BackgroundThere is growing interest in utilizing artificial intelligence, and particularly deep learning, for computer vision in histopathology. While accumulating studies highlight expert-level performance of convolutional neural networks (CNNs) on focused classification tasks, most studies rely on probability distribution scores with empirically defined cutoff values based on post-hoc analysis. More generalizable tools that allow humans to visualize histology-based deep learning inferences and decision making are scarce.ResultsHere, we leverage t-distributed Stochastic Neighbor Embedding (t-SNE) to reduce dimensionality and depict how CNNs organize histomorphologic information. Unique to our workflow, we develop a quantitative and transparent approach to visualizing classification decisions prior to softmax compression. By discretizing the relationships between classes on the t-SNE plot, we show we can super-impose randomly sampled regions of test images and use their distribution to render statistically-driven classifications. Therefore, in addition to providing intuitive outputs for human review, this visual approach can carry out automated and objective multi-class classifications similar to more traditional and less-transparent categorical probability distribution scores. Importantly, this novel classification approach is driven by a priori statistically defined cutoffs. It therefore serves as a generalizable classification and anomaly detection tool less reliant on post-hoc tuning.ConclusionRoutine incorporation of this convenient approach for quantitative visualization and error reduction in histopathology aims to accelerate early adoption of CNNs into generalized real-world applications where unanticipated and previously untrained classes are often encountered.Electronic supplementary materialThe online version of this article (10.1186/s12859-018-2184-4) contains supplementary material, which is available to authorized users.

No abstract

INTRODUCTION: Intratumoral (epi)genetic heterogeneity is recognized to be an important driver in glioma therapy resistance. A single biopsy is unlikely to represent the true spatial and temporal heterogeneity. Therefore, multiple sampling schemes guided by imaging are needed. METHODS:We obtained 74 stereotactic image-guided biopsies preceding craniotomy in eight patients with glioma. Imaging included standard MRI, diffusion and perfusion weighted MRI, MR spectroscopy and PET FET and CHO. We performed multi-region genome-wide methylation profiling of all samples with DNA copy number profiles inferred from methylation array data. To assess variability in methylation profiles from Ceccarelli et al, Cell, 2016, we conducted supervised classification of each sample biopsy. Phylo(epi)genetic trees were constructed to investigate tumor evolutionary paths. Multimodality imaging data was used to predict (epi) genetic characteristics. Data was validated in two independent populations of respectively 32 multi-region samples in 5 patients and 80 single and multi-region initial and patient-matched recurrence samples in 19 patients. RESULTS: Six out of eight gliomas demonstrated spatial heterogeneity of epigenetic classification. In three IDH mutant gliomas of the G-CIMP high subtype (LGm2) regions were classified as G-CIMP low (LGm1) or Codel subtype (LGm3). In three IDH wild type gliomas of the Mesenchymal subtype (LGm5) regions were classified as Classic subtype (LGm4). This was validated in the second dataset in which one of six gliomas showed spatial and three of twenty gliomas showed temporal heterogeneity. Phyloepigenetic and phylogenetic trees showed high concordance. IDH status of the samples could be predicted using multimodality imaging. DISCUSSION: These findings provide valuable information concerning intratumoral heterogeneity in gliomas. Moreover, imaging was able to predict molecular characteristics, which could lead to multi-region sampling schemes that direct future therapy development.