An interpretable deep hierarchical semantic convolutional neural network for lung nodule malignancy classification

Shen, Shiwen; Han, Steve; Aberle, Denise R.; Bui, Alex A. T.; Hsu, William

doi:10.1016/j.eswa.2019.01.048

Cited by 218 publications

(151 citation statements)

References 39 publications

Supporting

Mentioning

149

Contrasting

Unclassified

Order By: Relevance

“…More specifically, predictive models for EGFR and KRAS mutation status in lung cancer were developed. Following the current direction in the literature, where the analysis only focuses on the nodule structure and texture 36,37 , we started by using objective radiomic features directly extracted from nodules in CT scans. Then, semantic features, annotated during radiologist evaluation, were used as input.…”

mentioning

confidence: 99%

Identifying relationships between imaging phenotypes and lung cancer-related mutation status: EGFR and KRAS

Pinheiro

Pereira

Dias

et al. 2020

Sci Rep

View full text Add to dashboard Cite

EGFR and KRAS are the most frequently mutated genes in lung cancer, being active research topics in targeted therapy. The biopsy is the traditional method to genetically characterise a tumour. However, it is a risky procedure, painful for the patient, and, occasionally, the tumour might be inaccessible. This work aims to study and debate the nature of the relationships between imaging phenotypes and lung cancer-related mutation status. Until now, the literature has failed to point to new research directions, mainly consisting of results-oriented works in a field where there is still not enough available data to train clinically viable models. We intend to open a discussion about critical points and to present new possibilities for future radiogenomics studies. We conducted high-dimensional data visualisation and developed classifiers, which allowed us to analyse the results for EGFR and KRAS biological markers according to different combinations of input features. We show that EGFR mutation status might be correlated to CT scans imaging phenotypes; however, the same does not seem to hold for KRAS mutation status. Also, the experiments suggest that the best way to approach this problem is by combining nodule-related features with features from other lung structures. Lung cancer is the cancer type leading the incidence and mortality rates 1,2. This is linked to the fact that it is often diagnosed in an advanced stage, with 15% or less chance of a 5-year survival 3 , which magnifies the importance of treatments for advanced-stage disease. In Non-small-cell lung cancer (NSCLC), which constitutes 85% of all cases of lung cancer, certain genomic biomarkers are now considered predictive biomarkers and critical for the prognostic 4. Epidermal Growth Factor Receptor (EGFR) and Kristen Rat Sarcoma Viral Oncogene Homolog KRAS are the most frequently mutated gene in lung cancer 5. EGFR mutated is present in 15 to 50% of NSCLC patients from never-smokers 5. The two most common EGFR mutations (deletions in exon 19 and the single amino acid substitution L858R in exon 21) correspond to approximately 85% of the EGFR mutations in NSCLC. The other low frequency mutations include: point mutations, deletions, insertions, and duplications correspond to approximated 15% of EGFR mutations in NSCLC 6. Unlike the previous marker, KRAS is associated with tobacco use, with only 5 to 10% of KRAS-mutant lung cancers arising in never or light smokers 5,7. Surgically treated NSCLC patients with EGFR mutations showed better disease-free survival (DFS) and overall survival (OS) and the opposite was verified for KRAS, with worse DFS and OS 8. For cytotoxic chemotherapy, the role of EGFR and KRAS as a predictive marker is still unclear 9 ; however, it appears that mutant KRAS may predict a lack of response to chemotherapy 10. Regarding target therapy, tumour driver mutations have reliable predictive value and, in fact, they guide treatment decision in clinics 11. EGFR gene is a paradigmatic example, since its activating mutations, namely those loca...

show abstract

mentioning

confidence: 99%

Identifying relationships between imaging phenotypes and lung cancer-related mutation status: EGFR and KRAS

Pinheiro

Pereira

Dias

et al. 2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Different researchers studied different factors relevant to wound healing e.g., wet and dry environment effect on different type of skin wound, role of pH in wound healing, role of moisture balance in continuous healing. In the recent past, a few wound care solutions focused on exterior factors effect on wound healing or sensor-based wound care solutions [11][12][13][14][15][16][17][18][19] have been presented, however, the previous methods are only focused on single factor effect at a time and didn't provide and learning based investigation for wound care, as showed in Table 1.…”

