Recent developments in deep learning (DL) techniques have led to great performance improvement in medical image segmentation tasks, especially with the latest Transformer model and its variants. While labels from fusing multi-rater manual segmentations are often employed as ideal ground truths in DL model training, inter-rater variability due to factors such as training bias, image noise, and extreme anatomical variability can still affect the performance and uncertainty of the resulting algorithms. Knowledge regarding how inter-rater variability affects the reliability of the resulting DL algorithms, a key element in clinical deployment, can help inform better training data construction and DL models, but has not been explored extensively. In this paper, we measure aleatoric and epistemic uncertainties using test-time augmentation (TTA), test-time dropout (TTD), and deep ensemble to explore their relationship with inter-rater variability. Furthermore, we compare UNet and TransUNet to study the impacts of Transformers on model uncertainty with two label fusion strategies. We conduct a case study using multi-class paraspinal muscle segmentation from T2w MRIs. Our study reveals the interplay between inter-rater variability and uncertainties, affected by choices of label fusion strategies and DL models.
Automatic modulation classification (AMC) plays a critical role in wireless communications by autonomously classifying signals transmitted over the radio spectrum. Deep learning (DL) techniques are increasingly being used for AMC due to their ability to extract complex wireless signal features. However, DL models are computationally intensive and incur high inference latencies. This paper proposes the application of early exiting (EE) techniques for DL models used for AMC to accelerate inference. We present and analyze four early exiting architectures and a customized multi-branch training algorithm for this problem. Through extensive experimentation, we show that signals with moderate to high signal-to-noise ratios (SNRs) are easier to classify, do not require deep architectures, and can therefore leverage the proposed EE architectures. Our experimental results demonstrate that EE techniques can significantly reduce the inference speed of deep neural networks without sacrificing classification accuracy. We also thoroughly study the trade-off between classification accuracy and inference time when using these architectures. To the best of our knowledge, this work represents the first attempt to apply early exiting methods to AMC, providing a foundation for future research in this area.
T1 mapping is a quantitative magnetic resonance imaging (qMRI) technique that has emerged as a valuable tool in the diagnosis of diffuse myocardial diseases. However, prevailing approaches have relied heavily on breath-hold sequences to eliminate respiratory motion artifacts. This limitation hinders accessibility and effectiveness for patients who cannot tolerate breath-holding. Image registration can be used to enable free-breathing T1 mapping. Yet, inherent intensity differences between the different time points make the registration task challenging. We introduce PCMC-T1, a physically-constrained deep-learning model for motion correction in free-breathing T1 mapping. We incorporate the signal decay model into the network architecture to encourage physically-plausible deformations along the longitudinal relaxation axis. We compared PCMC-T1 to baseline deep-learning-based image registration approaches using a 5-fold experimental setup on a publicly available dataset of 210 patients. PCMC-T1 demonstrated superior model fitting quality (R2: 0.955) and achieved the highest clinical impact (clinical score: 3.93) compared to baseline methods (0.941, 0.946 and 3.34, 3.62 respectively). Anatomical alignment results were comparable (Dice score: 0.9835 vs. 0.984, 0.988). Our code and trained models are available at https://github.com/eyalhana/PCMC-T1.
Generating video stories from text prompts is a complex task. In addition to having high visual quality, videos need to realistically adhere to a sequence of text prompts whilst being consistent throughout the frames. Creating a benchmark for video generation requires data annotated over time, which contrasts with the single caption used often in video datasets. To fill this gap, we collect comprehensive human annotations on three existing datasets, and introduce StoryBench: a new, challenging multi-task benchmark to reliably evaluate forthcoming text-to-video models. Our benchmark includes three video generation tasks of increasing difficulty: action execution, where the next action must be generated starting from a conditioning video; story continuation, where a sequence of actions must be executed starting from a conditioning video; and story generation, where a video must be generated from only text prompts. We evaluate small yet strong text-to-video baselines, and show the benefits of training on story-like data algorithmically generated from existing video captions. Finally, we establish guidelines for human evaluation of video stories, and reaffirm the need of better automatic metrics for video generation. StoryBench aims at encouraging future research efforts in this exciting new area.
Language use has been shown to correlate with depression, but large-scale validation is needed. Traditional methods like clinic studies are expensive. So, natural language processing has been employed on social media to predict depression, but limitations remain-lack of validated labels, biased user samples, and no context. Our study identified 29 topics in 3919 smartphone-collected speech recordings from 265 participants using the Whisper tool and BERTopic model. Six topics with a median PHQ-8 greater than or equal to 10 were regarded as risk topics for depression: No Expectations, Sleep, Mental Therapy, Haircut, Studying, and Coursework. To elucidate the topic emergence and associations with depression, we compared behavioral (from wearables) and linguistic characteristics across identified topics. The correlation between topic shifts and changes in depression severity over time was also investigated, indicating the importance of longitudinally monitoring language use. We also tested the BERTopic model on a similar smaller dataset (356 speech recordings from 57 participants), obtaining some consistent results. In summary, our findings demonstrate specific speech topics may indicate depression severity. The presented data-driven workflow provides a practical approach to collecting and analyzing large-scale speech data from real-world settings for digital health research.
