power of .5 and 1 mu resolution are found to be smaller than the differences corresponding to 1 and 2 mu resolution. While 2 mu resolution may be marginally acceptable in certain situations, the 4 mu resolution did not render acceptably good classification rates.
Computers↗
Computer recognition of abnormal ectocervical cells. Comparison of the efficacy of contour and textural features.
The efficacy of contour and texture features for the computer recognition of abnormal ectocervical cells is compared. For this comparison, three cell groups were formed: benign, suspicious and malignant. Nine decision rules were derived. The discrimination attained by features derived from the contours of nucleus and cytoplasm is comparable to that achieved by textural features. A combination of both kinds of features gives best results.
Cell Nucleus↗
Computer recognition of binucleation with overlapping in epithelial cells.
A technique for computer recognition of binucleation with overlapping in epithelial cells was derived. The computer program uses analytic features related to nuclear contour only. The feature set includes moments, Fourier transform features and linguistic features. The algorithm works satisfactorily in instances where the human eye would detect the presence of two nuclei at first glance.
Cell Nucleus↗
A system for scanning biological cells in three colors.
A system for scanning biological cells under high magnification has been developed which utilizes a three-color photometer. The spectral information has proved useful in improving computer recognition of certain cell types.
Cytodiagnosis↗
Computer discriminant analysis of atypical urothelial cells.
Prior computer studies of digitized cell images by the TICAS system have shown that the category of urothelial cells classified visually as atypical may be composed of 2 subgroups, one clustering mainly with benign cells and the other with malignant cells. As a consequence, a visual review of the group of atypical cells was conducted and tested by computer discriminant analysis. The computer classification confirmed the visual reclassification and subdivision of atypical urothelial cells into 2 subgroups, ATY I and ATY II. This is yet another example of feedback from computer diagnosis to visual assessment of cells. The significance of these observations in terms of diagnosis will be the subject of subsequent communications.
Computers↗
Indices of cell atypia in urinary tract cytology.
A sample profile based on assessment of digitized cell images of normal (NEG), malignant (POS) and two classes of atypical (ATY I and ATY II) urothelial cells was established. Data distributed in a multidimensional feature space were projected into a two-dimensional display space using linear discriminant functions as composite features. Fitting of a polynomial led to the derivation of an atypicality index for each cell and to statistically clearly significant differences in the atypicality values for the four groups of urothelial cells. The application of these findings to patients' profiles will be examined in a subsequent communication.
Computers↗
Diagnostic cytologic sample profiles in patients with bladder cancer using TICAS system.
Computer analysis of digitized cell images was applied to consecutively encountered, well preserved urothelial cells in the urinary sediment of 12 patients with bladder cancer. It was shown that the composition of the cell sample, and notably the proportion cells classified in the ATY II and POS (malignant) categories, were of diagnostic significance. This work indicated that a computer-generated diagnosis based on the cells in a urinary sample could probably be achieved with a relatively small number of urothelial cells. The study also suggested that computer-generated cytologic profiles of patients with non-papillary carcinoma in situ can be distinguished from other forms of urothelial cancer. The atypicality indices computed for all cells from each patient provided important information but were per se insufficient for diagnostic purposes. This preliminary study, based on a small group of patients, suggests that high resolution scanning offers a promising approach to automation of cytology of the urinary sediment.
Carcinoma↗
Cytoplasmic and nuclear shape analysis for computerized cell recognition.
Explore the source record for details and available documents.
Cell Nucleus↗