Deep Learning of Cancer Stem Cell Morphology.

Knowledge regarding cancer stem cell (CSC) morphology is limited, and more extensive studies are therefore required. Image recognition technologies using artificial intelligence (AI) require no previous expertise in image annotation. Herein, we describe the construction of AI models that recognize the CSC morphology in cultures and tumor tissues. The visualization of the AI deep learning process enables insight to be obtained regarding unrecognized structures in an image.

Deep Learning of Phase-Contrast Images of Cancer Stem Cells Using a Selected Dataset of High Accuracy Value Using Conditional Generative Adversarial Networks.

Artificial intelligence (AI) technology for image recognition has the potential to identify cancer stem cells (CSCs) in cultures and tissues. CSCs play an important role in the development and relapse of tumors. Although the characteristics of CSCs have been extensively studied, their morphological features remain elusive. The attempt to obtain an AI model identifying CSCs in culture showed the importance of images from spatially and temporally grown cultures of CSCs for deep learning to improve accuracy, but was insufficient. This study aimed to identify a process that is significantly efficient in increasing the accuracy values of the AI model output for predicting CSCs from phase-contrast images. An AI model of conditional generative adversarial network (CGAN) image translation for CSC identification predicted CSCs with various accuracy levels, and convolutional neural network classification of CSC phase-contrast images showed variation in the images. The accuracy of the AI model of CGAN image translation was increased by the AI model built by deep learning of selected CSC images with high accuracy previously calculated by another AI model. The workflow of building an AI model based on CGAN image translation could be useful for the AI prediction of CSCs.

Non-Gaussian model-based diffusion-weighted imaging of oral squamous cell carcinoma: associations with Ki-67 proliferation status.

OBJECTIVES: To investigate possible associations between diffusion-weighted imaging (DWI) parameters derived from a non-Gaussian model fitting and Ki-67 status in patients with oral squamous cell carcinoma (OSCC). METHODS: Twenty-four patients with newly diagnosed OSCC were prospectively recruited. DWI was performed using six b-values (0-2500). The diffusion-related parameters of kurtosis value (K), kurtosis-corrected diffusion coefficient (DK), diffusion heterogeneity (α), distributed diffusion coefficient (DDC), slow diffusion coefficient (Dslow), and apparent diffusion coefficient (ADC) were calculated from four diffusion fitting models. Ki-67 status was categorized as low (Ki-67 percentage score < 20%), middle (20-50%), or high (> 50%). Kruskal-Wallis tests were performed between each non-Gaussian diffusion model parameters and Ki-67 grade. RESULTS: The Kruskal-Wallis tests revealed that multiple parameters (K, ADC, Dk, DDC and Dslow) showed statistically significant differences between the three levels of Ki-67 status (K: p = 0.020, ADC: p = 0.012, Dk: p = 0.027, DDC: p = 0.007 and Dslow: p = 0.026). CONCLUSIONS: Several non-Gaussian diffusion model parameters and ADC values were significantly associated with Ki-67 status and have potential as promising prognostic biomarkers in patients with OSCC.

Establishment of a New Qualitative Evaluation Method for Articular Cartilage by Dynamic T2w MRI Using a Novel Contrast Medium as a Water Tracer.

OBJECTIVE: In the early stages of cartilage damage, diagnostic methods focusing on the mechanism of maintaining the hydrostatic pressure of cartilage are thought to be useful. 17O-labeled water, which is a stable isotope of oxygen, has the advantage of no radiation exposure or allergic reactions and can be detected by magnetic resonance imaging (MRI). This study aimed to evaluate MRI images using 17O-labeled water in a rabbit model. DESIGN: Contrast MRI with 17O-labeled water and macroscopic and histological evaluations were performed 4 and 8 weeks after anterior cruciate ligament transection surgery in rabbits. A total of 18 T2-weighted images were acquired, and 17O-labeled water was manually administered on the third scan. The 17O concentration in each phase was calculated from the signal intensity at the articular cartilage. Macroscopic and histological grades were evaluated and compared with the 17O concentration. RESULTS: An increase in 17O concentration in the macroscopic and histologically injured areas was observed by MRI. Macroscopic evaluation showed that the 17O concentration significantly increased in the damaged site group. Histological evaluations also showed that 17O concentrations significantly increased at 36 minutes 30 seconds after initiating MRI scanning in the Osteoarthritis Research Society International (OARSI) grade 3 (0.493 in grade 0, 0.659 in grade 1, 0.4651 in grade 2, and 0.9964 in grade 3, P < 0.05). CONCLUSION: 17O-labeled water could visualize earlier articular cartilage damage, which is difficult to detect by conventional methods.