Section: Related Workmentioning

confidence: 99%

An Intelligent and Smart Environment Monitoring System for Healthcare

et al. 2019

View full text Add to dashboard Cite

Skin wound healing is influenced by two kinds of environment i.e., exterior environment that is nearby to wound surface and interior environment that is the environment of the adjacent part under wound surface. Both types of environment play a vital role in wound healing, which may contribute to continuous or impaired wound healing. Although, different previous studies provided wound care solutions, but they focused on single environmental factors either wound moisture level, pH value or healing enzymes. Practically, it is insignificant to consider environmental effect by determination of single factors or two, as both types of environment contain a lot of other factors which must be part of investigation e.g., smoke, air pollution, air humidity, temperature, hydrogen gases etc. Also, previous studies didn’t classify overall healing either as continuous or impaired based on exterior environment effect. In current research work, we proposed an effective wound care solution based on exterior environment monitoring system integrated with Neural Network Model to consider exterior environment effect on wound healing process, either as continuous or impaired. Current research facilitates patients by providing them intelligent wound care solution to monitor and control wound healing at their home.

show abstract

“…Therefore, many attempts have been made to develop computer-aided diagnosis (CAD) systems for automatic discrimination. [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] Conventional CAD systems first use classical image processing techniques, such as morphologic operators, 3-5 region growing, 6 energy optimization, 7,8 and statistical learning, 9,10 to segment a region of interest (ROI) that includes the nodule. Then, handcrafted features are extracted from the ROI, which are then fed to a classifier for nodule classification.…”

Section: Introductionmentioning

confidence: 99%

Feature‐shared adaptive‐boost deep learning for invasiveness classification of pulmonary subsolid nodules in CT images

Wang

Chen

et al. 2020

Medical Physics

View full text Add to dashboard Cite

Purpose: In clinical practice, invasiveness is an important reference indicator for differentiating the malignant degree of subsolid pulmonary nodules. These nodules can be classified as atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or invasive adenocarcinoma (IAC). The automatic determination of a nodule's invasiveness based on chest CT scans can guide treatment planning. However, it is challenging, owing to the insufficiency of training data and their interclass similarity and intraclass variation. To address these challenges, we propose a two-stage deep learning strategy for this task: prior-feature learning followed by adaptive-boost deep learning. Methods: The adaptive-boost deep learning is proposed to train a strong classifier for invasiveness classification of subsolid nodules in chest CT images, using multiple 3D convolutional neural network (CNN)-based weak classifiers. Because ensembles of multiple deep 3D CNN models have a huge number of parameters and require large computing resources along with more training and testing time, the prior-feature learning is proposed to reduce the computations by sharing the CNN layers between all weak classifiers. Using this strategy, all weak classifiers can be integrated into a single network. Results: Tenfold cross validation of binary classification was conducted on a total of 1357 nodules, including 765 noninvasive (AAH and AIS) and 592 invasive nodules (MIA and IAC). Ablation experimental results indicated that the proposed binary classifier achieved an accuracy of 73:4% AE 1:4 with an AUC of 81.3% AE 2:2. These results are superior compared to those achieved by three experienced chest imaging specialists who achieved an accuracy of 69:1% , 69:3% , and 67:9% , respectively. About 200 additional nodules were also collected. These nodules covered 50 cases for each category (AAH, AIS, MIA, and IAC, respectively). Both binary and multiple classifications were performed on these data and the results demonstrated that the proposed method definitely achieves better performance than the performance achieved by nonensemble deep learning methods. Conclusions: It can be concluded that the proposed adaptive-boost deep learning can significantly improve the performance of invasiveness classification of pulmonary subsolid nodules in CT images, while the prior-feature learning can significantly reduce the total size of deep models. The promising results on clinical data show that the trained models can be used as an effective lung cancer screening tool in hospitals. Moreover, the proposed strategy can be easily extended to other similar classification tasks in 3D medical images.

show abstract

An interpretable deep hierarchical semantic convolutional neural network for lung nodule malignancy classification

Cited by 218 publications

References 39 publications

Identifying relationships between imaging phenotypes and lung cancer-related mutation status: EGFR and KRAS

Identifying relationships between imaging phenotypes and lung cancer-related mutation status: EGFR and KRAS

An Intelligent and Smart Environment Monitoring System for Healthcare

Feature‐shared adaptive‐boost deep learning for invasiveness classification of pulmonary subsolid nodules in CT images

Contact Info

Product

Resources

About