Microcalcification is an effective indicator of early breast cancer. To improve the diagnostic accuracy of microcalcifications, this study evaluates the performance of deep learning-based models on large datasets for its discrimination. A semi-automated segmentation method was used to characterize all microcalcifications. A discrimination classifier model was constructed to assess the accuracies of microcalcifications and breast masses, either in isolation or combination, for classifying breast lesions. Performances were compared to benchmark models. Our deep learning model achieved a discriminative accuracy of 87.3% if microcalcifications were characterized alone, compared to 85.8% with a support vector machine. The accuracies were 61.3% for both methods with masses alone and improved to 89.7% and 85.8% after the combined analysis with microcalcifications. Image segmentation with our deep learning model yielded 15, 26 and 41 features for the three scenarios, respectively. Overall, deep learning based on large datasets was superior to standard methods for the discrimination of microcalcifications. Accuracy was increased by adopting a combinatorial approach to detect microcalcifications and masses simultaneously. This may have clinical value for early detection and treatment of breast cancer.

The prognostic values of staging parameters require continual re-assessment amid changes in diagnostic and therapeutic methods. This study aimed to identify the prognostic factors and failure patterns of non-metastatic nasopharyngeal carcinoma (NPC) in the intensity-modulated radiotherapy (IMRT) era. We reviewed the data from 749 patients with newly diagnosed, biopsy-proven, non-metastatic NPC in our cancer center (South China, an NPC endemic area) between January 2003 and December 2007. All patients underwent magnetic resonance imaging (MRI) before receiving IMRT. The actuarial survival rates were estimated using the Kaplan–Meier method, and survival curves were compared using the log-rank test. Multivariate analyses with the Cox proportional hazards model were used to test for the independent prognostic factors by backward eliminating insignificant explanatory variables. The 5-year occurrence rates of local failure, regional failure, locoregional failure, and distant failure were 5.4, 3.0, 7.4, and 17.4%, respectively. The 5-year survival rates were as follows: local relapse-free survival, 94.6%; nodal relapse-free survival, 97.0%; distant metastasis-free survival, 82.6%; disease-free survival, 75.1%; and overall survival, 82.0%. Multivariate Cox regression analysis revealed that orbit involvement was the only significant prognostic factor for local failure (P = 0.011). Parapharyngeal tumor extension, retropharyngeal lymph node involvement, and the laterality, longest diameter, and Ho’s location of the cervical lymph nodes were significant prognostic factors for both distant failure and disease failure (all P < 0.05). Intracranial extension had significant prognostic value for distant failure (P = 0.040). The key failure pattern for NPC was distant metastasis in the IMRT era. With changes in diagnostic and therapeutic technologies as well as treatment modalities, the significant prognostic parameters for local control have also been altered substantially.

Mammography is successfully used as an effective screening tool for cancer diagnosis. A calcification cluster on mammography is a primary sign of cancer. Early researches have proved the diagnostic value of the calcification, yet their performance is highly dependent on handcrafted image descriptors. Characterizing the calcification mammography in an automatic and robust way remains a challenge. In this paper, the calcification was characterized by descriptors obtained from deep learning and handcrafted descriptors. We compared the performances of different image feature sets on digital mammograms. The feature sets included the deep features alone, the handcrafted features, their combination, and the filtered deep features. Experimental results have demonstrated that the deep features outperform handcrafted features, but the handcrafted features can provide complementary information for deep features. We achieved a classification precision of 89.32% and sensitivity of 86.89% using the filtered deep features, which is the best performance among all the feature sets.

Recent studies have reported miR-145 dysregulated in colorectal cancer (CRC). In this study, miR-145 profiles were compared between CRC and corresponding non-tumour tissues. The expression levels of miR-145 were analysed in CRC cell lines and tumour tissues by real-time PCR. A luciferase reporter assay confirmed direct targets. The functional effects of miR-145 were examined in transfected CRC cells in vitro and in vivo using established assays. Downregulation of miR-145 was detected in most primary CRC tumours, and was significantly correlated with a more aggressive phenotype of CRC in patients. In CRC cell lines, ectopic overexpression of miR-145 inhibited cell proliferation, motility and invasion in vitro. Stable overexpression of miR-145 suppressed tumour growth and pulmonary metastasis in vivo. Further studies indicated that miR-145 may directly interact with the 3′-untranslated region (3′-UTR) of Fascin-1 messenger RNA (mRNA), downregulating its mRNA and protein expression levels. In clinical specimens, Fascin-1 expression was negatively correlated with miR-145 expression. MiR-145 has a critical role in the inhibition of invasive and metastatic capacities of CRC, probably through directly targeting Fascin-1. This miRNA may be involved in the development and progression of CRC.

Herein we developed a new “smart” Gd-based MR contrast agent (i.e., 1) which is susceptive to furin, a protease overexpressed in tumor. Under the action of furin, 1 condenses to form dimers (1-Ds) and the latter self-assemble into gadolinium nanparticles (Gd-NPs). Relaxivity of 1-D is more than 2 folds of those of 1 and magnevist at 1.5 T, and 1.4 folds of that of 1 at 3 T. Intracellular condensation of 1 in furin-overexpressed MDA-MB-468 cells was proven with direct two-photon laser microscopy (TPLM) fluorescence imaging of the cells incubated with the europium analog of 1 (i.e., 2). Intracellular Gd-NPs of 1 were uncovered and characterized for the first time. MRI of MDA-MB-468 tumors showed that 1 has enhanced MR contrast within the tumors than that of its scrambled control 1-Scr.

Mn-based nanoparticles (NPs) have emerged as new class of probes for magnetic resonance imaging due to the impressive contrast ability. However, the reported Mn-based NPs possess low relaxivity and there are no immunotoxicity data regarding Mn-based NPs as contrast agents. Here, we demonstrate the ultrahigh relaxivity of water protons of 8.26 mM⁻¹s⁻¹ from the Mn₃O₄ NPs synthesized by a simple and green technique, which is twice higher than that of commercial gadolinium (Gd)-based contrast agents (4.11 mM⁻¹s⁻¹) and the highest value reported to date for Mn-based NPs. We for the first time demonstrate these Mn₃O₄ NPs biocompatibilities both in vitro and in vivo are satisfactory based on systematical studies of the intrinsic toxicity including cell viability of human nasopharyngeal carcinoma cells, normal nasopharyngeal epithelium, apoptosis in cells and in vivo immunotoxicity. These findings pave the way for the practical clinical diagnosis of Mn based NPs as safe probes for in vivo imaging.