H-score is a semi-quantitative method used to assess the presence and distribution of proteins in tissue samples by combining the intensity of staining and percentage of stained nuclei. It is widely used but time-consuming and can be limited in accuracy and precision. Computer-aided methods may help overcome these limitations and improve the efficiency of pathologists' workflows. In this work, we developed a model EndoNet for automatic calculation of H-score on histological slides. Our proposed method uses neural networks and consists of two main parts. The first is a detection model which predicts keypoints of centers of nuclei. The second is a H-score module which calculates the value of the H-score using mean pixel values of predicted keypoints. Our model was trained and validated on 1780 annotated tiles with a shape of 100x100 $\mu m$ and performed 0.77 mAP on a test dataset. Moreover, the model can be adjusted to a specific specialist or whole laboratory to reproduce the manner of calculating the H-score. Thus, EndoNet is effective and robust in the analysis of histology slides, which can improve and significantly accelerate the work of pathologists.
In the cancer diagnosis pipeline, digital pathology plays an instrumental role in the identification, staging, and grading of malignant areas on biopsy tissue specimens. High resolution histology images are subject to high variance in appearance, sourcing either from the acquisition devices or the H\&E staining process. Nuclei segmentation is an important task, as it detects the nuclei cells over background tissue and gives rise to the topology, size, and count of nuclei which are determinant factors for cancer detection. Yet, it is a fairly time consuming task for pathologists, with reportedly high subjectivity. Computer Aided Diagnosis (CAD) tools empowered by modern Artificial Intelligence (AI) models enable the automation of nuclei segmentation. This can reduce the subjectivity in analysis and reading time. This paper provides an extensive review, beginning from earlier works use traditional image processing techniques and reaching up to modern approaches following the Deep Learning (DL) paradigm. Our review also focuses on the weak supervision aspect of the problem, motivated by the fact that annotated data is scarce. At the end, the advantages of different models and types of supervision are thoroughly discussed. Furthermore, we try to extrapolate and envision how future research lines will potentially be, so as to minimize the need for labeled data while maintaining high performance. Future methods should emphasize efficient and explainable models with a transparent underlying process so that physicians can trust their output.
Conventional time series classification approaches based on bags of patterns or shapelets face significant challenges in dealing with a vast amount of feature candidates from high-dimensional multivariate data. In contrast, deep neural networks can learn low-dimensional features efficiently, and in particular, Convolutional Neural Networks (CNN) have shown promising results in classifying Multivariate Time Series (MTS) data. A key factor in the success of deep neural networks is this astonishing expressive power. However, this power comes at the cost of complex, black-boxed models, conflicting with the goals of building reliable and human-understandable models. An essential criterion in understanding such predictive deep models involves quantifying the contribution of time-varying input variables to the classification. Hence, in this work, we introduce a new framework for interpreting multivariate time series data by extracting and clustering the input representative patterns that highly activate CNN neurons. This way, we identify each signal's role and dependencies, considering all possible combinations of signals in the MTS input. Then, we construct a graph that captures the temporal relationship between the extracted patterns for each layer. An effective graph merging strategy finds the connection of each node to the previous layer's nodes. Finally, a graph embedding algorithm generates new representations of the created interpretable time-series features. To evaluate the performance of our proposed framework, we run extensive experiments on eight datasets of the UCR/UEA archive, along with HAR and PAM datasets. The experiments indicate the benefit of our time-aware graph-based representation in MTS classification while enriching them with more interpretability.
In recent years, many NLP studies have focused solely on performance improvement. In this work, we focus on the linguistic and scientific aspects of NLP. We use the task of generating referring expressions in context (REG-in-context) as a case study and start our analysis from GREC, a comprehensive set of shared tasks in English that addressed this topic over a decade ago. We ask what the performance of models would be if we assessed them (1) on more realistic datasets, and (2) using more advanced methods. We test the models using different evaluation metrics and feature selection experiments. We conclude that GREC can no longer be regarded as offering a reliable assessment of models' ability to mimic human reference production, because the results are highly impacted by the choice of corpus and evaluation metrics. Our results also suggest that pre-trained language models are less dependent on the choice of corpus than classic Machine Learning models, and therefore make more robust class predictions.
Computer-assisted treatment has emerged as a viable application of medical imaging, owing to the efficacy of deep learning models. Real-time inference speed remains a key requirement for such applications to help medical personnel. Even though there generally exists a trade-off between performance and model size, impressive efforts have been made to retain near-original performance by compromising model size. Neural network pruning has emerged as an exciting area that aims to eliminate redundant parameters to make the inference faster. In this study, we show an application of neural network pruning in polyp segmentation. We compute the importance score of convolutional filters and remove the filters having the least scores, which to some value of pruning does not degrade the performance. For computing the importance score, we use the Taylor First Order (TaylorFO) approximation of the change in network output for the removal of certain filters. Specifically, we employ a gradient-normalized backpropagation for the computation of the importance score. Through experiments in the polyp datasets, we validate that our approach can significantly reduce the parameter count and FLOPs retaining similar performance.