Miguel A. González Ballester

Adrian Galdran, Johan Verjans, Gustavo Carneiro et al.

Delivering meaningful uncertainty estimates is essential for a successful deployment of machine learning models in the clinical practice. A central aspect of uncertainty quantification is the ability of a model to return predictions that are well-aligned with the actual probability of the model being correct, also known as model calibration. Although many methods have been proposed to improve calibration, no technique can match the simple, but expensive approach of training an ensemble of deep neural networks. In this paper we introduce a form of simplified ensembling that bypasses the costly training and inference of deep ensembles, yet it keeps its calibration capabilities. The idea is to replace the common linear classifier at the end of a network by a set of heads that are supervised with different loss functions to enforce diversity on their predictions. Specifically, each head is trained to minimize a weighted Cross-Entropy loss, but the weights are different among the different branches. We show that the resulting averaged predictions can achieve excellent calibration without sacrificing accuracy in two challenging datasets for histopathological and endoscopic image classification. Our experiments indicate that Multi-Head Multi-Loss classifiers are inherently well-calibrated, outperforming other recent calibration techniques and even challenging Deep Ensembles' performance. Code to reproduce our experiments can be found at \url{https://github.com/agaldran/mhml_calibration} .

Adrian Galdran, Katherine J. Hewitt, Narmin L. Ghaffari et al.

Tissue typology annotation in Whole Slide histological images is a complex and tedious, yet necessary task for the development of computational pathology models. We propose to address this problem by applying Open Set Recognition techniques to the task of jointly classifying tissue that belongs to a set of annotated classes, e.g. clinically relevant tissue categories, while rejecting in test time Open Set samples, i.e. images that belong to categories not present in the training set. To this end, we introduce a new approach for Open Set histopathological image recognition based on training a model to accurately identify image categories and simultaneously predict which data augmentation transform has been applied. In test time, we measure model confidence in predicting this transform, which we expect to be lower for images in the Open Set. We carry out comprehensive experiments in the context of colorectal cancer assessment from histological images, which provide evidence on the strengths of our approach to automatically identify samples from unknown categories. Code is released at https://github.com/agaldran/t3po .

Valentin Comte, Mireia Alenya, Andrea Urru et al.

Accurate segmentation of fetal brain magnetic resonance images is crucial for analyzing fetal brain development and detecting potential neurodevelopmental abnormalities. Traditional deep learning-based automatic segmentation, although effective, requires extensive training data with ground-truth labels, typically produced by clinicians through a time-consuming annotation process. To overcome this challenge, we propose a novel unsupervised segmentation method based on multi-atlas segmentation, that accurately segments multiple tissues without relying on labeled data for training. Our method employs a cascaded deep learning network for 3D image registration, which computes small, incremental deformations to the moving image to align it precisely with the fixed image. This cascaded network can then be used to register multiple annotated images with the image to be segmented, and combine the propagated labels to form a refined segmentation. Our experiments demonstrate that the proposed cascaded architecture outperforms the state-of-the-art registration methods that were tested. Furthermore, the derived segmentation method achieves similar performance and inference time to nnU-Net while only using a small subset of annotated data for the multi-atlas segmentation task and none for training the network. Our pipeline for registration and multi-atlas segmentation is publicly available at https://github.com/ValBcn/CasReg.

Adrian Galdran, Miguel A. González Ballester

As current computing capabilities increase, modern machine learning and computer vision system tend to increase in complexity, mostly by means of larger models and advanced optimization strategies. Although often neglected, in many problems there is also much to be gained by considering potential improvements in understanding and better leveraging already-available training data, including annotations. This so-called data-centric approach can lead to substantial performance increases, sometimes beyond what can be achieved by larger models. In this paper we adopt such an approach for the task of justifiable glaucoma screening from retinal images. In particular, we focus on how to combine information from multiple annotators of different skills into a tailored label smoothing scheme that allows us to better employ a large collection of fundus images, instead of discarding samples suffering from inter-rater variability. Internal validation results indicate that our bespoke label smoothing approach surpasses the performance of a standard resnet50 model and also the same model trained with conventional label smoothing techniques, in particular for the multi-label scenario of predicting clinical reasons of glaucoma likelihood in a highly imbalanced screening context. Our code is made available at github.com/agaldran/justraigs .

