Deep radial basis function networks with subcategorization for mitosis detection in breast histopathology images.

Due to the intra-class diversity of mitotic cells and the morphological overlap with similarly looking imposters, automatic mitosis detection in histopathology slides is still a challenging task. In this paper, we propose a novel mitosis detection model in a weakly supervised way, which consists of a candidate proposal network and a verification network. The candidate proposal network based on patch learning aims to separate both mitotic cells and their mimics from the background as candidate objects, which substantially reduces missed detections in the screening process of candidates. These obtained candidate results are then fed into the verification network for mitosis refinement. The verification network adopts an RBF-based subcategorization scheme to deal with the problems of high intra-class variability of mitosis and the mimics with similar appearance. We utilize the RBF centers to define subcategories containing mitotic cells with similar properties and capture representative RBF center locations through joint training of classification and clustering. Due to the lower intra-class variation within a subcategory, the localized feature space at subcategory level can better characterize a certain type of mitotic figures and can provide a better similarity measurement for distinguishing mitotic cells from nonmitotic cells. Our experiments manifest that this subcategorization scheme helps improve the performance of mitosis detection and achieves state-of-the-art results on the publicly available mitosis datasets using only weak labels.

Role of m6A modifications in immune evasion and immunotherapy.

RNA modification has garnered increasing attention in recent years due to its pivotal role in tumorigenesis and immune surveillance. N6-methyladenosine (m6A) modification is the most prevalent RNA modification, which can affect the expression of RNA by methylating adenylate at the sixth N position to regulate the occurrence and development of tumors. Dysregulation of m6A affects the activation of cancer-promoting pathways, destroys immune cell function, maintains immunosuppressive microenvironment, and promotes tumor cell growth. In this review, we delve into the latest insights into how abnormalities in m6A modification in both tumor and immune cells orchestrate immune evasion through the activation of signaling pathways. Furthermore, we explore how dysregulated m6A modification in tumor cells influences immune cells, thereby regulating tumor immune evasion via interactions within the tumor microenvironment (TME). Lastly, we highlight recent discoveries regarding specific inhibitors of m6A modulators and the encapsulation of m6A-targeting nanomaterials for cancer therapy, discussing their potential applications in immunotherapy.

Palmitic acid-capped MIL-101-Al as a nano-adjuvant to amplify immune responses against Pseudomonas aeruginosa.

As a highly contagious opportunistic pathogen, Pseudomonas aeruginosa (P. aeruginosa) is one of the main causes of healthcare-associated infections. The drug-resistant nature of P. aeruginosa can render antibiotic treatments ineffective, leading to a high morbidity and mortality. Higher specificity and reduced toxicity are features of immunotherapy, which can generate robust immune responses and preserve long-term immunological memory to completely eradicate infections. In this study, we developed a type of P. aeruginosa vaccine based on a metal-organic framework. Specifically, MIL-101-Al nanoparticles were synthesized to encapsulate antigens derived from the bacterial lysate (BL) of PAO1, a drug-resistant P. aeruginosa, and the adjuvant unmethylated cytosine-phosphate-guanine oligonucleotide (CpG), which were then modified with palmitic acid (PAA) to obtain MIL-BC@PAA. The stability and biocompatibility were significantly increased by capping with PAA. Moreover, MIL-BC@PAA showed significantly enhanced uptake by antigen presenting cells (APCs), and promoted their maturation. Importantly, immunity studies revealed the greatly elicited antigen-specific humoral and cellular responses, and a protection rate of about 70% was observed in P. aeruginosa-challenged mice. Overall, these results demonstrate the promising potential of MIL-BC@PAA as an ideal nanovaccine for P. aeruginosa vaccination.

Nuclei detection in breast histopathology images with iterative correction.

This work presents a deep network architecture to improve nuclei detection performance and achieve the high localization accuracy of nuclei in breast cancer histopathology images. The proposed model consists of two parts, generating nuclear candidate module and refining nuclear localization module. We first design a novel patch learning method to obtain high-quality nuclear candidates, where in addition to categories, location representations are also added to the patch information to implement the multi-task learning process of nuclear classification and localization; meanwhile, the deep supervision mechanism is introduced to obtain the coherent contributions from each scale layer. In order to refine nuclear localization, we propose an iterative correction strategy to make the prediction progressively closer to the ground truth, which significantly improves the accuracy of nuclear localization and facilitates neighbor size selection in the nonmaximum suppression step. Experimental results demonstrate the superior performance of our method for nuclei detection on the H&E stained histopathological image dataset as compared to previous state-of-the-art methods, especially in multiple cluttered nuclei detection, can achieve better results than existing techniques.