The utility of diffusion-weighted T2 mapping for the prediction of histological tumor grade in patients with head and neck squamous cell carcinoma.

Background: In head and neck cancers, histopathological information is important for the determination of the tumor characteristics and for predicting the prognosis. The aim of this study was to assess the utility of diffusion-weighted T2 (DW-T2) mapping for the evaluation of tumor histological grade in patients with head and neck squamous cell carcinoma (SCC). Methods: The cases of 41 patients with head and neck SCC (21 well/moderately and 17 poorly differentiated SCC) were retrospectively analyzed. All patients received MR scanning using a 3-Tesla MR unit. The conventional T2 value, DW-T2 value, ratio of DW-T2 value to conventional T2 value, and apparent diffusion coefficient (ADC) were calculated using signal information from the DW-T2 mapping sequence with a manually placed region of interest (ROI). Results: ADC values in the poorly differentiated SCC group were significantly lower than those in the moderately/well differentiated SCC group (P<0.05). The ratio of DW-T2 value to conventional T2 value was also significantly different between poorly and moderately/well differentiated SCC groups (P<0.01). Receiver operating characteristic (ROC) curve analysis of ADC values showed a sensitivity of 0.76, specificity of 0.67, positive predictive value (PPV) of 0.62, negative predictive value (NPV) of 0.8, accuracy of 0.71 and area under the curve (AUC) of 0.73, whereas the ROC curve analysis of the ratio of DW-T2 value to conventional T2 value showed a sensitivity of 0.76, specificity of 0.83, PPV of 0.76, NPV of 0.83, accuracy of 0.8 and AUC of 0.82. Conclusions: DW-T2 mapping might be useful as supportive information for the determination of tumor histological grade in patients with head and neck SCC.

Amide proton transfer imaging for the determination of human papillomavirus status in patients with oropharyngeal squamous cell carcinoma.

The aim of this study was to investigate the utility of amide proton transfer (APT) imaging for the determination of human papillomavirus (HPV) status in patients with oropharyngeal squamous cell carcinoma (SCC). Thirty-one patients with oropharyngeal SCC were retrospectively evaluated. All patients underwent amide proton transfer imaging using a 3T magnetic resonance (MR) unit. Patients were divided into HPV-positive and -negative groups depending on the pathological findings in their primary tumor. In APT imaging, the primary tumor was delineated with a polygonal region of interest (ROI). Signal information in the ROI was used to calculate the mean, standard deviation (SD) and coefficient of variant (CV) of the APT signals (APT mean, APT SD, and APT CV, respectively). The value of APT CV in the HPV-positive group (0.43 ±â€…0.04) was significantly lower than that in the HPV-negative group (0.48 ±â€…0.04) (P = .01). There was no significant difference in APT mean (P = .82) or APT SD (P = .13) between the HPV-positive and -negative groups. Receiver operating characteristic (ROC) curve analysis of APT CV had a sensitivity of 0.75, specificity of 0.8, positive predictive value of 0.75, negative predictive value of 0.8, accuracy of 0.77 and area under the curve (AUC) of 0.8. The APT signal in the HPV-negative group was considered heterogeneous compared to the HPV-positive group. This information might be useful for the determination of HPV status in patients with oropharyngeal SCC.

Temporal and Locational Values of Images Affecting the Deep Learning of Cancer Stem Cell Morphology.

