Mutations in isocitrate dehydrogenase genes IDH1 and IDH2 are frequently found in diffuse and anaplastic astrocytic and oligodendroglial tumours as well as in secondary glioblastomas. As IDH is a very important prognostic, diagnostic and therapeutic biomarker for glioma, it is of paramount importance to determine its mutational status. The haematoxylin and eosin (H&E) staining is a valuable tool in precision oncology as it guides histopathology-based diagnosis and proceeding patient’s treatment. However, H&E staining alone does not determine the IDH mutational status of a tumour. Deep learning methods applied to MRI data have been demonstrated to be a useful tool in IDH status prediction, however the effectiveness of deep learning on H&E slides in the clinical setting has not been investigated so far. Furthermore, the performance of deep learning methods in medical imaging has been practically limited by small sample sizes currently available. Here we propose a data augmentation method based on the Generative Adversarial Networks (GAN) deep learning methodology, to improve the prediction performance of IDH mutational status using H&E slides. The H&E slides were acquired from 266 grade II-IV glioma patients from a mixture of public and private databases, including 130 IDH-wildtype and 136 IDH-mutant patients. A baseline deep learning model without data augmentation achieved an accuracy of 0.794 (AUC = 0.920). With GAN-based data augmentation, the accuracy of the IDH mutational status prediction was improved to 0.853 (AUC = 0.927) when the 3,000 GAN generated training samples were added to the original training set (24,000 samples). By integrating also patients’ age into the model, the accuracy improved further to 0.882 (AUC = 0.931). Our findings show that deep learning methodology, enhanced by GAN data augmentation, can support physicians in gliomas’ IDH status prediction.

Background: Modeling the complex development and growth of tumor angiogenesis using mathematics and biological data is a burgeoning area of cancer research. Architectural complexity is the main feature of every anatomical system, including organs, tissues, cells and sub-cellular entities. The vascular system is a complex network whose geometrical characteristics cannot be properly defined using the principles of Euclidean geometry, which is only capable of interpreting regular and smooth objects that are almost impossible to find in Nature. However, fractal geometry is a more powerful means of quantifying the spatial complexity of real objects. Methods: This paper introduces the surface fractal dimension (D _s ) as a numerical index of the two-dimensional (2-D) geometrical complexity of tumor vascular networks, and their behavior during computer-simulated changes in vessel density and distribution. Results: We show that D _s significantly depends on the number of vessels and their pattern of distribution. This demonstrates that the quantitative evaluation of the 2-D geometrical complexity of tumor vascular systems can be useful not only to measure its complex architecture, but also to model its development and growth. Conclusions: Studying the fractal properties of neovascularity induces reflections upon the real significance of the complex form of branched anatomical structures, in an attempt to define more appropriate methods of describing them quantitatively. This knowledge can be used to predict the aggressiveness of malignant tumors and design compounds that can halt the process of angiogenesis and influence tumor growth.

It is well known that angiogenesis is a complex process that accompanies neoplastic growth, but pituitary tumours are less vascularized than normal pituitary glands. Several analytical methods aimed at quantifying the vascular system in two‐dimensional histological sections have been proposed, with very discordant results. In this study we investigated the non‐Euclidean geometrical complexity of the two‐dimensional microvasculature of normal pituitary glands and pituitary adenomas by quantifying the surface fractal dimension that measures its space‐filling property. We found a statistical significant difference between the mean vascular surface fractal dimension estimated in normal versus adenomatous tissues (P = 0.01), normal versus secreting adenomatous tissues (P = 0.0003), and normal versus non‐secreting adenomatous tissues (P = 0.047), whereas the difference between the secreting and non‐secreting adenomatous tissues was not statistically significant. This study provides the first demonstration that fractal dimension is an objective and valid quantitator of the two‐dimensional geometrical complexity of the pituitary gland microvascular network in physiological and pathological states. Further studies are needed to compare the vascular surface fractal dimension estimates in different subtypes of pituitary tumours and correlate them with clinical parameters in order to evaluate whether the distribution pattern of vascular growth is related to a particular state of the pituitary gland.

Cancer remains one of the most complex diseases affecting humans and, despite the impressive advances that have been made in molecular and cell biology, how cancer cells progress through carcinogenesis and acquire their metastatic ability is still widely debated. There is no doubt that human carcinogenesis is a dynamic process that depends on a large number of variables and is regulated at multiple spatial and temporal scales. Viewing cancer as a system that is dynamically complex in time and space will, however, probably reveal more about its underlying behavioural characteristics. It is encouraging that mathematicians, biologists and clinicians continue to contribute together towards a common quantitative understanding of cancer complexity. This way of thinking may further help to clarify concepts, interpret new and old experimental data, indicate alternative experiments and categorize the acquired knowledge on the basis of the similarities and/or shared behaviours of very different tumours.