Functional Connectivity Networks with Latent Distributions for Mild Cognitive Impairment Identification.

This work presents a novel approach to estimate brain functional connectivity networks via generative learning. Due to the complexity and variability of rs-fMRI signal, we consider it as a random variable, and utilize variational autoencoder networks to encode it as a confidence distribution in the latent space rather than as a fixed vector, so as to establish the relationship between them. First, the mean time series of each brain region of interest is mapped into a multivariate Gaussian distribution. The correlation between two brain regions is measured by the Jensen-Shannon divergence that describes the statistical similarity between two probability distributions, and then the adjacency matrix is created to indicate the functional connectivity strength of pairwise brain regions. Meanwhile, our findings show that the adjacency matrices obtained at VAE latent spaces of different dimensionalities have good complementarity for MCI identification in precision and recall, and the classification performance can be further boosted by an efficient cascade of classifiers. This proposal constructs brain functional networks from a statistical modeling standpoint, improving the statistical ability of population data and the generalization ability of observation data variability. We evaluate the proposed framework over the task of identifying subjects with MCI from normal controls, and the experimental results on the public dataset show that our method significantly outperforms both the baseline and current state-of-the-art methods.

RNA modifications in cancer.

Currently, more than 170 modifications have been identified on RNA. Among these RNA modifications, various methylations account for two-thirds of total cases and exist on almost all RNAs. Roles of RNA modifications in cancer are garnering increasing interest. The research on m6A RNA methylation in cancer is in full swing at present. However, there are still many other popular RNA modifications involved in the regulation of gene expression post-transcriptionally besides m6A RNA methylation. In this review, we focus on several important RNA modifications including m1A, m5C, m7G, 2'-O-Me, Ψ and A-to-I editing in cancer, which will provide a new perspective on tumourigenesis by peeking into the complex regulatory network of epigenetic RNA modifications, transcript processing, and protein translation.

BAK plays a key role in A-1331852-induced apoptosis in senescent chondrocytes.

Osteoarthritis occurs when the number of senescent chondrocytes in the joints reaches an intolerable level. The purpose of our study was to explore the therapeutic effect and mechanism of action of A-1331852 in osteoarthritis. Doxorubicin and etoposide were used to induce cell senescence as determined by the cessation of cell proliferation, augmented senescence-associated beta-galactosidase (SA-ß-Gal) staining, and increased p53 expression levels. The CCK-8 cytotoxicity assay and SA-ß-Gal staining demonstrated that Bcl-xL inhibitors could selectively remove senescent chondrocytes without damaging healthy chondrocytes. A-1331852 induced caspase-dependent death of senescent chondrocytes with decreased mitochondrial membrane potential, nuclear concentration, plasma membrane rupture, and PARP cleavage. Most importantly, A-1331852 upregulated BAK expression levels, indicating that BAK plays a key role in the A-1331852-induced apoptosis of senescent chondrocytes. Live-cell fluorescence resonance energy transfer showed that A-1331852 detached the binding of Bcl-xL to BAK and promoted the oligomerization of BAK on the mitochondrial membrane. In conclusion, this study provides the first evidence that A-1331852 selectively promotes apoptosis in senescent chondrocytes by interfering with the interaction between Bcl-xL and BAK.

Bim- and Bax-mediated mitochondrial pathway dominates abivertinib-induced apoptosis and ferroptosis.

