Abstract
The increasing popularity of attention mechanisms in deep learning algorithms for computer vision and natural language processing made these models attractive to other research domains. In healthcare, there is a strong need for tools that may improve the routines of the clinicians and the patients. Naturally, the use of attention-based algorithms for medical applications occurred smoothly. However, being healthcare a domain that depends on high-stake decisions, the scientific community must ponder if these high-performing algorithms fit the needs of medical applications. With this motto, this paper extensively reviews the use of attention mechanisms in machine learning methods (including Transformers) for several medical applications based on the types of tasks that may integrate several works pipelines of the medical domain. This work distinguishes itself from its predecessors by proposing a critical analysis of the claims and potentialities of attention mechanisms presented in the literature through an experimental case study on medical image classification with three different use cases. These experiments focus on the integrating process of attention mechanisms into established deep learning architectures, the analysis of their predictive power, and a visual assessment of their saliency maps generated by post-hoc explanation methods. This paper concludes with a critical analysis of the claims and potentialities presented in the literature about attention mechanisms and proposes future research lines in medical applications that may benefit from these frameworks.

Abstract
Every year, the VISion Understanding and Machine intelligence (VISUM) summer school runs a competition where participants can learn and share knowledge about Computer Vision and Machine Learning in a vibrant environment. 2021 VISUM's focused on applying those methodologies in fashion. Recently, there has been an increase of interest within the scientific community in applying computer vision methodologies to the fashion domain. That is highly motivated by fashion being one of the world's largest industries presenting a rapid development in e-commerce mainly since the COVID-19 pandemic. Computer Vision for Fashion enables a wide range of innovations, from personalized recommendations to outfit matching. The competition enabled students to apply the knowledge acquired in the summer school to a real-world problem. The ambition was to foster research and development in fashion outfit complementary product retrieval by leveraging vast visual and textual data with domain knowledge. For this, a new fashion outfit dataset (acquired and curated by FARFETCH) for research and benchmark purposes is introduced. Additionally, a competitive baseline with an original negative sampling process for triplet mining was implemented and served as a starting point for participants. The top 3 performing methods are described in this paper since they constitute the reference state-of-the-art for this particular problem. To our knowledge, this is the first challenge in fashion outfit complementary product retrieval. Moreover, this joint project between academia and industry brings several relevant contributions to disseminating science and technology, promoting economic and social development, and helping to connect early-career researchers to real-world industry challenges.

Abstract
This paper presents the main contributions of the VCMI Team to the ImageCLEFmedical Caption 2023 task. We addressed both the concept detection and caption prediction tasks. Regarding concept detection, our team employed different approaches to assign concepts to medical images: multi-label classification, adversarial training, autoregressive modelling, image retrieval, and concept retrieval. We also developed three model ensembles merging the results of some of the proposed methods. Our best submission obtained an F1-score of 0.4998, ranking 3rd among nine teams. Regarding the caption prediction task, our team explored two main approaches based on image retrieval and language generation. The language generation approaches, based on a vision model as the encoder and a language model as the decoder, yielded the best results, allowing us to rank 5th among thirteen teams, with a BERTScore of 0.6147. © 2023 Copyright for this paper by its authors.

Abstract
This paper presents the main contributions of the VCMI Team to the ImageCLEFmedical GANs 2023 task. This task aims to evaluate whether synthetic medical images generated using Generative Adversarial Networks (GANs) contain identifiable characteristics of the training data. We propose various approaches to classify a set of real images as having been used or not used in the training of the model that generated a set of synthetic images. We use similarity-based approaches to classify the real images based on their similarity to the generated ones. We develop autoencoders to classify the images through outlier detection techniques. Finally, we develop patch-based methods that operate on patches extracted from real and generated images to measure their similarity. On the development dataset, we attained an F1-score of 0.846 and an accuracy of 0.850 using an autoencoder-based method. On the test dataset, a similarity-based approach achieved the best results, with an F1-score of 0.801 and an accuracy of 0.810. The empirical results support the hypothesis that medical data generated using deep generative models trained without privacy constraints threatens the privacy of patients in the training data. © 2023 Copyright for this paper by its authors.

Abstract
Post-hoc explanation methods have often been criticised for abstracting away the decision-making process of deep neural networks. In this work, we would like to provide natural language descriptions for what different layers of a vision backbone have learned. Our DeViL method generates textual descriptions of visual features at different layers of the network as well as highlights the attribution locations of learned concepts. We train a transformer network to translate individual image features of any vision layer into a prompt that a separate off-the-shelf language model decodes into natural language. By employing dropout both per-layer and per-spatial-location, our model can generalize training on image-text pairs to generate localized explanations. As it uses a pre-trained language model, our approach is fast to train and can be applied to any vision backbone. Moreover, DeViL can create open-vocabulary attribution maps corresponding to words or phrases even outside the training scope of the vision model. We demonstrate that DeViL generates textual descriptions relevant to the image content on CC3M, surpassing previous lightweight captioning models and attribution maps, uncovering the learned concepts of the vision backbone. Further, we analyze fine-grained descriptions of layers as well as specific spatial locations and show that DeViL outperforms the current state-of-the-art on the neuron-wise descriptions of the MILANNOTATIONS dataset.

Abstract
In fields that rely on high-stakes decisions, such as medicine, interpretability plays a key role in promoting trust and facilitating the adoption of deep learning models by the clinical communities. In the medical image analysis domain, gradient-based class activation maps are the most widely used explanation methods and the field lacks a more in depth investigation into inherently interpretable models that focus on integrating knowledge that ensures the model is learning the correct rules. A new approach, B-cos networks, for increasing the interpretability of deep neural networks by inducing weight-input alignment during training showed promising results on natural image classification. In this work, we study the suitability of these B-cos networks to the medical domain by testing them on different use cases (skin lesions, diabetic retinopathy, cervical cytology, and chest X-rays) and conducting a thorough evaluation of several explanation quality assessment metrics. We find that, just like in natural image classification, B-cos explanations yield more localised maps, but it is not clear that they are better than other methods' explanations when considering more explanation properties. © 2024 IEEE.