The early detection of glaucoma is essential in preventing visual impairment. Artificial intelligence (AI) can be used to analyze color fundus photographs (CFPs) in a cost-effective manner, making glaucoma screening more accessible. While AI models for glaucoma screening from CFPs have shown promising results in laboratory settings, their performance decreases significantly in real-world scenarios due to the presence of out-of-distribution and low-quality images. To address this issue, we propose the Artificial Intelligence for Robust Glaucoma Screening (AIROGS) challenge. This challenge includes a large dataset of around 113,000 images from about 60,000 patients and 500 different screening centers, and encourages the development of algorithms that are robust to ungradable and unexpected input data. We evaluated solutions from 14 teams in this paper and found that the best teams performed similarly to a set of 20 expert ophthalmologists and optometrists. The highest-scoring team achieved an area under the receiver operating characteristic curve of 0.99 (95% CI: 0.98-0.99) for detecting ungradable images on-the-fly. Additionally, many of the algorithms showed robust performance when tested on three other publicly available datasets. These results demonstrate the feasibility of robust AI-enabled glaucoma screening.

Diabetic retinopathy (DR) grading is crucial in determining the adequate treatment and follow up of patient, but the screening process can be tiresome and prone to errors. Deep learning approaches have shown promising performance as computer-aided diagnosis (CAD) systems, but their black-box behaviour hinders clinical application. We propose DR|GRADUATE, a novel deep learning-based DR grading CAD system that supports its decision by providing a medically interpretable explanation and an estimation of how uncertain that prediction is, allowing the ophthalmologist to measure how much that decision should be trusted. We designed DR|GRADUATE taking into account the ordinal nature of the DR grading problem. A novel Gaussian-sampling approach built upon a Multiple Instance Learning framework allow DR|GRADUATE to infer an image grade associated with an explanation map and a prediction uncertainty while being trained only with image-wise labels. DR|GRADUATE was trained on the Kaggle DR detection training set and evaluated across multiple datasets. In DR grading, a quadratic-weighted Cohen's kappa (κ) between 0.71 and 0.84 was achieved in five different datasets. We show that high κ values occur for images with low prediction uncertainty, thus indicating that this uncertainty is a valid measure of the predictions' quality. Further, bad quality images are generally associated with higher uncertainties, showing that images not suitable for diagnosis indeed lead to less trustworthy predictions. Additionally, tests on unfamiliar medical image data types suggest that DR|GRADUATE allows outlier detection. The attention maps generally highlight regions of interest for diagnosis. These results show the great potential of DR|GRADUATE as a second-opinion system in DR severity grading.

Early diagnosis of lung cancer via computed tomography can significantly reduce the morbidity and mortality rates associated with the pathology. However, searching lung nodules is a high complexity task, which affects the success of screening programs. Whilst computer-aided detection systems can be used as second observers, they may bias radiologists and introduce significant time overheads. With this in mind, this study assesses the potential of using gaze information for integrating automatic detection systems in the clinical practice. For that purpose, 4 radiologists were asked to annotate 20 scans from a public dataset while being monitored by an eye tracker device, and an automatic lung nodule detection system was developed. Our results show that radiologists follow a similar search routine and tend to have lower fixation periods in regions where finding errors occur. The overall detection sensitivity of the specialists was 0.67±0.07, whereas the system achieved 0.69. Combining the annotations of one radiologist with the automatic system significantly improves the detection performance to similar levels of two annotators. Filtering automatic detection candidates only for low fixation regions still significantly improves the detection sensitivity without increasing the number of false-positives.

Breast cancer is the most common invasive cancer in women, affecting more than 10% of women worldwide. Microscopic analysis of a biopsy remains one of the most important methods to diagnose the type of breast cancer. This requires specialized analysis by pathologists, in a task that i) is highly time- and cost-consuming and ii) often leads to nonconsensual results. The relevance and potential of automatic classification algorithms using hematoxylin-eosin stained histopathological images has already been demonstrated, but the reported results are still sub-optimal for clinical use. With the goal of advancing the state-of-the-art in automatic classification, the Grand Challenge on BreAst Cancer Histology images (BACH) was organized in conjunction with the 15th International Conference on Image Analysis and Recognition (ICIAR 2018). BACH aimed at the classification and localization of clinically relevant histopathological classes in microscopy and whole-slide images from a large annotated dataset, specifically compiled and made publicly available for the challenge. Following a positive response from the scientific community, a total of 64 submissions, out of 677 registrations, effectively entered the competition. The submitted algorithms improved the state-of-the-art in automatic classification of breast cancer with microscopy images to an accuracy of 87%. Convolutional neuronal networks were the most successful methodology in the BACH challenge. Detailed analysis of the collective results allowed the identification of remaining challenges in the field and recommendations for future developments. The BACH dataset remains publicly available as to promote further improvements to the field of automatic classification in digital pathology.

