Search

Cancer originates from the uncontrolled growth of healthy cells into a mass. Chromophores, such as hemoglobin and melanin, characterize skin spectral properties, allowing the classification of lesions into different etiologies. Hyperspectral imaging systems gather skin-reflected and transmitted light into several wavelength ranges of the electromagnetic spectrum, enabling potential skin-lesion differentiation through machine learning algorithms. Challenged by data availability and tiny inter and intra-tumoral variability, here we introduce a pipeline based on deep neural networks to diagnose hyperspectral skin cancer images, targeting a handheld device equipped with a low-power graphical processing unit for routine clinical testing. Enhanced by data augmentation, transfer learning, and hyperparameter tuning, the proposed architectures aim to meet and improve the well-known dermatologist-level detection performances concerning both benign-malignant and multiclass classification tasks, being able to diagnose hyperspectral data considering real-time constraints. Experiments show 87% sensitivity and 88% specificity for benign-malignant classification and specificity above 80% for the multiclass scenario. AUC measurements suggest classification performance improvement above 90% with adequate thresholding. Concerning binary segmentation, we measured skin DICE and IOU higher than 90%. We estimated 1.21 s, at most, consuming 5 Watts to segment the epidermal lesions with the U-Net++ architecture, meeting the imposed time limit. Hence, we can diagnose hyperspectral epidermal data assuming real-time constraints.

Accurate DNA replication and segregation is key to reproduction and cell viability in all organisms. Autonomously replicating sequence-binding factor 1 (Abf1) is a multifunctional protein that has essential roles in replication, transcription, and regional silencing in the model yeast Saccharomyces cerevisiae. In the opportunistic pathogenic fungus Candida glabrata, which is closely related to S. cerevisiae, these processes are important for survival within the host, for example, the regulation of transcription of virulence-related genes like those involved in adherence. Here, we describe that CgABF1 is an essential gene required for cell viability and silencing near the telomeres, where many adhesin-encoding genes reside. CgAbf1 mediated subtelomeric silencing depends on the 43 C-terminal amino acids. We also found that abnormal expression, depletion, or overexpression of Abf1, results in defects in nuclear morphology, nuclear segregation, and transit through the cell cycle. In the absence of ABF1, cells are arrested in G2 but start cycling again after 9 h, coinciding with the loss of cell viability and the appearance of cells with higher DNA content. Overexpression of CgABF1 causes defects in nuclear segregation and cell cycle progression. We suggest that these effects could be due to the deregulation of DNA replication.

Candida albicans, Candida glabrata, Candida parapsilosis and Candida tropicalis are the four most common human fungal pathogens isolated that can cause superficial and invasive infections. It has been shown that specific metabolites present in the secretomes of these fungal pathogens are important for their virulence. C. glabrata is the second most common isolate world-wide and has an innate resistance to azoles, xenobiotics and oxidative stress that allows this fungal pathogen to evade the immune response and persist within the host. Here, we analyzed and compared the C. glabrata secretome with those of C. albicans, C. parapsilosis, C. tropicalis and the non-pathogenic yeast Saccharomyces cerevisiae. In C. glabrata, we identified a different number of metabolites depending on the growth media: 12 in synthetic complete media (SC), 27 in SC-glutamic acid and 23 in rich media (YPD). C. glabrata specific metabolites are 1-dodecene (0.09 ± 0.11%), 2,5-dimethylundecane (1.01 ± 0.19%), 3,7-dimethyldecane (0.14 ± 0.15%), and octadecane (0.4 ± 0.53%). The metabolites that are shared with C. albicans, C. glabrata, C. parapsilosis, C. tropicalis and S. cerevisiae are phenylethanol, which is synthesized from phenylalanine, and eicosane and nonanoic acid (identified as trimethylsilyl ester), which are synthesized from fatty acid metabolism. Phenylethanol is the most abundant metabolite in all fungi tested: 26.36 ± 17.42% (C. glabrata), 46.77 ± 15.58% (C. albicans), 49.76 ± 18.43% (C. tropicalis), 5.72 ± 0.66% (C. parapsilosis.) and 44.58 ± 27.91% (S. cerevisiae). The analysis of C. glabrata’s secretome will allow us to further our understanding of the possible role these metabolites could play in its virulence.

This paper shows new contributions in the detection of skin cancer, where we present the use of a customized hyperspectral system that captures images in the spectral range from 450 to 950 nm. By choosing a 7 × 7 sub-image of each channel in the hyperspectral image (HSI) and then taking the mean and standard deviation of these sub-images, we were able to make fits of the resulting curves. These fitted curves had certain characteristics, which then served as a basis of classification. The most distinct fit was for the melanoma pigmented skin lesions (PSLs), which is also the most aggressive malignant cancer. Furthermore, we were able to classify the other PSLs in malignant and benign classes. This gives us a rather complete classification method for PSLs with a novel perspective of the classification procedure by exploiting the variability of each channel in the HSI.

Candida glabrata is the second most common cause of candidemia, and its ability to adhere to different host cell types, to microorganisms, and to medical devices are important virulence factors. Here, we consider three characteristics that confer extraordinary advantages to C. glabrata within the host. (1) C. glabrata has a large number of genes encoding for adhesins most of which are localized at subtelomeric regions. The number and sequence of these genes varies substantially depending on the strain, indicating that C. glabrata can tolerate high genomic plasticity; (2) The largest family of CWPs (cell wall proteins) is the EPA (epithelial adhesin) family of adhesins. Epa1 is the major adhesin and mediates adherence to epithelial, endothelial and immune cells. Several layers of regulation like subtelomeric silencing, cis-acting regulatory regions, activators, nutritional signaling, and stress conditions tightly regulate the expression of many adhesin-encoding genes in C. glabrata, while many others are not expressed. Importantly, there is a connection between acquired resistance to xenobiotics and increased adherence; (3) Other subfamilies of adhesins mediate adherence to Candida albicans, allowing C. glabrata to efficiently invade the oral epithelium and form robust biofilms. It is noteworthy that every C. glabrata strain analyzed presents a unique pattern of CWPs at the cell surface.