Abivertinib (AC) is a novel epidermal growth factor receptor tyrosine kinase inhibitor with highly efficient antitumor activity. Here, we report the capacity of AC to induce both reactive oxygen species (ROS)-dependent apoptosis and ferroptosis in tumor cells. Our data showed that AC induced iron- and ROS-dependent cytotoxicity in MCF7, HeLa, and A549 cell lines. Flow cytometry analyses showed that AC increased ferrous ions and ROS and induced ferroptosis in MCF-7 cells. This was confirmed by the findings that AC not only decreased solute carrier family 7 member 11 (SLC7A11) and glutathione peroxidase 4 (GPX4) expression but also induced iron- and ROS-dependent aggrandized lipid ROS accumulation and plasma membrane damage. Meanwhile, AC induced nuclear condensation and increased ROS-dependent phosphatidylserine (PS) eversion, caspase-3 activation, and cleaved-PARP expression, suggesting that AC also induced ROS-dependent apoptosis. In addition, mitochondrial depletion significantly inhibited AC-induced cytotoxicity, including ferroptosis and apoptosis, indicating the key role of mitochondria in AC-induced ferroptosis and apoptosis. Moreover, knockout of Bim or Bax not only remarkably inhibited AC-induced apoptosis, but also markedly inhibited AC-triggered downregulation of SLC711 and GPX4, accumulation of lipid ROS, and damage to the plasma membrane. This suggests that Bim and Bax act upstream of SLC7A11 and GPX4 to mediate AC-induced ferroptosis. Collectively, AC induces ferroptosis and apoptosis, in which the Bim- and Bax-mediated mitochondrial pathways play a dominant role.

Bak instead of Bax plays a key role in metformin-induced apoptosis s in HCT116 cells.

Metformin (Met) exhibits anticancer ability in various cancer cell lines. This report aims to explore the exact molecular mechanism of Met-induced apoptosis in HCT116 cells, a human colorectal cancer cell line. Met-induced reactive oxygen species (ROS) increase and ROS-dependent cell death accompanied by plasma membrane blistering, mitochondrial swelling, loss of mitochondrial membrane potential, and release of cytochrome c. Western blotting analysis showed that Met upregulated Bak expression but downregulated Bax expression. Most importantly, silencing Bak instead of Bax inhibited Met-induced loss of mitochondrial membrane potential, indicating the key role of Bak in Met-induced apoptosis. Live-cell fluorescence resonance energy transfer (FRET) analysis showed that Met unlocked the binding of Mcl-1 to Bak, and enhanced the binding of Bim to Bak and subsequent Bak homo-oligomerization. Western blotting analysis showed that Met enhanced AMPK phosphorylation and Bim expression, and compound C, an inhibitor of AMPK, inhibited Met-induced Bim upregulation. Although Met increased the expression of Bcl-xL, overexpression of Bcl-xL did not prevent Met-induced apoptosis. In summary, our data demonstrate for the first time that Met promotes ROS-dependent apoptosis by regulating the Mcl-1-Bim-Bak axis.

Evaluating the inhibitory priority of Bcl-xL to Bad, tBid and Bax by using live-cell imaging assay.

Molecular regulatory network among the B cell leukemia-2 (Bcl-2) family proteins is a research hotspot on apoptosis. The inhibitory priority of anti-apoptotic Bcl-2 family proteins (such as Bcl-xL) to pro-apoptotic Bcl-2 family proteins (such as Bad, tBid and Bax) determines the outcome of their interactions. Based on over-expression model system, we here evaluate the inhibitory priority of Bcl-xL to Bad, tBid and Bax by using live-cell imaging assay on cell viability. Fluorescence images of living cells co-expressing CFP-Bcl-xL and YFP-Bad or YFP-tBid or YFP-Bax showed that Bcl-xL markedly inhibited Bad/tBid/Bax-mediated apoptosis, revealing that Bcl-xL inhibits the proapoptotic function of Bad, tBid and Bax. In the case of equimolar co-expression of Bad and CFP-Bcl-xL, the inhibition of Bcl-xL on tBid/Bax mediate-apoptosis was completely relieved. Moreover, co-expression of tBid-P2A-CFP-Bcl-xL significantly relieved the inhibition of Bcl-xL on the pro-apoptotic ability Bax, suggesting that Bcl-xL preferentially inhibits the pro-apoptotic ability of Bad over tBid, subsequently to Bax.

Measuring calibration factors by imaging a dish of cells expressing different tandem constructs plasmids.

Three-cube Förster resonance energy transfer (FRET) method is the most extensively applied approach for live-cell FRET quantification. Reliable measurements of calibration factors are crucial for quantitative FRET measurement. We here proposed a modified TA-G method (termed as mTA-G) to simultaneously obtain the FRET-sensitized quenching transition factor (G) and extinction coefficients ratio (γ) between donor and acceptor. mTA-G method includes four steps: (1) predetermining the ratio ranges of the sensitized emission of acceptor (FC ) to the donor excitation and donor channel image (IDD [(DA])) for all FRET plasmids; (2) culturing the cells which express every FRET plasmid in one dish respectively; (3) distinguishing and marking the cells expressing different FRET plasmids by detecting their FC /IDD (DA) values; (4) linearly fitting FC /IAA (DA) (acceptor excitation and acceptor channel image) to IDD (DA)/IAA (DA) for different kinds of cells. We implemented mTA-G method by imaging tandem constructs cells with different FRET efficiency cultured in one dish on different days, and obtained consistent G and γ values. mTA-G method not only circumvents switchover of different culture dishes but also keep the constant imaging conditions, exhibiting excellent robustness, and thus will expands the biological applications of quantitative FRET analysis in living cells.

