Label noise is prevalent in machine learning datasets. It is crucial to identify and remove label noise because models trained on noisy data can have substantially reduced accuracy and generalizability. Most existing label noise detection approaches are designed for classification tasks, and data cleaning for outcome prediction analysis is relatively unexplored. Inspired by the fluctuations in performance across different folds in cross-validation, we propose Repeated Cross-Validations for label noise estimation (ReCoV) to address this gap. ReCoV constructs a noise histogram that ranks the noise level of samples based on a large number of cross-validations by recording sample IDs in each worst-performing fold. We further propose three approaches for identifying noisy samples based on noise histograms to address increasingly complex noise distributions. We show that ReCoV outperforms state-of-the-art algorithms for label cleaning in a classification task benchmark. More importantly, we show that removing ReCoV-identified noisy samples in two medical imaging outcome prediction datasets significantly improves model performance on test sets. As a statistical approach that does not rely on hyperparameters, noise distributions, or model structures, ReCoV is compatible with any machine learning analysis.
Cross-modal retrieval has drawn much attention in both computer vision and natural language processing domains. With the development of convolutional and recurrent neural networks, the bottleneck of retrieval across image-text modalities is no longer the extraction of image and text features but an efficient loss function learning in embedding space. Many loss functions try to closer pairwise features from heterogeneous modalities. This paper proposes a method for learning joint embedding of images and texts using an intra-modal constraint loss function to reduce the violation of negative pairs from the same homogeneous modality. Experimental results show that our approach outperforms state-of-the-art bi-directional image-text retrieval methods on Flickr30K and Microsoft COCO datasets. Our code is publicly available: https://github.com/CanonChen/IMC.
Cancer stage is a large determinant of patient prognosis and management in many cancer types, and is often assessed using medical imaging modalities, such as CT and MRI. These medical images contain rich information that can be explored to stratify patients within each stage group to further improve prognostic algorithms. Although the majority of cancer deaths result from metastatic and multifocal disease, building imaging biomarkers for patients with multiple tumors has been a challenging task due to the lack of annotated datasets and standard study framework. In this paper, we process two public datasets to set up a benchmark cohort of 341 patient in total for studying outcome prediction of multifocal metastatic cancer. We identify the lack of expressiveness in common multiple instance classification networks and propose two injective multiple instance pooling functions that are better suited to outcome prediction. Our results show that multiple instance learning with injective pooling functions can achieve state-of-the-art performance in the non-small-cell lung cancer CT and head and neck CT outcome prediction benchmarking tasks. We will release the processed multifocal datasets, our code and the intermediate files i.e. extracted radiomic features to support further transparent and reproducible research.
The MICCAI conference has encountered tremendous growth over the last years in terms of the size of the community, as well as the number of contributions and their technical success. With this growth, however, come new challenges for the community. Methods are more difficult to reproduce and the ever-increasing number of paper submissions to the MICCAI conference poses new questions regarding the selection process and the diversity of topics. To exchange, discuss, and find novel and creative solutions to these challenges, a new format of a hackathon was initiated as a satellite event at the MICCAI 2020 conference: The MICCAI Hackathon. The first edition of the MICCAI Hackathon covered the topics reproducibility, diversity, and selection of MICCAI papers. In the manner of a small think-tank, participants collaborated to find solutions to these challenges. In this report, we summarize the insights from the MICCAI Hackathon into immediate and long-term measures to address these challenges. The proposed measures can be seen as starting points and guidelines for discussions and actions to possibly improve the MICCAI conference with regards to reproducibility, diversity, and selection of papers.
Colorectal cancer is one of the most common and lethal cancers and colorectal cancer liver metastases (CRLM) is the major cause of death in patients with colorectal cancer. Multifocality occurs frequently in CRLM, but is relatively unexplored in CRLM outcome prediction. Most existing clinical and imaging biomarkers do not take the imaging features of all multifocal lesions into account. In this paper, we present an end-to-end autoencoder-based multiple instance neural network (AMINN) for the prediction of survival outcomes in multifocal CRLM patients using radiomic features extracted from contrast-enhanced MRIs. Specifically, we jointly train an autoencoder to reconstruct input features and a multiple instance network to make predictions by aggregating information from all tumour lesions of a patient. In addition, we incorporate a two-step normalization technique to improve the training of deep neural networks, built on the observation that the distributions of radiomic features are almost always severely skewed. Experimental results empirically validated our hypothesis that incorporating imaging features of all lesions improves outcome prediction for multifocal cancer. The proposed ADMINN framework achieved an area under the ROC curve (AUC) of 0.70, which is 19.5% higher than baseline methods. We built a risk score based on the outputs of our network and compared it to other clinical and imaging biomarkers. Our risk score is the only one that achieved statistical significance in univariate and multivariate cox proportional hazard modeling in our cohort of multifocal CRLM patients. The effectiveness of incorporating all lesions and applying two-step normalization is demonstrated by a series of ablation studies. Our code will be released after the peer-review process.