The almost Gaussian nature of OFDM (Orthogonal Frequency Division Multiplexing) signals with high number of subcarriers N is widely employed to characterize nonlinearly distorted OFDM signals and to evaluate the corresponding performance. In this paper we study the accuracy of the Gaussian approach when evaluating nonlinear effects in OFDM signals with finite number of subcarriers, showing the strengths and limitations of this approach. It is shown that the decomposition in useful and self-interference components is valid even for a reduced number of subcarriers. The Gaussian approximation of the nonlinear self-interference at the subcarrier level is very accurate provided that N2is high. However, the nonlinear distortion levels slightly lower than the ones obtained with the Gaussian approximation, with relative errors dropping with 1/N, leading to somewhat pessimistic SIR levels (Signal to Interference Ratio).

The almost Gaussian nature of OFDM (Orthogonal Frequency Division Multiplexing) signals with high number of subcarriers is widely employed to characterize nonlinearly distorted OFDM signals and to evaluate the corresponding performance. In this paper we study the accuracy of the Gaussian approach when evaluating nonlinear effects in OFDM signals with finite number of subcarriers, showing the strengths and limitations of this approach.

As with other multicarrier modulations, OFDMA signals (Orthogonal Frequency Division Multiple Access) have large envelope fluctuations making them very prone to nonlinear distortion effects. In this paper we study the impact of nonlinear devices on OFDMA signals. We present an analytical characterization of the signals at the output of a nonlinear device that can be used for obtaining the spectral characterization of the transmitted signals, as well as the computation of the nonlinear interference levels on the received signals. It is shown that the power allocated to each user has a key impact on the nonlinear distortion effects. For this reason, we should avoid having low-power and high-power users in adjacent subcarriers. It is also shown that nonlinear distortion levels are lower when just a small fraction of the subcarriers is used (i.e., when the system load is small).

In this paper, we consider MC-CDM schemes (MultiCarrier Code Division Multiplexing) where clipping techniques are employed to reduce the envelope fluctuations of the transmitted signals. Both CP-assisted (Cyclic Prefix) and ZP (Zero- Padded) MC-CDM schemes are studied. We develop frequencydomain turbo equalizers combined with an iterative estimation and cancelation of nonlinear distortion effects, with relatively low complexity since they allow FFT-based (Fast Fourier Transform), frequency-domain implementations. Our performance results show that the proposed receivers allow significant performance improvements at low and moderate SNR (Signal-to-Noise Ratio), even when strongly nonlinear transmitters are employed. The receiver for ZP MC-CDM is of special interest for systems where the duration of the channel impulse response is not a small fraction of the duration of the MC-CDM blocks, being suitable to MC-CDM systems with very large blocks (hundreds or even thousands of subcarriers), since they do not require the inversion nor the multiplication of large matrixes.

In this paper we consider the evaluation of nonlinear distortion effects on the uplink transmission of systems employing OFDMA signals (orthogonal frequency division multiple access). Our results allow an analytical spectral characterization of the transmitted signals, as well as the computation of the nonlinear interference levels on the received signals. It is shown that the power allocated to each user has a key impact on the nonlinear distortion effects. It is also shown that nonlinear distortion levels are lower when just a small fraction of the subcarriers is used (i.e., when the system load is small).

We consider MC-CDMA schemes, with reduced envelope fluctuations. Both CP-assisted (cyclic prefix) and ZP (zero-padded) MC-CDMA schemes are addressed. We develop turbo FDE (frequency-domain equalization) schemes, combined with cancelation of nonlinear distortion effects. The proposed turbo receivers allow significant performance improvements at low and moderate SNR, even when the transmitted signals have reduced envelope fluctuations.

The software radio concept is a topic of widespread interest in wireless cellular systems. Due to the high sampling ratio and quantization requirements, the ADC (analog-to-digital converter) is as key component of any software radio architecture. In this paper we present an analytical tool for evaluating the quantization requirements within the ADC used in software radio architectures. For this purpose, we take advantage of the Gaussian behavior of the multi-band/multi-user signal at the input of the wideband ADC and employ results on Gaussian signals and memoryless nonlinearities. This characterization is then used for the performance evaluation and optimization of different quantization characteristics.

In this paper we consider a broadband wireless system employing SC (single-carrier) modulations combined with FDE (frequency-domain equalization). We propose a receiver structure for joint carrier synchronization and equalization which combines IB-DFE (iterative block decision feedback equalization) and post-equalization carrier synchronization schemes. The proposed receiver can be regarded as a modified IB-DFE where we perform a decision-directed frequency offset estimation within each iteration of the equalizer. Our results show that the proposed receiver structure has a good performance, allowing BER (bit error rate) close to the MFB (matched filter bound), even with moderate frequency offsets and severe time-dispersive channels.