An official website of the United States government
Abstract BackgroundLung cancer is the deadliest and second most common cancer in the United States due to the lack of symptoms for early diagnosis. Pulmonary nodules are small abnormal regions that can be potentially correlated to the occurrence of lung cancer. Early detection of these nodules is critical because it can significantly improve the patient's survival rates. Thoracic thin‐sliced computed tomography (CT) scanning has emerged as a widely used method for diagnosing and prognosis lung abnormalities. PurposeThe standard clinical workflow of detecting pulmonary nodules relies on radiologists to analyze CT images to assess the risk factors of cancerous nodules. However, this approach can be error‐prone due to the various nodule formation causes, such as pollutants and infections. Deep learning (DL) algorithms have recently demonstrated remarkable success in medical image classification and segmentation. As an ever more important assistant to radiologists in nodule detection, it is imperative ensure the DL algorithm and radiologist to better understand the decisions from each other. This study aims to develop a framework integrating explainable AI methods to achieve accurate pulmonary nodule detection. MethodsA robust and explainable detection (RXD) framework is proposed, focusing on reducing false positives in pulmonary nodule detection. Its implementation is based on an explanation supervision method, which uses nodule contours of radiologists as supervision signals to force the model to learn nodule morphologies, enabling improved learning ability on small dataset, and enable small dataset learning ability. In addition, two imputation methods are applied to the nodule region annotations to reduce the noise within human annotations and allow the model to have robust attributions that meet human expectations. The 480, 265, and 265 CT image sets from the public Lung Image Database Consortium and Image Database Resource Initiative (LIDC‐IDRI) dataset are used for training, validation, and testing. ResultsUsing only 10, 30, 50, and 100 training samples sequentially, our method constantly improves the classification performance and explanation quality of baseline in terms of Area Under the Curve (AUC) and Intersection over Union (IoU). In particular, our framework with a learnable imputation kernel improves IoU from baseline by 24.0% to 80.0%. A pre‐defined Gaussian imputation kernel achieves an even greater improvement, from 38.4% to 118.8% from baseline. Compared to the baseline trained on 100 samples, our method shows less drop in AUC when trained on fewer samples. A comprehensive comparison of interpretability shows that our method aligns better with expert opinions. ConclusionsA pulmonary nodule detection framework was demonstrated using public thoracic CT image datasets. The framework integrates the robust explanation supervision (RES) technique to ensure the performance of nodule classification and morphology. The method can reduce the workload of radiologists and enable them to focus on the diagnosis and prognosis of the potential cancerous pulmonary nodules at the early stage to improve the outcomes for lung cancer patients.
more »
« less
Both fluid intelligence and visual object recognition ability relate to nodule detection in chest radiographs
Summary Recent work suggests that some aspects of lung nodule detection ability may relate to object recognition ability. However, this work only sampled radiological novices. Here, we further investigate whether object recognition ability predicts lung nodule detection ability (as measured by the Vanderbilt Chest Radiograph Test or VCRT), after controlling for experience and fluid intelligence, in a sample of radiologists and nonradiologists. We find that radiological experience accounts for approximately 50% of VCRT variance. After controlling for experience, fluid intelligence and object recognition ability account for an additional 15% of VCRT variance. These results suggest that while training is key in learning to detect nodules, given the same experience level, those with higher fluid intelligence and object recognition ability perform better. The recently proposed construct of visual object recognition ability may add unique information relative to general cognitive skills in assessing aptitude for a career in radiology.