Deep learning is being increasingly applied for obtaining digital microscopy image data of cells. Well-defined annotated cell images have contributed to the development of the technology. Cell morphology is an inherent characteristic of each cell type. Moreover, the morphology of a cell changes during its lifetime because of cellular activity. Artificial intelligence (AI) capable of recognizing a mouse-induced pluripotent stem (miPS) cell cultured in a medium containing Lewis lung cancer (LLC) cell culture-conditioned medium (cm), miPS-LLCcm cell, which is a cancer stem cell (CSC) derived from miPS cell, would be suitable for basic and applied science. This study aims to clarify the limitation of AI models constructed using different datasets and the versatility improvement of AI models. The trained AI was used to segment CSC in phase-contrast images using conditional generative adversarial networks (CGAN). The dataset included blank cell images that were used for training the AI but they did not affect the quality of predicting CSC in phase contrast images compared with the dataset without the blank cell images. AI models trained using images of 1-day culture could predict CSC in images of 2-day culture; however, the quality of the CSC prediction was reduced. Convolutional neural network (CNN) classification indicated that miPS-LLCcm cell image classification was done based on cultivation day. By using a dataset that included images of each cell culture day, the prediction of CSC remains to be improved. This is useful because cells do not change the characteristics of stem cells owing to stem cell marker expression, even if the cell morphology changes during culture.

Variations and natural history of primary intraparenchymal lesions associated with neurofibromatosis type 2.

The study aimed to investigate the clinical implications and natural history of primary intraparenchymal lesions in patients with neurofibromatosis type 2. Radiological findings of 15 neurofibromatosis type 2 cases were retrospectively collected. Twenty-seven primary intraparenchymal lesions were observed in 7 out of 15 patients (47%). Cortical/subcortical T2 hyperintense lesions and enlarged Virchow-Robin spaces were the most common findings in five and four patients, respectively. During the follow-up period (median 84 months), one new primary intraparenchymal lesion was identified and increased lesions were observed in two cases on contrast-enhanced MRI. Surgical resection was performed in one case pathologically diagnosed with atypical meningioma. Twenty-five other lesions without contrast enhancement presented no apparent growth during follow-up. Although most primary intraparenchymal lesions are benign, a subset of cases would present newly developed or increased lesions on contrast-enhanced MRI. Careful monitoring is necessary for such cases, and pathological confirmation should be considered.

Deep Learning of Cancer Stem Cell Morphology Using Conditional Generative Adversarial Networks.

Deep-learning workflows of microscopic image analysis are sufficient for handling the contextual variations because they employ biological samples and have numerous tasks. The use of well-defined annotated images is important for the workflow. Cancer stem cells (CSCs) are identified by specific cell markers. These CSCs were extensively characterized by the stem cell (SC)-like gene expression and proliferation mechanisms for the development of tumors. In contrast, the morphological characterization remains elusive. This study aims to investigate the segmentation of CSCs in phase contrast imaging using conditional generative adversarial networks (CGAN). Artificial intelligence (AI) was trained using fluorescence images of the Nanog-Green fluorescence protein, the expression of which was maintained in CSCs, and the phase contrast images. The AI model segmented the CSC region in the phase contrast image of the CSC cultures and tumor model. By selecting images for training, several values for measuring segmentation quality increased. Moreover, nucleus fluorescence overlaid-phase contrast was effective for increasing the values. We show the possibility of mapping CSC morphology to the condition of undifferentiation using deep-learning CGAN workflows.

Prediction of Hypoxia in Brain Tumors Using a Multivariate Model Built from MR Imaging and 18F-Fluorodeoxyglucose Accumulation Data.

PURPOSE: The aim of this study was to generate a multivariate model using various MRI markers of blood flow and vascular permeability and accumulation of 18F-fluorodeoxyglucose (FDG) to predict the extent of hypoxia in an 18F-fluoromisonidazole (FMISO)-positive region. METHODS: Fifteen patients aged 27-74 years with brain tumors (glioma, n = 13; lymphoma, n = 1; germinoma, n = 1) were included. MRI scans were performed using a 3T scanner, and dynamic contrast-enhanced (DCE) perfusion and arterial spin labeling images were obtained. Ktrans and Vp maps were generated using the DCE images. FDG and FMISO positron emission tomography scans were also obtained. A model for predicting FMISO positivity was generated on a voxel-by-voxel basis by a multivariate logistic regression model using all the MRI parameters with and without FDG. Receiver-operating characteristic curve analysis was used to detect FMISO positivity with multivariate and univariate analysis of each parameter. Cross-validation was performed using the leave-one-out method. RESULTS: The area under the curve (AUC) was highest for the multivariate prediction model with FDG (0.892) followed by the multivariate model without FDG and univariate analysis with FDG and Ktrans (0.844 for all). In cross-validation, the multivariate model with FDG had the highest AUC (0.857 ± 0.08) followed by the multivariate model without FDG (0.834 ± 0.119). CONCLUSION: A multivariate prediction model created using blood flow, vascular permeability, and glycometabolism parameters can predict the extent of hypoxia in FMISO-positive areas in patients with brain tumors.

