Search

The Facial Action Coding System (FACS) [23] is an objective method for quantifying facial movement in terms of component actions. This system is widely used in behavioral investigations of emotion, cognitive processes, and social interaction. The coding is presently performed by highly trained human experts. This paper explores and compares techniques for automatically recognizing facial actions in sequences of images. These techniques include analysis of facial motion through estimation of optical flow; holistic spatial analysis, such as principal component analysis, independent component analysis, local feature analysis, and linear discriminant analysis; and methods based on the outputs of local filters, such as Gabor wavelet representations and local principal components. Performance of these systems is compared to naive and expert human subjects. Best performances were obtained using the Gabor wavelet representation and the independent component representation, both of which achieved 96 percent accuracy for classifying 12 facial actions of the upper and lower face. The results provide converging evidence for the importance of using local filters, high spatial frequencies, and statistical independence for classifying facial actions.

Facial expressions provide an important behavioral measure for the study of emotion, cognitive processes, and social interaction. The Facial Action Coding System (Ekman & Friesen, 1978) is an objective method for quantifying facial movement in terms of component actions. We applied computer image analysis to the problem of automatically detecting facial actions in sequences of images. Three approaches were compared: holistic spatial analysis, explicit measurement of features such as wrinkles, and estimation of motion flow fields. The three methods were combined in a hybrid system that classified six upper facial actions with 91% accuracy. The hybrid system outperformed human nonexperts on this task and performed as well as highly trained experts. An automated system would make facial expression measurement more widely accessible as a research tool in behavioral science and investigations of the neural substrates of emotion.

Use of valvulotomes in non-reversed venous conduits carries the potential of venous endothelial cell injury. Earlier studies have shown that there is a significant decrease in the number of endothelial cells present in the human saphenous vein when employing the circular (LeMaitre) valvulotome. The present study was performed to evaluate and compare the LeMaitre and modified Hall valvulotome techniques on vascular endothelial cells from human saphenous vein. The results of the present study indicate that while both valvulotomes caused a significant decrease in the number of endothelial cells, the modified Hall instrument was less damaging to the vascular endothelium than the LeMaitre valvulotome. These results suggest that the modified Hall valvulotome technique may be more beneficial in maintaining endothelial cell function when the use of a valvulotome is warranted.

Mouse mammary tumor cells were cultured on a three-dimensional bundle of semipermeable hollow fibers as a prototype system for large-scale in vitro production of mammary tumor-associated antigens. Cells were seeded onto the surface of the hollow fibers that served as an artificial capillary network. The cells grew to tissue-like densities within 3 weeks, achieving the high density and three-dimensional intercellular relationships known to potentiate expression of murine mammary tumor virus (MuMTV) antigens. Significant levels of a 52,000-relative-molecular-weight (Mr) glycoprotein antigen of MuMTV (gp52) were detected on the cell side of the semipermeable membrane of the hollow fibers in a yield at least fiftyfold greater than that produced by monolayer cultures. Antigen could be collected from capillary cultures for 6-12 weeks, rather than days from monolayer cultures. Further, gp52 was collected in an undegraded state. Perfusates contained lesser amounts of cross-reactive antigenic species that were predominantly smaller than 50,000 Mr, which suggests that the fibers may be used to separate antigens from their degradation products. The production of nonviral, mammary tumor-associated antigens by cells on artificial capillaries was also demonstrated. Artificial capillary cell culture systems provide a means to obtain significant quantities of tumor-relevant antigens in a semipurified natural state.

The establishment of a maturation-induction model using human colon cancer cells as targets is reported. Two colon carcinoma cell lines were established from human tumors; one line was heterogeneous and was cloned into two distant subpopulations. Cells from these lines and clones and cells from an established human colon carcinoma cell line were treated in vitro with N,N-dimethylformamide (DMF) and were compared to untreated cells according to two general sets of criteria. One set contains characteristics that define a cell as transformed (i.e., anchorage independence and tumorigenicity for nude mice), and the second set contains three antigenic marker systems that would provide evidence that maturation is occurring in treated human colon cancer cells. These colon tissue-or tumor-related markers include carcinoembryonic antigen (CEA), colonic mucoprotein antigen (CMA), and the human blood group determinants. DMF-treated cells are less malignant than untreated cells; the treated cells show a marked reduction in tumorigenicity and in clonogenicity in soft agar. Each of the markers indicates that the treated cells are better differentiated than their untreated counterparts. For example, treated cells show increased expression of normal-CMA and decreased expression of tumor-CMA compared to untreated cells. Cells removed from DMF reverted to express the tumorigenicity, growth properties, and antigens characteristic of their untreated counterparts. Therefore, DMF reversibly induces in cultured colon cancer cells a less malignant phenotype with concomitant maturational effects. These results indicate that this model is appropriate for study of maturation-induction in human colon tumor cells, and has potential application to other types of human carcinomas.