Adrian Galdran, Gustavo Carneiro, Miguel A. González Ballester

Highly imbalanced datasets are ubiquitous in medical image classification problems. In such problems, it is often the case that rare classes associated to less prevalent diseases are severely under-represented in labeled databases, typically resulting in poor performance of machine learning algorithms due to overfitting in the learning process. In this paper, we propose a novel mechanism for sampling training data based on the popular MixUp regularization technique, which we refer to as Balanced-MixUp. In short, Balanced-MixUp simultaneously performs regular (i.e., instance-based) and balanced (i.e., class-based) sampling of the training data. The resulting two sets of samples are then mixed-up to create a more balanced training distribution from which a neural network can effectively learn without incurring in heavily under-fitting the minority classes. We experiment with a highly imbalanced dataset of retinal images (55K samples, 5 classes) and a long-tail dataset of gastro-intestinal video frames (10K images, 23 classes), using two CNNs of varying representation capabilities. Experimental results demonstrate that applying Balanced-MixUp outperforms other conventional sampling schemes and loss functions specifically designed to deal with imbalanced data. Code is released at https://github.com/agaldran/balanced_mixup .

Amelia Jiménez-Sánchez, Mickael Tardy, Miguel A. González Ballester et al.

For early breast cancer detection, regular screening with mammography imaging is recommended. Routinary examinations result in datasets with a predominant amount of negative samples. A potential solution to such class-imbalance is joining forces across multiple institutions. Developing a collaborative computer-aided diagnosis system is challenging in different ways. Patient privacy and regulations need to be carefully respected. Data across institutions may be acquired from different devices or imaging protocols, leading to heterogeneous non-IID data. Also, for learning-based methods, new optimization strategies working on distributed data are required. Recently, federated learning has emerged as an effective tool for collaborative learning. In this setting, local models perform computation on their private data to update the global model. The order and the frequency of local updates influence the final global model. Hence, the order in which samples are locally presented to the optimizers plays an important role. In this work, we define a memory-aware curriculum learning method for the federated setting. Our curriculum controls the order of the training samples paying special attention to those that are forgotten after the deployment of the global model. Our approach is combined with unsupervised domain adaptation to deal with domain shift while preserving data privacy. We evaluate our method with three clinical datasets from different vendors. Our results verify the effectiveness of federated adversarial learning for the multi-site breast cancer classification. Moreover, we show that our proposed memory-aware curriculum method is beneficial to further improve classification performance. Our code is publicly available at: https://github.com/ameliajimenez/curriculum-federated-learning.

Amelia Jiménez-Sánchez, Diana Mateus, Sonja Kirchhoff et al.

An adequate classification of proximal femur fractures from X-ray images is crucial for the treatment choice and the patients' clinical outcome. We rely on the commonly used AO system, which describes a hierarchical knowledge tree classifying the images into types and subtypes according to the fracture's location and complexity. In this paper, we propose a method for the automatic classification of proximal femur fractures into 3 and 7 AO classes based on a Convolutional Neural Network (CNN). As it is known, CNNs need large and representative datasets with reliable labels, which are hard to collect for the application at hand. In this paper, we design a curriculum learning (CL) approach that improves over the basic CNNs performance under such conditions. Our novel formulation reunites three curriculum strategies: individually weighting training samples, reordering the training set, and sampling subsets of data. The core of these strategies is a scoring function ranking the training samples. We define two novel scoring functions: one from domain-specific prior knowledge and an original self-paced uncertainty score. We perform experiments on a clinical dataset of proximal femur radiographs. The curriculum improves proximal femur fracture classification up to the performance of experienced trauma surgeons. The best curriculum method reorders the training set based on prior knowledge resulting into a classification improvement of 15%. Using the publicly available MNIST dataset, we further discuss and demonstrate the benefits of our unified CL formulation for three controlled and challenging digit recognition scenarios: with limited amounts of data, under class-imbalance, and in the presence of label noise. The code of our work is available at: https://github.com/ameliajimenez/curriculum-learning-prior-uncertainty.

Meritxell Riera-Marín, Javier García López, Júlia Rodríguez-Comas et al.

