A convolutional neural network-based linguistic steganalysis for synonym substitution steganography.

In this paper, a linguistic steganalysis method based on two-level cascaded convolutional neural networks (CNNs) is proposed to improve the system's ability to detect stego texts, which are generated via synonym substitutions. The first-level network, sentence-level CNN, consists of one convolutional layer with multiple convolutional kernels in different window sizes, one pooling layer to deal with variable sentence lengths, and one fully connected layer with dropout as well as a softmax output, such that two final steganographic features are obtained for each sentence. The unmodified and modified sentences, along with their words, are represented in the form of pre-trained dense word embeddings, which serve as the input of the network. Sentence-level CNN provides the representation of a sentence, and can thus be utilized to predict whether a sentence is unmodified or has been modified by synonym substitutions. In the second level, a text-level CNN exploits the predicted representations of sentences obtained from the sentence-level CNN to determine whether the detected text is a stego text or cover text. Experimental results indicate that the proposed sentence-level CNN can effectively extract sentence features for sentence-level steganalysis tasks and reaches an average accuracy of 82.245%. Moreover, the proposed steganalysis method achieves greatly improved detection performance when distinguishing stego texts from cover texts.

[Correlation of immunophenotype to cytogenetics and clinical features of adult acute myeloid leukemia].

BACKGROUND & OBJECTIVE: New WHO classification has been rapidly used in diagnosis of leukemia. Based on coexpression and correlation of lineage-associated antigens, multiparameter high-resolution flow cytometry has been developed to precisely identify lineage characteristics of leukemia. Some immunophenotypes correlate with cytogenetic abnormality and prognosis. This study was to analyze immunophenotype of naive acute myeloid leukemia (AML), and explore its correlations to cytogenetics, clinical features, and FAB subtype of AML. METHODS: Multiparameter high-resolution flow cytometry with a panel of 25 different monoclonal antibodies was used to analyze the surface and cytoplasmic antigens expressions of 96 adults with AML; G-binding technique was used to analyze karyotype of 73 of the 96 patients. RESULTS: In these AML patients, some antigens were correlated with FAB subtypes:expression of CD2 was enhanced in AML-M3; HLA-DR, CD34, and CD56 were absent in AML-M3; expression of CD19 was increased in AML-M2; expressions of CD14 and CD56 were enhanced in AML-M5; MPO was absent in AML-M0. Karyotype abnormality was detected in 40(54.8%) patients. CD22, CD56, and TdT expressions were correlated with karyotype abnormality. t(8; 21) was only detected in 10 AML-M2 patients with high expressions of CD15, CD19, CD34, and CD56; no lymphoid lineage antigens were detected in 7 AML-M3 patients with t (15; 17). Expressions of CD4 and TdT were positively correlated with patient's age; expressions of CD7 and CD14 were positively correlated with high white blood cell count; expressions of CD4, CD14, and CD56 were positively correlated with high platelet count. CONCLUSIONS: The abnormal antigen expression of AML is tightly linked with karyotype abnormality. Detection of immunophenotype may help to diagnose and classify AML.

[Immunophenotypes in 115 patients with acute myeloid leukemia by multi-color flow cytometry].

Immunophenotyping has become common in the diagnosis and classification of leukemia. To evaluate the immunophenotype of acute myeloid leukemia (AML), multiparameter flow cytometry and CD45/SSC gating were used to analyze the surface and cytoplasmic antigen expressions in 115 cases of AML. The results were compared with the French-American-British (FAB) Cooperative Group classification to help define the best use and role of multiparameter flow cytometry in the diagnosis and proper classification of AML. The results showed that CD38, CD38 and CD13 were the most commonly expressed antigen (94.8%, 91.3% and 89.6%, respectively). CD7 was the most commonly expressed lymphoid antigen (20.2%), followed by CD19 (16.5%) and CD2 (15%). Some immunophenotypes correlated with FAB type, including increased frequency of CD2 in M(3); lack of HLA-DR, CD34 and CD56 expression in M(3); increased frequency of CD19 in M(2), CD14 and CD56 in M(5) and lack of MPO in M(0). In conclusion, multiparameter flow cytometry is a reliable technique in the diagnosis of AML, and some immunophenotypes correlate with FAB type.