Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural network

Background

Ocular images play an essential role in ophthalmological diagnoses. Having an imbalanced dataset is an inevitable issue in automated ocular diseases diagnosis; the scarcity of positive samples always tends to result in the misdiagnosis of severe patients during the classification task. Exploring an effective computer-aided diagnostic method to deal with imbalanced ophthalmological dataset is crucial.

Methods

In this paper, we develop an effective cost-sensitive deep residual convolutional neural network (CS-ResCNN) classifier to diagnose ophthalmic diseases using retro-illumination images. First, the regions of interest (crystalline lens) are automatically identified via twice-applied Canny detection and Hough transformation. Then, the localized zones are fed into the CS-ResCNN to extract high-level features for subsequent use in automatic diagnosis. Second, the impacts of cost factors on the CS-ResCNN are further analyzed using a grid-search procedure to verify that our proposed system is robust and efficient.

Results

Qualitative analyses and quantitative experimental results demonstrate that our proposed method outperforms other conventional approaches and offers exceptional mean accuracy (92.24%), specificity (93.19%), sensitivity (89.66%) and AUC (97.11%) results. Moreover, the sensitivity of the CS-ResCNN is enhanced by over 13.6% compared to the native CNN method.

Conclusion

Our study provides a practical strategy for addressing imbalanced ophthalmological datasets and has the potential to be applied to other medical images. The developed and deployed CS-ResCNN could serve as computer-aided diagnosis software for ophthalmologists in clinical application.

     

Active use of the metapleural glands by ants in controlling fungal infection

Insect societies face constant challenges from disease agents. Ants deploy diverse antimicrobial compounds against pathogens and the key sources are metapleural glands (MGs). Are MG products passively secreted and used indiscriminately or are they selectively used when ants are challenged by pathogens? In 26 species from five subfamilies, ants use foreleg movements to precisely groom the MG opening. In the absence of experimental infection, MG grooming rates are low and workers groom themselves after contacting the MGs. The derived leaf-cutter ants (Atta and Acromyrmex) also groom their fungal gardens, substrata (leaves), queens and nest-mates after MG grooming. Atta respond to a challenge by fungal conidia by increasing the rate of MG grooming, but do not do so when an inert powder is applied. This increase occurs in the first hour after a potential infection, after which it returns to baseline levels. Ants with open MGs produce more infrabuccal pellets (IP) than ants with sealed MGs and conidia within pellets from the former are less likely to germinate. Thus, ants selectively groom their MGs when disease agents are present, suggesting that they also selectively use their MG secretions, which has important implications for understanding the evolution of hygienic behaviour in social groups.     

Comparison of different neural network algorithms in the diagnosis of acute appendicitis

Four different neural network algorithms, binary adaptive resonance theory (ART1), self-organizing map, learning vector quantization and back-propagation, were compared in the diagnosis of acute appendicitis with different parameter groups. The results show that supervised learning algorithms learning vector quantization and back-propagation were better than unsupervised algorithms in this medical decision making problem. The best results were obtained with the learning vector quantization. The self-organizing map algorithm showed good specificity, but this was in conjunction with lower sensitivity. The best parameter group was found to be the clinical signs. It seems beneficial to design a decision support system which uses these methods in the decision making process.     

Segmentation of organs-at-risk in cervical cancer CT images with a convolutional neural network

PurposeWe introduced and evaluated an end-to-end organs-at-risk (OARs) segmentation model that can provide accurate and consistent OARs segmentation results in much less time.MethodsWe collected 105 patients’ Computed Tomography (CT) scans that diagnosed locally advanced cervical cancer and treated with radiotherapy in one hospital. Seven organs, including the bladder, bone marrow, left femoral head, right femoral head, rectum, small intestine and spinal cord were defined as OARs. The annotated contours of the OARs previously delineated manually by the patient’s radiotherapy oncologist and confirmed by the professional committee consisted of eight experienced oncologists before the radiotherapy were used as the ground truth masks. A multi-class segmentation model based on U-Net was designed to fulfil the OARs segmentation task. The Dice Similarity Coefficient (DSC) and 95th Hausdorff Distance (HD) are used as quantitative evaluation metrics to evaluate the proposed method.ResultsThe mean DSC values of the proposed method are 0.924, 0.854, 0.906, 0.900, 0.791, 0.833 and 0.827 for the bladder, bone marrow, femoral head left, femoral head right, rectum, small intestine, and spinal cord, respectively. The mean HD values are 5.098, 1.993, 1.390, 1.435, 5.949, 5.281 and 3.269 for the above OARs respectively.ConclusionsOur proposed method can help reduce the inter-observer and intra-observer variability of manual OARs delineation and lessen oncologists’ efforts. The experimental results demonstrate that our model outperforms the benchmark U-Net model and the oncologists’ evaluations show that the segmentation results are highly acceptable to be used in radiation therapy planning.     