Diagnostic value of tumor blood flow and its histogram analysis obtained with pCASL to differentiate sinonasal malignant lymphoma from squamous cell carcinoma.

OBJECTIVES: To investigate the diagnostic value of tumor blood flow (TBF) obtained with pseudo-continuous arterial spin labeling (pCASL) for the differentiation of squamous cell carcinoma (SCC) and malignant lymphoma (ML) in the nasal or sinonasal cavity. METHODS: Thirty-three patients with SCC and 6 patients with ML in the nasal or sinonasal cavity were retrospectively analyzed. Quantitative TBF values were obtained using whole-tumor region of interest (ROI) from pCASL data. The histogram analysis of TBF values within the tumor ROI was also performed by calculating the coefficient of variation (CV), kurtosis, and skewness. The mean TBF value, histogram CV, kurtosis and skewness of the patients with SCC were compared with those of the ML patients. The diagnostic accuracy to differentiate SCC from ML was also calculated by receiver operating characteristic (ROC) curve analysis. In addition, multiple logistic regression models were also performed to determine their independent predictive value, and diagnostic accuracy with the combined use of these parameters. RESULTS: Between the SCC and ML groups, significant differences were observed in mean TBF, CV, and kurtosis, but not in skewness. In ROC curve analysis, the diagnostic accuracy values for the differentiation of SCC from ML in mean TBF, CV, and kurtosis were all 0.87, respectively. Multiple logistic regression models revealed TBF and CV were respectively independent predictive value. With the combination of these parameters, the diagnostic accuracy was elevated to 0.97. CONCLUSIONS: The TBF value and its histogram analysis obtained with pCASL can help differentiate SCC and ML.

TRIM67 protein negatively regulates Ras activity through degradation of 80K-H and induces neuritogenesis.

Tripartite motif (TRIM)-containing proteins, which are defined by the presence of a common domain structure composed of a RING finger, one or two B-box motifs and a coiled-coil motif, are involved in many biological processes including innate immunity, viral infection, carcinogenesis, and development. Here we show that TRIM67, which has a TRIM motif, an FN3 domain and a SPRY domain, is highly expressed in the cerebellum and that TRIM67 interacts with PRG-1 and 80K-H, which is involved in the Ras-mediated signaling pathway. Ectopic expression of TRIM67 results in degradation of endogenous 80K-H and attenuation of cell proliferation and enhances neuritogenesis in the neuroblastoma cell line N1E-115. Furthermore, morphological and biological changes caused by knockdown of 80K-H are similar to those observed by overexpression of TRIM67. These findings suggest that TRIM67 regulates Ras signaling via degradation of 80K-H, leading to neural differentiation including neuritogenesis.

Inhibition of NF-kappaB signaling via tyrosine phosphorylation of Ymer.

Cytoplasmic zinc finger protein A20 functionally dampens inflammatory signals and apoptosis via inhibition of NF-kappaB activation. We have reported that Ymer interacts with A20 and lysine (K)-63-linked polyubiquitin chain and that Ymer inhibits NF-kappaB signaling in collaboration with A20. It has also been reported that Ymer is phosphorylated by EGF stimulation. We found that Ymer was considerably phosphorylated on tyrosine residues also via Src family kinases such as Lck. A luciferase reporter assay showed that mutation of tyrosines on Ymer (YmerY217/279/304F) results in loss of the inhibitory activity for NF-kappaB signaling. Furthermore, a soft agar colony formation assay showed that the combination of SrcY527F and YmerY217/279/304F has no ability for anchorage-independent growth, suggesting that tyrosine phosphorylation of Ymer is important for inhibition of the NF-kappaB-mediated apoptotic pathway. These findings demonstrate that Ymer is likely to be a negative regulator for the NF-kappaB signaling pathway.