Objective: Accurate probability estimates are essential for the safe deployment of medical image segmentation models in clinical decision-making. However, modern deep segmentation networks are often poorly calibrated, a problem exacerbated when multiple expert annotations exhibit substantial disagreement. While inter-rater variability is typically treated as noise, it provides valuable information about intrinsic annotation ambiguity that must be reflected in model confidence. Methods: We improve the probabilistic calibration of medical image segmentation models by reformulating multi-rater supervision as an ordinal learning problem. Voxel-wise annotator agreement is treated as an ordered target, linking predictive confidence to the empirical variability in training data. This formulation allows the use of ordinal-aware scoring rules, such as the Ranked Probability Score ordinal loss, combined with a standard binary objective to preserve discriminative performance. Results: We evaluated the proposed approach across four public segmentation benchmarks spanning ophthalmology, histopathology, and thoracic imaging. Calibration was assessed using a multi-rater extension of expected calibration error. Results consistently show that ordinal-aware training yields substantially improved calibration with respect to inter-rater agreement without degrading segmentation accuracy. Conclusions: Treating multi-rater annotations as ordered information provides a principled and architecture-agnostic route to more reliable probabilistic segmentation models.

Francesco Aldo Venturelli, Emanuele Costa, Sikha O K et al.

Deep learning models are used in critical applications, in which mistakes can have serious consequences. Therefore, it is crucial to understand how and why models generate predictions. This understanding provides useful information to check whether the model is learning the right patterns, detect biases in the data, improve model design, and build systems that can be trusted. This work proposes a new method for interpreting Convolutional Neural Networks in image classification tasks. The approach works by selecting the most important feature maps that contribute to each prediction. To solve this combinatorial problem, we encode it into a quantum constrained optimization problem and propose to solve it using quantum annealing. We evaluate our method against the state-of-the-art explainable AI techniques, specifically GradCAM and GradCAM++, and observe an improved class disentanglement, i.e. the model's decision boundaries become more distinct and its reasoning more transparent. This demonstrates that our approach enhances the quality of explanations, making it easier to understand which features the model relies on for specific predictions. In addition, we study the computational behavior of the quantum annealing algorithm. Specifically, we analyze the minimum energy gap of the system during computation and the probability that the algorithm finds the correct solution. These analyses provide theoretical insight into why the method works effectively in practice.

Sikha O K, Meritxell Riera-Marín, Adrian Galdran et al.

Image segmentation is a critical step in computational biomedical image analysis, typically evaluated using metrics like the Dice coefficient during training and validation. However, in clinical settings without manual annotations, assessing segmentation quality becomes challenging, and models lacking reliability indicators face adoption barriers. To address this gap, we propose a novel framework for predicting segmentation quality without requiring ground truth annotations during test time. Our approach introduces two complementary frameworks: one leveraging predicted segmentation and uncertainty maps, and another integrating the original input image, uncertainty maps, and predicted segmentation maps. We present Bayesian adaptations of two benchmark segmentation models-SwinUNet and Feature Pyramid Network with ResNet50-using Monte Carlo Dropout, Ensemble, and Test Time Augmentation to quantify uncertainty. We evaluate four uncertainty estimates: confidence map, entropy, mutual information, and expected pairwise Kullback-Leibler divergence on 2D skin lesion and 3D liver segmentation datasets, analyzing their correlation with segmentation quality metrics. Our framework achieves an R2 score of 93.25 and Pearson correlation of 96.58 on the HAM10000 dataset, outperforming previous segmentation quality assessment methods. For 3D liver segmentation, Test Time Augmentation with entropy achieves an R2 score of 85.03 and a Pearson correlation of 65.02, demonstrating cross-modality robustness. Additionally, we propose an aggregation strategy that combines multiple uncertainty estimates into a single score per image, offering a more robust and comprehensive assessment of segmentation quality. Finally, we use Grad-CAM and UMAP-based embedding analysis to interpret the model's behavior and reliability, highlighting the impact of uncertainty integration.

Adrian Galdran, Gustavo Carneiro, Miguel A. González Ballester

This paper compares well-established Convolutional Neural Networks (CNNs) to recently introduced Vision Transformers for the task of Diabetic Foot Ulcer Classification, in the context of the DFUC 2021 Grand-Challenge, in which this work attained the first position. Comprehensive experiments demonstrate that modern CNNs are still capable of outperforming Transformers in a low-data regime, likely owing to their ability for better exploiting spatial correlations. In addition, we empirically demonstrate that the recent Sharpness-Aware Minimization (SAM) optimization algorithm considerably improves the generalization capability of both kinds of models. Our results demonstrate that for this task, the combination of CNNs and the SAM optimization process results in superior performance than any other of the considered approaches.