Development and validation of a nomogram to predict overall survival for patients with metastatic renal cell carcinoma.

BACKGROUND: Heterogeneity of metastatic renal cell carcinoma (RCC) constraints accurate prognosis prediction of the tumor. We therefore aimed at developing a novel nomogram for accurate prediction of overall survival (OS) of patients with metastatic RCC. METHODS: We extracted 2010 to 2016 data for metastatic RCC patients in the Surveillance, Epidemiology, and End Results (SEER) database, and randomly stratified them equally into training and validation sets. Prognostic factors for OS were analyzed using Cox regression models, and thereafter integrated into a 1, 3 and 5-year OS predictive nomogram. The nomogram was validated using the training and validation sets. The performance of this model was evaluated by the Harrell's concordance index (C-index), calibration curve, integrated discrimination improvement (IDI), category-free net reclassification improvement (NRI), index of prediction accuracy (IPA), and decision curve analysis (DCA). RESULTS: Overall, 2315 metastatic RCC patients in the SEER database who fulfilled our inclusion criteria were utilized in constructing a nomogram for predicting OS of newly diagnosed metastatic RCC patients. The nomogram incorporated eight clinical factors: Fuhrman grade, lymph node status, sarcomatoid feature, cancer-directed surgery and bone, brain, liver, and lung metastases, all significantly associated with OS. The model was superior to the American Joint Committee on Cancer (AJCC) staging system (7th edition) both in training (C-indices, 0.701 vs. 0.612, P < 0.001) and validation sets (C-indices, 0.676 vs. 0.600, P < 0.001). The calibration plots of the nomogram corresponded well between predicted and observed values. NRI, IDI, and IPA further validated the superior predictive capability of the nomogram relative to the AJCC staging system. The DCA plots revealed reliable clinical application of our model in prognosis prediction of metastatic RCC patients. CONCLUSIONS: We developed and validated an accurate nomogram for individual OS prediction of metastatic RCC patients. This nomogram can be applied in design of clinical trials, patient counseling, and rationalizing therapeutic modalities.

Nomogram prediction of overall survival based on log odds of positive lymph nodes for patients with penile squamous cell carcinoma.

PURPOSE: This study aimed to establish a nomogram to predict the long-term overall survival (OS) for patients with penile squamous cell carcinoma (PSCC). METHOD: The PSCC patients receiving regional lymph node dissection (RLND) were enrolled from the Surveillance, Epidemiology, and End Results (SEER) database between 2004 and 2015. The dataset of all eligible patients were used to develop the predictive model. The significant independent predictors were identified through Cox regression modeling based on the Bayesian information criterion and then incorporated into a nomogram to predicted 1-, 3-, and 5-year OS. Internal validation was performed using the bootstrap resampling method. The model performance was evaluated using Harrell's concordance index (C-index), calibration plots, integrated discrimination improvement (IDI), net reclassification improvement (NRI), and decision curve analysis (DCA). RESULTS: Totally, 384 eligible PSCC patients were enrolled from the SEER database. A nomogram for OS prediction was developed, in which three clinical variables significantly associated with OS were integrated, including age, N classification, and log odds of positive lymph nodes (LODDS). The C-index of the nomogram (0.746, 95% CI: 0.702-0.790) was significantly higher than that of the American Joint Committee on Cancer (AJCC) staging system (0.692, 95% CI: 0.646-0.738, P = .020). The bootstrap optimism-corrected C-index for the nomogram was 0.739 (95% CI: 0.690-0.784). The bias-corrected calibration plots showed the predicted risks were in good accordance with the actual risks. The results of NRI, IDI, and DCA exhibited superior predictive capability and higher clinical use of the nomogram compared with the AJCC staging system. CONCLUSION: We successfully constructed a simple and reliable nomogram for OS prediction among PSCC patients receiving RLND, which would be beneficial to clinical trial design, patient counseling, and therapeutic modality selection.