The use of polymerase chain reaction in the diagnosis of diphtheria infection

624 Corynebacterium diphtheriae strains, newly isolated from patients and carriers, were studied with the use of the methods of gel immunodiffusion (Elek's test) and polymerase chain reaction (PCR). In the evaluation of 388 C. diptheriae strains, found to be toxigenic in PCR, the results of Elek's test coincided with those of PCR on 98% of cases. In 38 out of 143 strains (26.5%), nontoxigenic according to the results of Elek's test, the presence of the A-fragment of the tox-gene was established. Subculturing in nutrient media made it possible to determine the presence of toxin in 19 out of 38 of these strains; the remaining strains, isolated mainly from carriers, were found to have the "silent" gene. The advantage of using PCR for the detection of toxigenic and nontoxigenic C. diphtheriae strains of different origin was shown.     

Automated assessment of breast cancer margin in optical coherence tomography images via pretrained convolutional neural network

The benchmark method for the evaluation of breast cancers involves microscopic testing of a hematoxylin and eosin (H&E)‐stained tissue biopsy. Resurgery is required in 20% to 30% of cases because of incomplete excision of malignant tissues. Therefore, a more accurate method is required to detect the cancer margin to avoid the risk of recurrence. In the recent years, convolutional neural networks (CNNs) has achieved excellent performance in the field of medical images diagnosis. It automatically extracts the features from the images and classifies them. In the proposed study, we apply a pretrained Inception‐v3 CNN with reverse active learning for the classification of healthy and malignancy breast tissue using optical coherence tomography (OCT) images. This proposed method attained the sensitivity, specificity and accuracy is 90.2%, 91.7% and 90%, respectively, with testing datasets collected from 48 patients (22 normal fibro‐adipose tissue and 26 Invasive ductal carcinomas cancerous tissues). The trained network utilizes for the breast cancer margin assessment to predict the tumor with negative margins. Additionally, the network output is correlated with the corresponding histology image. Our results lay the foundation for the future that the proposed method can be used to perform automatic intraoperative identification of breast cancer margins in real‐time and to guide core needle biopsies.      

EMG scanning in the diagnosis of chronic pain

A surface EMG diagnostic protocol was developed to assess the neuromuscular/postural contributions to pain states. The EMG activity of the right and left aspects of 11 muscle groups were monitored while the patient was in the sitting and standing positions. The diagnostic protocol was evaluated by comparing the patterns of EMG activity in four diagnostic groups: headache only, neck/shoulder/upper back pain only, low back pain only, and mixed pain states. The results suggest that (1) bilateral levels of EMG activity in the frontalis and masseter groups are of primary importance for the headache patients, (2) the discrepancy between the right and left EMG activity in the lumbar and cervical paraspinal muscle groups are of primary importance for low back pain patients, (3) position (sit/stand) may provide important diagnostic information, and (4) the data appear to support the notion of a postural disturbance as a contributing factor in low back pain.     

Ecological factors in the inundative use of fungal entomopathogens

Fungal entomopathogens have been developed in numerous countries as biocontrol agents with more than 100 mycoinsecticide products commercially available in 2006. The chief, perhaps sole, use of these mycoinsecticides has been as inundative agents, within a chemical paradigm. Large numbers of propagules are applied in an attempt to overwhelm by brute force many of the factors that keep a pathogen in nonepizootic equilibrium with its host. This review attempts to summarize what we know about the abiotic and biotic factors that affect the efficacy of these mycoinsecticides in both foliar and soil applications. Sunlight, humidity, temperature, and phylloplane-associated factors can affect both immediate efficacy and persistence on plants. Likewise, soil texture-moisture interactions, temperature, and a host of biotic factors can affect mycoinsecticides in the soil. Despite much research, our understanding of these ecological aspects is imperfect, especially in a holistic, dynamic sense.     

Non-invasive observation of the development of fungal infection in fruit

Summary The progression of infection caused by a fungal pathogen,Botrytis cinerea Pers.: Fr., in a fruit of red raspberry (Rubus idaeus L.) was followed by nuclear magnetic resonance (NMR) microscopic imaging over a 4 day period. It was found that a standard gradient echo sequence discriminated clearly between infected and healthy tissue as mycelium spread across the fruit from a single woundinoculated drupelet.     