The booming successes of machine learning in different domains boost industry-scale deployments of innovative AI algorithms, systems, and architectures, and thus the importance of benchmarking grows. However, the confidential nature of the workloads, the paramount importance of the representativeness and diversity of benchmarks, and the prohibitive cost of training a state-of-the-art model mutually aggravate the AI benchmarking challenges. In this paper, we present a balanced AI benchmarking methodology for meeting the subtly different requirements of different stages in developing a new system/architecture and ranking/purchasing commercial off-the-shelf ones. Performing an exhaustive survey on the most important AI domain-Internet services with seventeen industry partners, we identify and include seventeen representative AI tasks to guarantee the representativeness and diversity of the benchmarks. Meanwhile, for reducing the benchmarking cost, we select a benchmark subset to a minimum-three tasks-according to the criteria: diversity of model complexity, computational cost, and convergence rate, repeatability, and having widely-accepted metrics or not. We contribute by far the most comprehensive AI benchmark suite-AIBench. The evaluations show AIBench outperforms MLPerf in terms of the diversity and representativeness of model complexity, computational cost, convergent rate, computation and memory access patterns, and hotspot functions. With respect to the AIBench full benchmarks, its subset shortens the benchmarking cost by 41%, while maintaining the primary workload characteristics. The specifications, source code, and performance numbers are publicly available from the web site http://www.benchcouncil.org/AIBench/index.html.
Accurate segmentation of lung and infection in COVID-19 CT scans plays an important role in the quantitative management of patients. Most of the existing studies are based on large and private annotated datasets that are impractical to obtain from a single institution, especially when radiologists are busy fighting the coronavirus disease. Furthermore, it is hard to compare current COVID-19 CT segmentation methods as they are developed on different datasets, trained in different settings, and evaluated with different metrics. In this paper, we created a COVID-19 3D CT dataset with 20 cases that contains 1800+ annotated slices and made it publicly available. To promote the development of annotation-efficient deep learning methods, we built three benchmarks for lung and infection segmentation that contain current main research interests, e.g., few-shot learning, domain generalization, and knowledge transfer. For a fair comparison among different segmentation methods, we also provide unified training, validation and testing dataset splits, and evaluation metrics and corresponding code. In addition, we provided more than 40 pre-trained baseline models for the benchmarks, which not only serve as out-of-the-box segmentation tools but also save computational time for researchers who are interested in COVID-19 lung and infection segmentation. To the best of our knowledge, this work presents the largest public annotated COVID-19 CT volume dataset, the first segmentation benchmark, and the most pre-trained models up to now. We hope these resources (\url{https://gitee.com/junma11/COVID-19-CT-Seg-Benchmark}) could advance the development of deep learning methods for COVID-19 CT segmentation with limited data.
Domain-specific software and hardware co-design is encouraging as it is much easier to achieve efficiency for fewer tasks. Agile domain-specific benchmarking speeds up the process as it provides not only relevant design inputs but also relevant metrics, and tools. Unfortunately, modern workloads like Big data, AI, and Internet services dwarf the traditional one in terms of code size, deployment scale, and execution path, and hence raise serious benchmarking challenges. This paper proposes an agile domain-specific benchmarking methodology. Together with seventeen industry partners, we identify ten important end-to-end application scenarios, among which sixteen representative AI tasks are distilled as the AI component benchmarks. We propose the permutations of essential AI and non-AI component benchmarks as end-to-end benchmarks. An end-to-end benchmark is a distillation of the essential attributes of an industry-scale application. We design and implement a highly extensible, configurable, and flexible benchmark framework, on the basis of which, we propose the guideline for building end-to-end benchmarks, and present the first end-to-end Internet service AI benchmark. The preliminary evaluation shows the value of our benchmark suite---AIBench against MLPerf and TailBench for hardware and software designers, micro-architectural researchers, and code developers. The specifications, source code, testbed, and results are publicly available from the web site \url{http://www.benchcouncil.org/AIBench/index.html}.
Quantitative medical image computing (radiomics) has been widely applied to build prediction models from medical images. However, overfitting is a significant issue in conventional radiomics, where a large number of radiomic features are directly used to train and test models that predict genotypes or clinical outcomes. In order to tackle this problem, we propose an unsupervised learning pipeline composed of an autoencoder for representation learning of radiomic features and a Gaussian mixture model based on minimum message length criterion for clustering. By incorporating probabilistic modeling, disease heterogeneity has been taken into account. The performance of the proposed pipeline was evaluated on an institutional MRI cohort of 108 patients with colorectal cancer liver metastases. Our approach is capable of automatically selecting the optimal number of clusters and assigns patients into clusters (imaging subtypes) with significantly different survival rates. Our method outperforms other unsupervised clustering methods that have been used for radiomics analysis and has comparable performance to a state-of-the-art imaging biomarker.
Despite their success in many computer vision tasks, convolutional networks tend to require large amounts of labeled data to achieve generalization. Furthermore, the performance is not guaranteed on a sample from an unseen domain at test time, if the network was not exposed to similar samples from that domain at training time. This hinders the adoption of these techniques in clinical setting where the imaging data is scarce, and where the intra- and inter-domain variance of the data can be substantial. We propose a domain adaptation technique that is especially suitable for deep networks to alleviate this requirement of labeled data. Our method utilizes gradient reversal layers and Squeezeand-Excite modules to stabilize the training in deep networks. The proposed method was applied to publicly available histopathology and chest X-ray databases and achieved superior performance to existing state-of-the-art networks with and without domain adaptation. Depending on the application, our method can improve multi-class classification accuracy by 5-20% compared to DANN introduced in (Ganin, 2014).