Cultured human colon carcinoma cells were induced by the polar solvent N,N-dimethylformamide (DMF) to express a more differentiated phenotype as indicated by three types of antigen markers. Carcinoembryonic antigen expression in all DMF-induced cell lines was enhanced. Exposure of cells to DMF resulted in a reduction in ability to bind antibody to tumor-derived colon mucoprotein antigen (CMA) with a concomitant gain in binding of antibody to normal CMA. DMF-induced differentiation was further indicated by modulation of blood group antigen expression. DMF treatment decreased the amount of expression of the H-gene determinant. When DMF-treated cells were cultured in the absence of DMF for 14 days, the levels of expression of the antigen markers reverted to those characteristic of their untreated counterparts.

We have described previously the isolation and characterization of five distinct subpopulations of tumor cells from a single spontaneous strain BALB/cfC3H mouse mammary tumor (Cancer Res., 38: 3174--3181, 3758--3763, 1978). Subpopulations 68H and 4.10 are polygonal and grow in epithelioid patterns in vitro, whereas subpopulations 66, 67, and 168 are fusiform and grow in lattice or fibroblast-like patterns. Line 4.10 produces tumors with distinctly glandular architecture, whereas the other four subpopulations produce poorly differentiated tumors with mixed epithelial-sarcomatous histological patterns. All five lines were evaluated for epithelial characteristics. Dome formation, characteristic of transporting epithelial cells, could be induced by dexamethasone or dimethyl sulfoxide only in line 4.10 cells. Antibodies to cell type-specific mammary epithelial antigens reacted with each of the subpopulations. All five subpopulations had ultrastructural features of epithelial cells, including desmosomes (all five lines), junctional complexes (68H, 4.10, early-passage 66 and 67 only; poorly defined in 168), and growth in cords demonstrating polarity (68H cells). Less definitive myoepithelial characteristics were also seen in four of the lines, including an incomplete reaction for Na+-K+-ATPase (4.10 cells), hemidesmosome-like junctions (168 and early-passage 66 cells), and pinocytotic vesicles at lower than normal frequency (66, 67, and 168 cells). Thus, none of the lines were distinctly myoepithelial. We conclude that the five subpopulations are epithelial cells that express a spectrum of epithelial characteristics.

An in vivo model is described for assessing the antitumor activity of chemotherapeutic agents. Tumors derived from human colon carcinoma cell lines injected into antithymocyte serum (ATS) immunosuppressed mice were used. In this system, both antitumor effects and host toxicity can be quantitated, permitting calculation of a Therapeutic Index. Compared with other xenograft models, the present system is simple, experiments are completed in less than 2 weeks, and the use of cultured cell lines allows in vitro studies to be performed. The in vitro sensitivities of one colon cell line to 22 chemotherapeutic agents and of four cell lines to three agents is reported. Four drugs used in treating colon cancer (Mitomycin C, 5-FU, BCNU, and methyl-CCNU) show antitumor activity in vivo in this system. Each has a low therapeutic index. Further work with this model is indicated, with the goal of finding new drugs with high Therapeutic Indices.

Indirect viable cell immunofluorescence studies revealed, on the plasma membrane of cultured adenocarcinoma cells from the human colon and rectum, a tissue-specific antigen that reacted with a component of normal rabbit serum. The use of G200 column chromatography and monospecific antisera prepared against class-specific heavy chains of rabbit immunoglobulin indentified the reactive components in rabbit serum as IgM. Examination of normal and malignant cells from various human tissues by immunofluorescence and adsorption studies showed that rabbit serum reacted only with cells derived from the intestinal epithelium.

Seven spontaneous BALB/cfC3H mouse mammary tumors were heterogeneous in expression of murine mammary tumor virus-associated cell surface antigens. To determine the basis of this heterogeneity, cells from spontaneous tumors and from five subpopulations isolated from a single spontaneous tumor were examined for expression of viral antigens under both conventional conditions and conditions known to induce synthesis of murine mammary tumor virus antigens (5-iodo-2'-deoxyuridine and dexamethasone). Induction with 5-iodo-2'-deoxyuridine resulted in further manifestation of the antigenic heterogeneity of spontaneous tumors. The five subpopulations from a single tumor differed in the amount of viral antigens present in untreated and in induced cultures. Coculturing showed that viral antigen expression was independent in each subpopulation within a heterogeneous mixture and was not influenced by the presence of other subpopulations with different potentials for viral antigen synthesis. The expression of murine mammary tumor virus structural antigens, a protein with a molecular weight of 28,000 and a glycoprotein with a molecular weight of 52,000, differed within the heterogeneous subpopulations, and was noncoordinate. The data suggest that the antigenic heterogeneity in spontaneous tumors reflects the existence of cells within them that differ in both expression of viral antigens and in their response to inducers of viral antigen synthesis.