Abstract
Backed by more powerful computational resources and optimized training routines, deep learning models have attained unprecedented performance in extracting information from chest X-ray data. Preceding other tasks, an automated abnormality detection stage can be useful to prioritize certain exams and enable a more efficient clinical workflow. However, the presence of image artifacts such as lettering often generates a harmful bias in the classifier, leading to an increase of false positive results. Consequently, healthcare would benefit from a system that selects the thoracic region of interest prior to deciding whether an image is possibly pathologic. The current work tackles this binary classification exercise using an attention-driven and spatially unsupervised Spatial Transformer Network (STN). The results indicate that the STN achieves similar results to using YOLO-cropped images, with fewer computational expenses and without the need for localization labels. More specifically, the system is able to distinguish between normal and abnormal CheXpert images with a mean AUC of 84.22%.

Abstract
Several systemic diseases affect the retinal blood vessels, and thus, their assessment allows an accurate clinical diagnosis. This assessment entails the estimation of the arteriolar-to-venular ratio (AVR), a predictive biomarker of cerebral atrophy and cardiovascular events in adults. In this context, different automatic and semiautomatic image-based approaches for artery/vein (A/V) classification and AVR estimation have been proposed in the literature, to the point of having become a hot research topic in the last decades. Most of these approaches use a wide variety of image properties, often redundant and/or irrelevant, requiring a training process that limits their generalization ability when applied to other datasets. This paper presents a new automatic method for A/V classification that just uses the local contrast between blood vessels and their surrounding background, computes a graph that represents the vascular structure, and applies a multilevel thresholding to obtain a preliminary classification. Next, a novel graph propagation approach was developed to obtain the final A/V classification and to compute the AVR. Our approach has been tested on two public datasets (INSPIRE and DRIVE), obtaining high classification accuracy rates, especially in the main vessels, and AVR ratios very similar to those provided by human experts. Therefore, our fully automatic method provides the reliable results without any training step, which makes it suitable for use with different retinal image datasets and as part of any clinical routine.

Abstract
Purpose This study aims to investigate retinal and choroidal vascular reactivity to carbogen in central serous chorioretinopathy (CSC) patients. Methods An experimental pilot study including 68 eyes from 20 CSC patients and 14 age and sex-matched controls was performed. The participants inhaled carbogen (5% CO2 + 95% O-2) for 2 min through a high-concentration disposable mask. A 30 degrees disc-centered fundus imaging using infra-red (IR) and macular spectral domain optical coherence tomography (SD-OCT) using the enhanced depth imaging (EDI) technique was performed, both at baseline and after a 2-min gas exposure. A parametric model fitting-based approach for automatic retinal blood vessel caliber estimation was used to assess the mean variation in both arterial and venous vasculature. Choroidal thickness was measured in two different ways: the subfoveal choroidal thickness (SFCT) was calculated using a manual caliper and the mean central choroidal thickness (MCCT) was assessed using an automatic software. Results No significant differences were detected in baseline hemodynamic parameters between both groups. A significant positive correlation was found between the participants' age and arterial diameter variation (p < 0.001, r= 0.447), meaning that younger participants presented a more vasoconstrictive response (negative variation) than older ones. No significant differences were detected in the vasoreactive response between CSC and controls for both arterial and venous vessels (p = 0.63 and p = 0.85, respectively). Although the vascular reactivity was not related to the activity of CSC, it was related to the time of disease, for both the arterial (p = 0.02, r = 0.381) and venous (p = 0.001, r= 0.530) beds. SFCT and MCCT were highly correlated (r= 0.830, p < 0.001). Both SFCT and MCCT significantly increased in CSC patients (p < 0.001 and p < 0.001) but not in controls (p = 0.059 and 0.247). A significant negative correlation between CSC patients' age and MCCT variation (r = - 0.340, p = 0.049) was detected. In CSC patients, the choroidal thickness variation was not related to the activity state, time of disease, or previous photodynamic treatment. Conclusion Vasoreactivity to carbogen was similar in the retinal vessels but significantly higher in the choroidal vessels of CSC patients when compared to controls, strengthening the hypothesis of a choroidal regulation dysfunction in this pathology.

Abstract
Advances in data-driven deep learning for chest X-ray image analysis underscore the need for explainability, privacy, large datasets and significant computational resources. We frame privacy and explainability as a lossy single-image compression problem to reduce both computational and data requirements without training. For Cardiomegaly detection in chest X-ray images, we propose HeartSpot and four spatial bias priors. HeartSpot priors define how to sample pixels based on domain knowledge from medical literature and from machines. HeartSpot privatizes chest X-ray images by discarding up to 97% of pixels, such as those that reveal the shape of the thoracic cage, bones, small lesions and other sensitive features. HeartSpot priors are ante-hoc explainable and give a human-interpretable image of the preserved spatial features that clearly outlines the heart. HeartSpot offers strong compression, with up to 32x fewer pixels and llx smaller filesize. Cardiomegaly detectors using HeartSpot are up to 9x faster to train or at least as accurate (up to +.01 AUC ROC) when compared to a baseline DenseNet121. HeartSpot is post-hoc explainable by re-using existing attribution methods without requiring access to the original non-privatized image. In summary, HeartSpot improves speed and accuracy, reduces image size, improves privacy and ensures explainability.

Abstract
Microscopy images have been increasingly analyzed quantitatively in biomedical research. Segmenting individual cell nucleus is an important step as many research studies involve counting cell nuclei and analysing their shape. We propose a novel weakly supervised instance segmentation method trained with image segmentation masks only. Our system comprises two models: an implicit shape Multi-Layer Perceptron (MLP) that learns the shape of the nuclei in canonical coordinates; and 2) an encoder that predicts the parameters of the affine transformation to deform the canonical shape into the correct location, scale, and orientation in the image. To further improve the performance of the model, we propose a loss that uses the total number of nuclei in an image as supervision. Our system is explainable, as the implicit shape MLP learns that the canonical shape of the cell nuclei is a circle, and interpretable as the output of the encoder are parameters of affine transformations. We obtain image segmentation performance close to DeepLabV3 and, additionally, obtain an F1-scoreIoU=0.5 of 68.47% at the instance segmentation task, even though the system was trained with image segmentations. © 2022 P. Costa, A. Gaudio, A. Campilho & J.S. Cardoso.

Abstract
Is there an initialization for deep networks that requires no learning? ExplainFix adopts two design principles: the fixed filters principle that all spatial filter weights of convolutional neural networks can be fixed at initialization and never learned, and the nimbleness principle that only few network parameters suffice. We contribute (a) visual model-based explanations, (b) speed and accuracy gains, and (c) novel tools for deep convolutional neural networks. ExplainFix gives key insights that spatially fixed networks should have a steered initialization, that spatial convolution layers tend to prioritize low frequencies, and that most network parameters are not necessary in spatially fixed models. ExplainFix models have up to x100 fewer spatial filter kernels than fully learned models and matching or improved accuracy. Our extensive empirical analysis confirms that ExplainFix guarantees nimbler models (train up to 17% faster with channel pruning), matching or improved predictive performance (spanning 13 distinct baseline models, four architectures and two medical image datasets), improved robustness to larger learning rate, and robustness to varying model size. We are first to demonstrate that all spatial filters in state-of-the-art convolutional deep networks can be fixed at initialization, not learned.This article is categorized under:Technologies > Machine LearningFundamental Concepts of Data and Knowledge > Explainable AIFundamental Concepts of Data and Knowledge > Key Design Issues in Data Mining