Learning Nonclassical Receptive Field Modulation for Contour Detection.

This work develops a biologically inspired neural network for contour detection in natural images by combining the nonclassical receptive field modulation mechanism with a deep learning framework. The input image is first convolved with the local feature detectors to produce the classical receptive field responses, and then a corresponding modulatory kernel is constructed for each feature map to model the nonclassical receptive field modulation behaviors. The modulatory effects can activate a larger cortical area and thus allow cortical neurons to integrate a broader range of visual information to recognize complex cases. Additionally, to characterize spatial structures at various scales, a multiresolution technique is used to represent visual field information from fine to coarse. Different scale responses are combined to estimate the contour probability. Our method achieves state-of-the-art results among all biologically inspired contour detection models. This study provides a method for improving visual modeling of contour detection and inspires new ideas for integrating more brain cognitive mechanisms into deep neural networks.

Inhibition of HDAC6 expression decreases brain injury induced by APOE4 and Aß co­aggregation in rats.

The present study aimed to explore the effects of histone deacetylase 6 (HDAC6) on brain injury in rats induced by apolipoprotein E4 (APOE4) and amyloid ß protein alloform 1­40 (Aß1­40) copolymerization. The rats were randomly divided into four groups: Control group, sham group, APOE4 + Aß1­40 co­injection group (model group) and HDAC6 inhibitor group (HDAC6 group). The brain injury model was established by co­injection of APOE4 + Aß1­40. Morris water maze experiment was used to observe the spatial memory and learning the ability of rats. Histological changes of the hippocampus were observed by hematoxylin and eosin staining. The mRNA expression levels of choline acetyltransferase (ChAT) and HDAC6 were detected by reverse transcription­quantitative PCR. Immunohistochemistry was used to detect the protein expression of HDAC6. Western blotting was used to detect the protein expression levels of HDAC6, microtubule­associated protein tau and glycogen synthase kinase 3ß (GSK3ß). APOE4 and Aß1­40 co­aggregation decreased the short­term spatial memory and learning ability of rats, whereas inhibition of HDAC6 activity attenuated the injury. Inhibition of HDAC6 activity resulted in an attenuation of the APOE4 and Aß1­40 co­aggregation­induced increase in the number of dysplastic hippocampal cells. Further experiments demonstrated that APOE4 and Aß1­40 co­aggregation decreased the expression levels of ChAT mRNA, and the phosphorylation levels of tau GSK3ß protein in the hippocampus, whereas inhibition of HDAC6 activity resulted in increased expression of ChAT mRNA, tau protein and GSK3ß phosphorylation. The inhibition of HDAC6 activity was also demonstrated to reduce brain injury induced by APOE4 and Aß1­40 co­aggregation in model rats.

Risk factors associated with suicide among kidney cancer patients: A Surveillance, Epidemiology, and End Results analysis.

BACKGROUND: The suicide risk was higher in kidney cancer patients than in the general population. The purpose of this study was to characterize the suicide rates among kidney cancer patients and to identify the potential risk factors associated with suicide from the Surveillance, Epidemiology, and End Results (SEER) database. METHODS: Kidney cancer patients were identified from the SEER database during 1973-2015. Suicide rates and standardized mortality ratios (SMRs) of this population were calculated, and the US general population during 1981-2015 was chosen as a reference. Univariable and multivariable Cox regression were performed to find out potential risk factors of suicide. RESULTS: There were 207 suicides identified among 171 819 individuals with kidney cancer observed for 948 272 person-years. The suicide rate was 21.83 per 100 000 person-years, and SMR was 1.83 (95% CI: 1.59-2.10). On Cox regression, diagnosis in early years (1973-1982 vs 2003-2015, HR: 2.03, 95% CI: 1.01-4.11, P = 0.048; 1983-1992 vs 2003-2015, HR: 1.99, 95% CI: 1.18-3.35, P = 0.010), male sex (vs female sex, HR: 4.43, 95% CI: 2.95-6.65, P < 0.001), unmarried status (vs married status, HR: 2.54, 95% CI: 1.91-3.38, P < 0.001), non-black race (white race vs black race, HR: 4.47, 95% CI: 2.09-9.58, P < 0.001; other races vs black race, HR: 3.01, 95% CI: 1.08-8.37, P = 0.035), higher histologic grade (grade IV vs grade I, HR: 3.27, 95% CI: 1.50-7.13, P = 0.003; grade III vs grade I, HR: 2.13, 95% CI: 1.19-3.81, P = 0.011) and cancer-directed surgery not performed (vs performed, HR: 2.78, 95% CI: 1.52-5.11, P < 0.001) were independent risk factors of suicide among kidney cancer patients. CONCLUSIONS: Diagnosis in early years, male sex, unmarried status, non-black race, higher histologic grade, and cancer-directed surgery not performed were significantly associated with suicide among kidney cancer patients. In order to prevent suicidal death, clinicians should pay more attention to patients with high-risk factors of suicide.