Adrian Galdran, Gustavo Carneiro, Miguel A. González Ballester

Polyps represent an early sign of the development of Colorectal Cancer. The standard procedure for their detection consists of colonoscopic examination of the gastrointestinal tract. However, the wide range of polyp shapes and visual appearances, as well as the reduced quality of this image modality, turn their automatic identification and segmentation with computational tools into a challenging computer vision task. In this work, we present a new strategy for the delineation of gastrointestinal polyps from endoscopic images based on a direct extension of common encoder-decoder networks for semantic segmentation. In our approach, two pretrained encoder-decoder networks are sequentially stacked: the second network takes as input the concatenation of the original frame and the initial prediction generated by the first network, which acts as an attention mechanism enabling the second network to focus on interesting areas within the image, thereby improving the quality of its predictions. Quantitative evaluation carried out on several polyp segmentation databases shows that double encoder-decoder networks clearly outperform their single encoder-decoder counterparts in all cases. In addition, our best double encoder-decoder combination attains excellent segmentation accuracy and reaches state-of-the-art performance results in all the considered datasets, with a remarkable boost of accuracy on images extracted from datasets not used for training.

Xavier Rafael-Palou, Anton Aubanell, Mario Ceresa et al.

Early detection and quantification of tumour growth would help clinicians to prescribe more accurate treatments and provide better surgical planning. However, the multifactorial and heterogeneous nature of lung tumour progression hampers identification of growth patterns. In this study, we present a novel method based on a deep hierarchical generative and probabilistic framework that, according to radiological guidelines, predicts tumour growth, quantifies its size and provides a semantic appearance of the future nodule. Unlike previous deterministic solutions, the generative characteristic of our approach also allows us to estimate the uncertainty in the predictions, especially important for complex and doubtful cases. Results of evaluating this method on an independent test set reported a tumour growth balanced accuracy of 74%, a tumour growth size MAE of 1.77 mm and a tumour segmentation Dice score of 78%. These surpassed the performances of equivalent deterministic and alternative generative solutions (i.e. probabilistic U-Net, Bayesian test dropout and Pix2Pix GAN) confirming the suitability of our approach.

Xavier Rafael-Palou, Anton Aubanell, Mario Ceresa et al.

We address the problem of supporting radiologists in the longitudinal management of lung cancer. Therefore, we proposed a deep learning pipeline, composed of four stages that completely automatized from the detection of nodules to the classification of cancer, through the detection of growth in the nodules. In addition, the pipeline integrated a novel approach for nodule growth detection, which relied on a recent hierarchical probabilistic U-Net adapted to report uncertainty estimates. Also, a second novel method was introduced for lung cancer nodule classification, integrating into a two stream 3D-CNN network the estimated nodule malignancy probabilities derived from a pretrained nodule malignancy network. The pipeline was evaluated in a longitudinal cohort and reported comparable performances to the state of art.

Xavier Rafael-Palou, Anton Aubanell, Ilaria Bonavita et al.

Nodule malignancy assessment is a complex, time-consuming and error-prone task. Current clinical practice requires measuring changes in size and density of the nodule at different time-points. State of the art solutions rely on 3D convolutional neural networks built on pulmonary nodules obtained from single CT scan per patient. In this work, we propose a two-stream 3D convolutional neural network that predicts malignancy by jointly analyzing two pulmonary nodule volumes from the same patient taken at different time-points. Best results achieve 77% of F1-score in test with an increment of 9% and 12% of F1-score with respect to the same network trained with images from a single time-point.

Amelia Jiménez-Sánchez, Diana Mateus, Sonja Kirchhoff et al.

Current deep-learning based methods do not easily integrate to clinical protocols, neither take full advantage of medical knowledge. In this work, we propose and compare several strategies relying on curriculum learning, to support the classification of proximal femur fracture from X-ray images, a challenging problem as reflected by existing intra- and inter-expert disagreement. Our strategies are derived from knowledge such as medical decision trees and inconsistencies in the annotations of multiple experts, which allows us to assign a degree of difficulty to each training sample. We demonstrate that if we start learning "easy" examples and move towards "hard", the model can reach a better performance, even with fewer data. The evaluation is performed on the classification of a clinical dataset of about 1000 X-ray images. Our results show that, compared to class-uniform and random strategies, the proposed medical knowledge-based curriculum, performs up to 15% better in terms of accuracy, achieving the performance of experienced trauma surgeons.

Ilaria Bonavita, Xavier Rafael-Palou, Mario Ceresa et al.

