Motivation: Tumor classification using Imaging Mass Spectrometry (IMS) data has a high potential for future applications in pathology. Due to the complexity and size of the data, automated feature extraction and classification steps are required to fully process the data. Deep learning offers an approach to learn feature extraction and classification combined in a single model. Commonly these steps are handled separately in IMS data analysis, hence deep learning offers an alternative strategy worthwhile to explore. Results: Methodologically, we propose an adapted architecture based on deep convolutional networks to handle the characteristics of mass spectrometry data, as well as a strategy to interpret the learned model in the spectral domain based on a sensitivity analysis. The proposed methods are evaluated on two challenging tumor classification tasks and compared to a baseline approach. Competitiveness of the proposed methods are shown on both tasks by studying the performance via cross-validation. Moreover, the learned models are analyzed by the proposed sensitivity analysis revealing biologically plausible effects as well as confounding factors of the considered task. Thus, this study may serve as a starting point for further development of deep learning approaches in IMS classification tasks. Source Code: https://gitlab.informatik.uni-bremen.de/digipath/Deep Learning for Tumor Classification in IMS Data: https://seafile.zfn.uni-bremen.de/d/85c915784e/

Despite their impressive performance, deep neural networks exhibit striking failures on out-of-distribution inputs. One core idea of adversarial example research is to reveal neural network errors under such distribution shifts. We decompose these errors into two complementary sources: sensitivity and invariance. We show deep networks are not only too sensitive to task-irrelevant changes of their input, as is well-known from -adversarial examples, but are also too invariant to a wide range of task-relevant changes, thus making vast regions in input space vulnerable to adversarial attacks. We show such excessive invariance occurs across various tasks and architecture types. On MNIST and ImageNet one can manipulate the class-specific content of almost any image without changing the hidden activations. We identify an insufficiency of the standard cross-entropy loss as a reason for these failures. Further, we extend this objective based on an informationtheoretic analysis so it encourages the model to consider all task-dependent features in its decision. This provides the first approach tailored explicitly to overcome excessive invariance and resulting vulnerabilities.

Subtyping of the most common non‐small cell lung cancer (NSCLC) tumor types adenocarcinoma (ADC) and squamous cell carcinoma (SqCC) is still a challenge in the clinical routine and a correct diagnosis is crucial for an adequate therapy selection. Matrix‐assisted laser desorption/ionization (MALDI) mass spectrometry imaging (MSI) has shown potential for NSCLC subtyping but is subject to strong technical variability and has only been applied to tissue samples assembled in tissue microarrays (TMAs). To our knowledge, a successful transfer of a classifier from TMAs to whole sections, which are generated in the standard clinical routine, has not been presented in the literature as of yet. We introduce a classification algorithm using extensive preprocessing and a classifier (either a neural network or a linear discriminant analysis (LDA)) to robustly classify whole sections of ADC and SqCC lung tissue. The classifiers were trained on TMAs and validated and tested on whole sections. Vital for a successful application on whole sections is the extensive preprocessing and the use of whole sections for hyperparameter selection. The classification system with the neural network/LDA results in 99.0%/98.3% test accuracy on spectra level and 100.0%/100.0% test accuracy on whole section level, respectively, and, therefore, provides a powerful tool to support the pathologist's decision making process. The presented method is a step further towards a clinical application of MALDI MSI and artificial intelligence for subtyping of NSCLC tissue sections.

Invertible neural networks (INNs) have been used to design generative models, implement memory-saving gradient computation, and solve inverse problems. In this work, we show that commonly-used INN architectures suffer from exploding inverses and are thus prone to becoming numerically non-invertible. Across a wide range of INN use-cases, we reveal failures including the non-applicability of the change-of-variables formula on in-and out-of-distribution (OOD) data, incorrect gradients for memory-saving backprop, and the inability to sample from normalizing flow models. We further derive bi-Lipschitz properties of atomic building blocks of common architectures. These insights into the stability of INNs then provide ways forward to remedy these failures. For tasks where local invertibility is sufficient, like memory-saving backprop, we propose a flexible and efficient regularizer. For problems where global invertibility is necessary, such as applying normalizing flows on OOD data, we show the importance of designing stable INN building blocks.Preprint. Under review.

Image reconstruction from computed tomography (CT) measurement is a challenging statistical inverse problem since a high-dimensional conditional distribution needs to be estimated. Based on training data obtained from high-quality reconstructions, we aim to learn a conditional density of images from noisy low-dose CT measurements. To tackle this problem, we propose a hybrid conditional normalizing flow, which integrates the physical model by using the filtered back-projection as conditioner. We evaluate our approach on a low-dose CT benchmark and demonstrate superior performance in terms of structural similarity of our flow-based method compared to other deep learning based approaches.