The use of the latex agglutination reaction for the diagnosis of a Brucella infection

The authors have developed the optimum conditions for the preparation of antigenic diagnosticum based on latex manufactured in the USSR. To sensitize latex with the diameter of microspheres equal to 0.83 microns, Brucella polysaccharide was used in a dose of 100 micrograms/ml. As stabilizer, polyvinylpyrrolidone at a concentration of 0.1% was used. The specificity and sensitivity of the diagnosticum were studied in analysis of serum samples taken from 102 healthy donors and patients with infectious diseases of nonbrucellar etiology and from 120 patients with different forms of brucellosis. The specificity of the diagnosticum was found to be 94.1% and its sensitivity, 77.5%. Comparative study of the latex agglutination test with other serological tests showed that the former test is highly effective both in acute and chronic forms of the disease. A high degree of correlation between the agglutination test, Coombs' test, the passive hemagglutination test and the latex agglutination test was established (r = 0.83, 0.72 and 0.62, respectively).     

Resolution of batch variations in pyrolysis mass spectrometry of bacteria by the use of artificial neural network analysis

A simple, but stringent, three group model of bacterial interstrain identity (two cultures of the same strain ofEscherichia coli) and difference (a culture of a serologically distinct strain) was used in multiple serial weekly subcultures for five weeks to demonstrate the effect of both growth-related (phenotypic) and machine-related variation on pyrolysis mass spectra. An aliquot of serum from a single sample was included in each pyrolysis batch to distinguish machine drift from culture drift. Conventional principal component (PC) canonical variate (CV) analysis was successful within each pyrolysis batch but the variations between batches precluded the use of data from more than one batch in successful PCCV analysis. In contrast, artificial neural networks (ANNs) trained with data from one batch could be successfully used to identify groups in data from non-contemporaneous pyrolysis batches. Although the ANN method will require validation in more complex settings than this simple model, it is a promising approach to the problem of batch constraint in pyrolysis mass spectrometry.     

Optimized BlaM-transposon shuttle mutagenesis of Helicobacter pylori allows the identification of novel genetic loci involved in bacterial virulence

Helicobacter pylori is an important etiologic agent of gastroduodenal disease in humans. In this report, we describe a general genetic approach for the identification of genes encoding exported proteins in H. pylori. The novel TnMax9 mini-blaM transposon was used for insertion mutagenesis of a H. pylori gene library established in Escherichia coli. A total of 192 E. coli clones expressing active β-lactamase fusion proteins (BlaM⁺) were obtained, indicating that the corresponding target plasmids carry H. pylori genes encoding putative extracytoplasmic proteins. Natural transformation of H. pylori P1 or P12 using the 192 mutant plasmids resulted in 135 distinct H. pylori mutant strains (70%). Screening of the H. pylori collection of mutant strains allowed the identification of mutant strains impaired in motility, in natural transformation competence and in adherence to gastric epithelial cell lines. Motility mutants could be grouped into distinct classes: (i) mutant strains lacking the major flagellin subunit FlaA and intact flagella (class I); (ii) mutant strains with apparently normal flagella, but reduced motility (class II), and (iii) mutant strains with obviously normal flagella, but completely abolished motility (class III). Two independent mutations that exhibited defects in natural competence for genetic transformation mapped to different genetic loci. In addition, two independent mutant strains were isolated by their failure to bind to the human gastric carcinoma cell line Katoill. Both mutant strains carried a transposon in the same gene, 0.8 kb apart, and showed decreased autoagglutination when compared to the wild-type strain.     

The role of immunostimulation in the treatment of invasive fungal infection

Invasive fungal infections (IFI) constitute a severe and increasing problem in an expanding patient population with immunodeficiency due to various reasons. Despite current management strategies, IFI remain a common and frequently devastating problem. New antifungals have been introduced in the medical armamentarium, but the results in patients’ survival are not yet optimal. The compromised host defense mechanisms of these patients contribute to the high failure rates. Immunoregulation includes strategies for restoration or augmentation of impaired immune responses that increase phagocytes’ number or enhance their function. Although there are extensive preclinical data, clinical investigations are very limited. Administration of certain beneficial cytokines, antibodies and other immune molecules, granulocyte transfusions, and stimulation of adaptive immunity through vaccination are potential, promising future preventive and therapeutic modalities and could be adjuncts to antifungal therapy for IFI. This review highlights and comments on preclinical and clinical studies published since 2005.