The early identification of malignant pulmonary nodules is critical for better lung cancer prognosis and less invasive chemo or radio therapies. Nodule malignancy assessment done by radiologists is extremely useful for planning a preventive intervention but is, unfortunately, a complex, time-consuming and error-prone task. This explains the lack of large datasets containing radiologists malignancy characterization of nodules. In this article, we propose to assess nodule malignancy through 3D convolutional neural networks and to integrate it in an automated end-to-end existing pipeline of lung cancer detection. For training and testing purposes we used independent subsets of the LIDC dataset. Adding the probabilities of nodules malignity in a baseline lung cancer pipeline improved its F1-weighted score by 14.7%, whereas integrating the malignancy model itself using transfer learning outperformed the baseline prediction by 11.8% of F1-weighted score. Despite the limited size of the lung cancer datasets, integrating predictive models of nodule malignancy improves prediction of lung cancer.

Víctor M. Campello, Carlos Martín-Isla, Cristian Izquierdo et al.

Accurate segmentation of the cardiac boundaries in late gadolinium enhancement magnetic resonance images (LGE-MRI) is a fundamental step for accurate quantification of scar tissue. However, while there are many solutions for automatic cardiac segmentation of cine images, the presence of scar tissue can make the correct delineation of the myocardium in LGE-MRI challenging even for human experts. As part of the Multi-Sequence Cardiac MR Segmentation Challenge, we propose a solution for LGE-MRI segmentation based on two components. First, a generative adversarial network is trained for the task of modality-to-modality translation between cine and LGE-MRI sequences to obtain extra synthetic images for both modalities. Second, a deep learning model is trained for segmentation with different combinations of original, augmented and synthetic sequences. Our results based on three magnetic resonance sequences (LGE, bSSFP and T2) from 45 different patients show that the multi-sequence model training integrating synthetic images and data augmentation improves in the segmentation over conventional training with real datasets. In conclusion, the accuracy of the segmentation of LGE-MRI images can be improved by using complementary information provided by non-contrast MRI sequences.

Karen López-Linares, Inmaculada García, Ainhoa García-Familiar et al.

An abdominal aortic aneurysm (AAA) is a focal dilation of the aorta that, if not treated, tends to grow and may rupture. A significant unmet need in the assessment of AAA disease, for the diagnosis, prognosis and follow-up, is the determination of rupture risk, which is currently based on the manual measurement of the aneurysm diameter in a selected Computed Tomography Angiography (CTA) scan. However, there is a lack of standardization determining the degree and rate of disease progression, due to the lack of robust, automated aneurysm segmentation tools that allow quantitatively analyzing the AAA. In this work, we aim at proposing the first 3D convolutional neural network for the segmentation of aneurysms both from preoperative and postoperative CTA scans. We extensively validate its performance in terms of diameter measurements, to test its applicability in the clinical practice, as well as regarding the relative volume difference, and Dice and Jaccard scores. The proposed method yields a mean diameter measurement error of 3.3 mm, a relative volume difference of 8.58 %, and Dice and Jaccard scores of 87 % and 77 %, respectively. At a clinical level, an aneurysm enlargement of 10 mm is considered relevant, thus, our method is suitable to automatically determine the AAA diameter and opens up the opportunity for more complex aneurysm analysis.

Karen López-Linares, Nerea Aranjuelo, Luis Kabongo et al.

Computerized Tomography Angiography (CTA) based follow-up of Abdominal Aortic Aneurysms (AAA) treated with Endovascular Aneurysm Repair (EVAR) is essential to evaluate the progress of the patient and detect complications. In this context, accurate quantification of post-operative thrombus volume is required. However, a proper evaluation is hindered by the lack of automatic, robust and reproducible thrombus segmentation algorithms. We propose a new fully automatic approach based on Deep Convolutional Neural Networks (DCNN) for robust and reproducible thrombus region of interest detection and subsequent fine thrombus segmentation. The DetecNet detection network is adapted to perform region of interest extraction from a complete CTA and a new segmentation network architecture, based on Fully Convolutional Networks and a Holistically-Nested Edge Detection Network, is presented. These networks are trained, validated and tested in 13 post-operative CTA volumes of different patients using a 4-fold cross-validation approach to provide more robustness to the results. Our pipeline achieves a Dice score of more than 82% for post-operative thrombus segmentation and provides a mean relative volume difference between ground truth and automatic segmentation that lays within the experienced human observer variance without the need of human intervention in most common cases.