Long non-coding RNA CRNDE promotes heptaocellular carcinoma cell proliferation by regulating PI3K/Akt /ß-catenin signaling.

Hepatocellular carcinoma (HCC) is one of the common malignant cancer worldwide, and its molecular pathogenesis remains elusive and recently long non-coding RNAs (lncRNAs) have been reported that play divergent roles in HCC tumorigenesis and development. In current study, we found a lncRNA, CRNDE is more commonly up-regulated in HCC malignant tissues and associated with poor clinical outcomes. Furthermore, both loss- and gain-functions assays revealed that CRNDE promotes HCC cell proliferation and growth in vitro and in vivo. In addition, we found that CRNDE regulates PI3K/Akt/GSK3ß-Wnt/ß-catenin axis to exert its oncogenic role in HCC cell proliferation and growth. In conclusion, our findings enlarged our knowledge in the roles of lncRNAs in the progression of HCC and CRNDE as a lncRNA could be a novel target for HCC treatment.

A preliminary study on the anticancer efficacy of Caulis spatholobi compound 1802.

In recent years, antitumor and antiviral effect of Caulis spatholobi becomes a hot topic of medical drug research. Experiment shows that the water extract of Caulis spatholobi compound 1802 showed the effect of inhibiting tumor growth, the difference was statistically significant (P<0.05); the rate of tumor inhibition was highest in the high dose group of compound 1802, which could reach 41.99%. Anti tumor drugs generally have high toxicity, therefore, low toxicity is the significant characteristic of Caulis spatholobi. In particular the use of Caulis spatholobi has certain practical significance for development of tumor patients' daily diet products. In summary, the chemical constituents of Caulis spatholobi complex, has extensive pharmacological effects and clinical application.

Computational Model Based on Neural Network of Visual Cortex for Human Action Recognition.

In this paper, we propose a bioinspired model for human action recognition through modeling neural mechanisms of information processing in two visual cortical areas: the primary visual cortex (V1) and the middle temporal cortex (MT) dedicated to motion. This model, named V1-MT, is composed of V1 and MT models (layers) corresponding to their cortical areas, which are built with layered spiking neural networks (SNNs). Some neuron properties in V1 and MT, such as direction and speed selectivity, spatiotemporal inseparability, and center surround suppression, are integrated into SNNs. Based on speed and direction selectivity, V1 and MT models contain multiple SNN channels, each of which processes motion information in sequences with spatiotemporal tunings of neurons at a certain speed and different directions. Therefore, we propose two operations, input signal perceiving with 3-D Gabor filters and surround inhibition processing with 3-D differences of Gaussian functions, to perform this task according to the spatiotemporal inseparability and center surround suppression of neurons. Then, neurons are modeled with our simplified integrate-and-fire model and motion information is transformed into spike trains. Afterward, we define a new feature vector: a mean motion map computed from spike trains in all channels to represent human actions. Finally, a support vector machine is trained to classify actions represented by the feature vectors. We conducted extensive experiments on public action databases, and the results show that our model outperforms other bioinspired models and rivals the state-of-the-art approaches.

Medical Image Retrieval Using Multi-Texton Assignment.

In this paper, we present a multi-texton representation method for medical image retrieval, which utilizes the locality constraint to encode each filter bank response within its local-coordinate system consisting of the k nearest neighbors in texton dictionary and subsequently employs spatial pyramid matching technique to implement feature vector representation. Comparison with the traditional nearest neighbor assignment followed by texton histogram statistics method, our strategies reduce the quantization errors in mapping process and add information about the spatial layout of texton distributions and, thus, increase the descriptive power of the image representation. We investigate the effects of different parameters on system performance in order to choose the appropriate ones for our datasets and carry out experiments on the IRMA-2009 medical collection and the mammographic patch dataset. The extensive experimental results demonstrate that the proposed method has superior performance.