more »
« less
- Award ID(s):
- 1640681
- PAR ID:
- 10075716
- Publisher / Repository:
- Wiley Blackwell (John Wiley & Sons)
- Date Published:
- Journal Name:
- Applied Cognitive Psychology
- Volume:
- 32
- Issue:
- 6
- ISSN:
- 0888-4080
- Page Range / eLocation ID:
- p. 755-762
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
One of the challenges of using machine learning techniques with medical data is the frequent dearth of source image data on which to train. A representative example is automated lung cancer diagnosis, where nodule images need to be classified as suspicious or benign. In this work we propose an automatic synthetic lung nodule image generator. Our 3D shape generator is designed to augment the variety of 3D images. Our proposed system takes root in autoencoder techniques, and we provide extensive experimental characterization that demonstrates its ability to produce quality synthetic images.more » « less
-
null (Ed.)In recent work, the Vanderbilt Holistic Processing Tests for novel objects (VHPT-NOs), were used to show that holistic processing for artificial objects increased as a function of parametric variation of experience. Here, novices are tested on the VHPT-Nos to address two questions. First, does the test detect any level of holistic processing for novel objects in novices? Second, how is part matching performance on this test related to object recognition ability, as measured by the Novel Object Memory Test (NOMT)? In a high-powered study, we provide substantial evidence of no holistic processing on the VHPT-NO in novices, including for arguably facelike symmetrical Greebles. Evidence of no correlations between measures of holistic processing suggests that these indices can be considered free of influences from domain-general selective attention. In contrast, overall performance in part matching in the VHPT-NO shows shared variance across categories, which we postulate is related to object recognition. A second study provides direct evidence that part matching measures to a large extent the same ability as whole object learning on the NOMT. Our results suggest that any holistic processing measured in the VHPT-NOs will not be contaminated by domain-general effects and can be considered entirely due to experience with a category. The VHPT-NO will therefore be useful in further examination of how different aspects of experience contribute to the development of holistic processing.more » « less
-
Individual differences in expertise with non-face objects has been positively related to neural selectivity for these objects in several brain regions, including in the fusiform face area (FFA). Recently, we reported that FFA’s cortical thickness is also positively correlated with expertise for non-living objects, while FFA’s cortical thickness is negatively correlated with face recognition ability. These opposite relations between structure and visual abilities, obtained in the same subjects, were postulated to reflect the earlier experience with faces relative to cars, with different mechanisms of plasticity operating at these different developmental times. Here we predicted that variability for faces, presumably reflecting pruning, would be found selectively in deep cortical layers. In 13 men selected to vary in their performance with faces, we used ultra-high field imaging (7 Tesla), we localized the FFA functionally and collected and averaged 6 ultra-high resolution susceptibility weighed images (SWI). Voxel dimensions were 0.194x0.194x1.00mm, covering 20 slices with 0.1mm gap. Images were then processed by two operators blind to behavioral results to define the gray matter/white matter (deep) and gray matter/CSF (superficial) cortical boundaries. Internal boundaries between presumed deep, middle and superficial cortical layers were obtained with an automated method based on image intensities. We used an extensive battery of behavioral tests to quantify both face and object recognition ability. We replicate prior work with face and non-living object recognition predicting large and independent parts of the variance in cortical thickness of the right FFA, in different directions. We also find that face recognition is specifically predicted by the thickness of the deep cortical layers in FFA, whereas recognition of vehicles relates to the thickness of all cortical layers. Our results represent the most precise structural correlate of a behavioral ability to date, linking face recognition ability to a specific layer of a functionally-defined area.more » « less
-
3D CT point clouds reconstructed from the original CT images are naturally represented in real-world coordinates. Compared with CT images, 3D CT point clouds contain invariant geometric features with irregular spatial distributions from multiple viewpoints. This paper rethinks pulmonary nodule detection in CT point cloud representations. We first extract the multi-view features from a sparse convolutional (SparseConv) encoder by rotating the point clouds with different angles in the world coordinate. Then, to simultaneously learn the discriminative and robust spatial features from various viewpoints, a nodule proposal optimization schema is proposed to obtain coarse nodule regions by aggregating consistent nodule proposals prediction from multi-view features. Last, the multi-level features and semantic segmentation features extracted from a SparseConv decoder are concatenated with multi-view features for final nodule region regression. Experiments on the benchmark dataset (LUNA16) demonstrate the feasibility of applying CT point clouds in lung nodule detection task. Furthermore, we observe that by combining multi-view predictions, the performance of the proposed framework is greatly improved compared to single-view, while the interior texture features of nodules from images are more suitable for detecting nodules in small sizes.more » « less
