COVID-19 Infection Percentage Estimation from Computed Tomography Scans: Results and Insights from the International Per-COVID-19 Challenge.

COVID-19 analysis from medical imaging is an important task that has been intensively studied in the last years due to the spread of the COVID-19 pandemic. In fact, medical imaging has often been used as a complementary or main tool to recognize the infected persons. On the other hand, medical imaging has the ability to provide more details about COVID-19 infection, including its severity and spread, which makes it possible to evaluate the infection and follow-up the patient's state. CT scans are the most informative tool for COVID-19 infection, where the evaluation of COVID-19 infection is usually performed through infection segmentation. However, segmentation is a tedious task that requires much effort and time from expert radiologists. To deal with this limitation, an efficient framework for estimating COVID-19 infection as a regression task is proposed. The goal of the Per-COVID-19 challenge is to test the efficiency of modern deep learning methods on COVID-19 infection percentage estimation (CIPE) from CT scans. Participants had to develop an efficient deep learning approach that can learn from noisy data. In addition, participants had to cope with many challenges, including those related to COVID-19 infection complexity and crossdataset scenarios. This paper provides an overview of the COVID-19 infection percentage estimation challenge (Per-COVID-19) held at MIA-COVID-2022. Details of the competition data, challenges, and evaluation metrics are presented. The best performing approaches and their results are described and discussed.

Convolutional neural network (CNN)-enabled electrocardiogram (ECG) analysis: a comparison between standard twelve-lead and single-lead setups.

Background: Artificial intelligence (AI) has shown promise in the early detection of various cardiac conditions from a standard 12-lead electrocardiogram (ECG). However, the ability of AI to identify abnormalities from single-lead recordings across a range of pathological conditions remains to be systematically investigated. This study aims to assess the performance of a convolutional neural network (CNN) using a single-lead (D1) rather than a standard 12-lead setup for accurate identification of ECG abnormalities. Methods: We designed and trained a lightweight CNN to identify 20 different cardiac abnormalities on ECGs, using data from the PTB-XL dataset. With a relatively simple architecture, the network was designed to accommodate different combinations of leads as input (<100,000 learnable parameters). We compared various lead setups such as the standard 12-lead, D1 alone, and D1 paired with an additional lead. Results: This has been corrected to "The CNN based on single-lead ECG (D1) achieved satisfactory performance compared to the standard 12-lead framework (average percentage AUC difference: −8.7%). Notably, for certain diagnostic classes, there was no difference in the diagnostic AUC between the single-lead and the standard 12-lead setups. When a second lead was detected in the CNN in addition to D1, the AUC gap was further reduced to an average percentage difference of -2.8% compared with that of the standard 12-lead setup. Conclusions: A relatively lightweight CNN can predict different classes of cardiac abnormalities from D1 alone and the standard 12-lead ECG. Considering the growing availability of wearable devices capable of recording a D1-like single-lead ECG, we discuss how our findings contribute to the foundation of a large-scale screening of cardiac abnormalities.

LOss-Based SensiTivity rEgulaRization: Towards deep sparse neural networks.

LOBSTER (LOss-Based SensiTivity rEgulaRization) is a method for training neural networks having a sparse topology. Let the sensitivity of a network parameter be the variation of the loss function with respect to the variation of the parameter. Parameters with low sensitivity, i.e. having little impact on the loss when perturbed, are shrunk and then pruned to sparsify the network. Our method allows to train a network from scratch, i.e. without preliminary learning or rewinding. Experiments on multiple architectures and datasets show competitive compression ratios with minimal computational overhead.

SeReNe: Sensitivity-Based Regularization of Neurons for Structured Sparsity in Neural Networks.

Deep neural networks include millions of learnable parameters, making their deployment over resource-constrained devices problematic. Sensitivity-based regularization of neurons (SeReNe) is a method for learning sparse topologies with a structure, exploiting neural sensitivity as a regularizer. We define the sensitivity of a neuron as the variation of the network output with respect to the variation of the activity of the neuron. The lower the sensitivity of a neuron, the less the network output is perturbed if the neuron output changes. By including the neuron sensitivity in the cost function as a regularization term, we are able to prune neurons with low sensitivity. As entire neurons are pruned rather than single parameters, practical network footprint reduction becomes possible. Our experimental results on multiple network architectures and datasets yield competitive compression ratios with respect to state-of-the-art references.

Learnable Descriptors for Visual Search.

This work proposes LDVS, a learnable binary local descriptor devised for matching natural images within the MPEG CDVS framework. LDVS descriptors are learned so that they can be sign-quantized and compared using the Hamming distance. The underlying convolutional architecture enjoys a moderate parameters count for operations on mobile devices. Our experiments show that LDVS descriptors perform favorably over comparable learned binary descriptors at patch matching on two different datasets. A complete pair-wise image matching pipeline is then designed around LDVS descriptors, integrating them in the reference CDVS evaluation framework. Experiments show that LDVS descriptors outperform the compressed CDVS SIFT-like descriptors at pair-wise image matching over the challenging CDVS image dataset.