RESUMEN
PURPOSE: Identifying optimal machine learning pipelines for computer-aided diagnosis is key for the development of robust, reproducible, and clinically relevant imaging biomarkers for endometrial carcinoma. The purpose of this study was to introduce the mathematical development of image descriptors computed from spherical harmonics (SPHARM) decompositions as well as the associated machine learning pipeline, and to evaluate their performance in predicting deep myometrial invasion (MI) and histopathological high-grade in preoperative multiparametric magnetic resonance imaging (MRI). PATIENTS AND METHODS: This retrospective study included 128 women with histopathology-confirmed endometrial carcinomas who underwent 1.5-T MRI before hysterectomy between January 2011 and July 2015. SPHARM descriptors of each tumor were computed on multiparametric MRI images (T2-weighted, diffusion-weighted, dynamic contrast-enhanced-MRI and apparent diffusion coefficient maps). Tensor-based logistic regression was used to classify two-dimensional SPHARM rotationally-invariant descriptors. Head-to-head comparisons with radiomics analyses were performed with DeLong tests with Bonferroni-Holm correction to compare diagnostic performances. RESULTS: With all MRI contrasts, SPHARM analysis resulted in area under the curve, sensitivity, specificity, and balanced accuracy values of 0.94 (95% confidence interval [CI]: 0.85, 1.00), 100% (95% CI: 100, 100), 74% (95% CI: 51, 92), 87% (95% CI: 78, 98), respectively, for predicting deep MI. For predicting high-grade tumor histology, the corresponding values for the same diagnostic metrics were 0.81 (95% CI: 0.64, 0.90), 93% (95% CI: 67, 100), 63% (95% CI: 45, 79) and 78% (95% CI: 64, 86). The corresponding values achieved via radiomics were 0.92 (95% CI: 0.82, 0.95), 82% (95% CI: 65, 93), 80% (95% CI: 51, 94), 81% (95% CI: 70, 91) for deep MI and 0.72 (95% CI: 0.58, 0.83), 93% (95% CI: 65, 100), 55% (95% CI: 41, 69), 74% (95% CI: 52, 88) for high-grade histology. The diagnostic performance of the SPHARM analysis was not significantly different (P = 0.62) from that of radiomics for predicting deep MI but was significantly higher (P = 0.044) for predicting high-grade histology. CONCLUSION: The proposed SPHARM analysis yields similar or higher diagnostic performance than radiomics in identifying deep MI and high-grade status in histology-proven endometrial carcinoma.
Asunto(s)
Neoplasias Endometriales , Imágenes de Resonancia Magnética Multiparamétrica , Humanos , Femenino , Imágenes de Resonancia Magnética Multiparamétrica/métodos , Estudios Retrospectivos , Curva ROC , Imagen por Resonancia Magnética/métodos , Neoplasias Endometriales/diagnóstico por imagen , Neoplasias Endometriales/patología , Imagen de Difusión por Resonancia Magnética/métodosRESUMEN
Two of the most common tasks in medical imaging are classification and segmentation. Either task requires labeled data annotated by experts, which is scarce and expensive to collect. Annotating data for segmentation is generally considered to be more laborious as the annotator has to draw around the boundaries of regions of interest, as opposed to assigning image patches a class label. Furthermore, in tasks such as breast cancer histopathology, any realistic clinical application often includes working with whole slide images, whereas most publicly available training data are in the form of image patches, which are given a class label. We propose an architecture that can alleviate the requirements for segmentation-level ground truth by making use of image-level labels to reduce the amount of time spent on data curation. In addition, this architecture can help unlock the potential of previously acquired image-level datasets on segmentation tasks by annotating a small number of regions of interest. In our experiments, we show using only one segmentation-level annotation per class, we can achieve performance comparable to a fully annotated dataset.
Asunto(s)
Neoplasias de la Mama , Aprendizaje Profundo , Neoplasias de la Mama/diagnóstico por imagen , Femenino , HumanosRESUMEN
Histopathological images contain rich phenotypic information that can be used to monitor underlying mechanisms contributing to disease progression and patient survival outcomes. Recently, deep learning has become the mainstream methodological choice for analyzing and interpreting histology images. In this paper, we present a comprehensive review of state-of-the-art deep learning approaches that have been used in the context of histopathological image analysis. From the survey of over 130 papers, we review the field's progress based on the methodological aspect of different machine learning strategies such as supervised, weakly supervised, unsupervised, transfer learning and various other sub-variants of these methods. We also provide an overview of deep learning based survival models that are applicable for disease-specific prognosis tasks. Finally, we summarize several existing open datasets and highlight critical challenges and limitations with current deep learning approaches, along with possible avenues for future research.
Asunto(s)
Algoritmos , Redes Neurales de la Computación , Técnicas Histológicas , Humanos , Procesamiento de Imagen Asistido por Computador , Aprendizaje AutomáticoRESUMEN
Whole slide images (WSIs) pose unique challenges when training deep learning models. They are very large which makes it necessary to break each image down into smaller patches for analysis, image features have to be extracted at multiple scales in order to capture both detail and context, and extreme class imbalances may exist. Significant progress has been made in the analysis of these images, thanks largely due to the availability of public annotated datasets. We postulate, however, that even if a method scores well on a challenge task, this success may not translate to good performance in a more clinically relevant workflow. Many datasets consist of image patches which may suffer from data curation bias; other datasets are only labelled at the whole slide level and the lack of annotations across an image may mask erroneous local predictions so long as the final decision is correct. In this paper, we outline the differences between patch or slide-level classification versus methods that need to localize or segment cancer accurately across the whole slide, and we experimentally verify that best practices differ in both cases. We apply a binary cancer detection network on post neoadjuvant therapy breast cancer WSIs to find the tumor bed outlining the extent of cancer, a task which requires sensitivity and precision across the whole slide. We extensively study multiple design choices and their effects on the outcome, including architectures and augmentations. We propose a negative data sampling strategy, which drastically reduces the false positive rate (25% of false positives versus 62.5%) and improves each metric pertinent to our problem, with a 53% reduction in the error of tumor extent. Our results indicate classification performances of image patches versus WSIs are inversely related when the same negative data sampling strategy is used. Specifically, injection of negatives into training data for image patch classification degrades the performance, whereas the performance is improved for slide and pixel-level WSI classification tasks. Furthermore, we find applying extensive augmentations helps more in WSI-based tasks compared to patch-level image classification.